AMD Predicts End of 32-bit Processors

DDumitru writes "Infoworld reports that AMD predicts it will stop producing 32-bit processors by the end of 2005. By depending on price cuts for Athlon-64 and Opteron, AMD is predicting that it's sales of 32-bit CPUs will fall off and obsolete 32-bit systems in less than 3 years. This is either a push forward, or a tactic to try to capture the 64/32 bit standard leaving Intel in the rear. Or it could just be hype." I'm not in a hurry to ditch any of my 32-bit machines, so long as I get them replaced by 2038.

Comments

32-bits is dieing

[Anonymous Coward]

AMD now confirms: 32 bits is dieing.

Re:32-bits is dieing

[Anonymous Coward]

You mean 32-bits are dying.

You know.....

[inteller]

.....forgive me for being captain obvious, but my old A7M266 board runs just fine with XP 2000+....they can continue dropping the price on these suckers so that eventually I can max it out with 2600+s and also plop two MP 2600+s in my A7M266-D.....I don't even use half the capacity now...I'll be blown away when I plop those in for $50 each in a year or so.....keep predicting the demise AMD, it's all fine with me.

Re:You know.....

[Anonymous Coward]

why would you want to ? for $50 you should be able to get a 64 bit CPU anyway in a year.

Re:You know.....

[addaon]

You have to realize there's a certain minimum cost to retail processors, the same way there is to, say, video cards. You can't sell an AGP video card for $30 MSRP. The cost of packaging, shelf space, support, etc. outweighs the cost of hardware by so much at that level that, even if the hardware were essentially free, the product price would be around $30 or so. Ditto for processors; once a processor goes below about $40, it seems like it's no longer worth producing; the cost of packaging and so on outweighs any possible further reduction in price. On the other end, at $5 or so they become worth producing again (see PICs and such) because, at that point, no packaging or support is expected... but don't expect to ever find a nice $40 CPU to put in your motherboard.

Re:You know.....

[Popadopolis]

Still, you will eventually be left behind...

Re:You know.....

[Naffer]

From what I've heard, the yields on the new Athlon 64 FX chips have been atrociously low. We're not talking high prices because of speed bining here, but high prices because of terrible yield. Someone needs to move to 300mm wafers.

Re:You know.....

[Uma Thurman]

Not a new $40 CPU, but there's plenty of used ones out there. Someday I'm going to pick up a 10 Ghz processor and motherboard for $20.

Re:You know.....

[sirsnork]

They're going down to 90nm early next year which should improve yeilds

Re:You know.....

[Popadopolis]

I believe it was low compared to what was predicted for them, but high in comparison to what is out there, but I could be wrong.

Re:You know.....

[randyest]

Why is that? Generally smaller geometries result in lower yields. It's worse when the process is first rolled out, and yield does improve as a process matures (which 90nm has by no means done yet), but in general, a larger geometry process results in higher yields.

If you care, this is because defect densities in the silicon remain relatively constant, and although die sizes may be reduced somewhat with smaller geometries (not as much as you would think, though, due to wiring density not scaling linearly with gate size), the odds of a defect being fatal (i.e., falling into one of the increasingly dense "wrong spots" on the silicon) increase exponentially with gate size decrease.

Re:You know.....

[sweetooth]

Not quite $40 but close enough for government work. Of course once you including shipping you are looking at $50.

http://www.computerbrain.com/cbisys/items.asp?main =CPUs&cat=AMD_Athlon_XP_CPUs&mfr=all&class =

Re:You know.....

[kayen_telva]

pricewatch

looks good to me

Re:You know.....

[ivan256]

I run my A7M266-D, with 2 Athlon MP 1800+ processors, plugged into a Kill-A-Watt meter. All signs point to it being cheaper to buy a dual opteron over the not so long term simply for the power saving features at idle. I figure that the new machine would pay for itself in less than a year. I'll be buying as soon as there are new opterons readily available so the price on the old ones will be low.

Perhaps you should try being captian not-so-obvious and look and see how much it costs to leave that dual Athlon MP, which uses about the same amount of power at idle as at full utilization, on all the time. Also, if you don't need the capacity, why did you buy it in the first place?

Re:You know.....

[tomhudson]

poster wrote:

You can't sell an AGP video card for $30 MSRP

Why not? GeForce 64meg w. tv-out are retailing for about that now.

Ditto for processors; once a processor goes below about $40, it seems like it's no longer worth producing;

You can get a motherboard w. duron 2000 cpu for $100 CAN (about $75.US). Split the cost in 2 - half for mb, half for cpu, we are under the $40.00 price point already.

Same w. CD-ROMS at $20.00 new, etc.

Re:You know.....

[Eccles]

Why is that? Generally smaller geometries result in lower yields.

Yes, but the greater number of chips/wafer almost always more than compensates. Remember the increase in the chips/wafer is the square of the inverse of the decrease. I.e., if you go from .15 nm to .1 nm, the chips/wafer increases by (.15/.1)^2 = 2.25 times as many chips.

Re:You know.....

[the_ghost226]

the cost of packaging and so on outweighs any possible further reduction in price Actually, once you already have a bunch in a warehouse that you don't expect to sell, you can sell them below cost; as anything you get for them is more than you expected.

Re:You know.....

[sfe_software]

The cost of packaging, shelf space, support, etc. outweighs the cost of hardware by so much at that level that, even if the hardware were essentially free, the product price would be around $30 or so.

Then:

On the other end, at $5 or so they become worth producing again (see PICs and such) because, at that point, no packaging or support is expected...

I'm not so sure I agree with this logic. You can buy OEM CPUs today, even for high-end CPUs, with no packaging other than a small grey box (no sink/fan/documentation/etc). I recently purchased an Athlon 2400+ like this, and it was about $20 less than the full packaged version with sink/fan.

I've seen older CPUs in computer shops going for $30 to $50, from old MMX 233 Pentiums on up to mid-range PIII chips. I see no reason the AMD 32-bit chips won't be sold at a low price once the manufacturer calls them "obsolete". It happens all the time...

Re:You know.....

[randyest]

Huh? Do you know what yield is in this context? (Hint: it's not the same as in recipes). It's not how many they can make, it's how many that work out of all they make.

Yield = (number of working dice) / (total number of dice manufactured)

Making more total dice does not increase your yield. You must improve the ratio of working to failing dice in order to increase yield.

Also important, recipe-style yield (total output) doesn't go up as you suggest either because, as I mentioned before, die size for a given design does not scale with gate size as you think it might. Wires in 90nm processes are not 1/2 the pitch of wires in 0.18um processes. It's more like a 25-30% reduction, and no matter how small your gates are, you still need the same number of wires (and the die area that does with it) to connect them. It's about time mfgr's started quoting wiring pitch along with gate size (some do) -- this might help avoid confusion from over simplifications such as yours.

Re:You know.....

[Ancient Devices King]

I regularly buy the cheapest AGP cards I can get. They go for anywhere from $5 from online discounters to $15 from in town places. Granted that they *suck*, but these are new, still in original box cards.

Re:You know.....

[TheLink]

I think you're missing his main point.

Sure his statement "you can't sell ABC at $X" was at face value incorrect. But you're missing his point by saying "ABC type XYZ is retailing for $Y" right now". You're talking about price details. He's talking about something else which in general is true.

Example: most people don't sell drinking straws in single units at low prices. It's not worth doing that. You bundle them together or with something else. Or you try to sell them as something else at a higher price.

The $X changes depending on where you are - India, US, Japan etc. But the principle is still useful, until someone comes up with something really different.

Re:You know.....

[Predius]

See my url... The world has yet to move to a point where my 386 can't operate. It may be slow, but anything you can do on a nice big fast 64bit box, my 386 can trudge through as well. If we haven't obsoleted the 386 completely yet, I think 32bit x86 has a LONG lifespan left yet.

Re:You know.....

[Anonymous Coward]

don't expect to ever find a nice $40 CPU

Cough. Excuse me? Have you ever tried looking for $40 processors? A Celeron or Athlon 1.3Ghz CPU is currently $40 - some 60% faster than my current 800Mhz desktop machine.

Re:You know.....

[tolan-b]

Perhaps. But CPUs are bulk produced, are bulk shipped, are sold in units of 1000, have virtually no packaging, and don't come with a support contract.

So the non-manufacturing and r&d costs are negligable.

Re:You know.....

[Jeff DeMaagd]

Well, there are people that thought that one couldn't make a PC for less than $2000 for many of the said reasons.

Re:You know.....

[Loki_1929]

"You can get a motherboard w. duron 2000 cpu for $100 CAN (about $75.US). Split the cost in 2 - half for mb, half for cpu, we are under the $40.00 price point already."

I'm sure you're getting a super board for that price too. What's it a 1/2 layer ECS fire-hazard? Perhaps something printed out by someone's inkjet printer?

The poster never said you couldn't possibly find any for under that price, they merely pointed out that it's no longer cost-efficient for a CPU manufacturer to produce processors that retail at under around $40. From what you've posted, it would seem as though AMD should continue producing and discounting Durons when they're $.25 each.

Re:You know.....

[Loki_1929]

" .....forgive me for being captain obvious, but my old A7M266 board runs just fine with XP 2000+....they can continue dropping the price on these suckers so that eventually I can max it out with 2600+s and also plop two MP 2600+s in my A7M266-D.....I don't even use half the capacity now...I'll be blown away when I plop those in for $50 each in a year or so.....keep predicting the demise AMD, it's all fine with me."

Wow, I'm not quite sure there are words to describe that kind of pretentiousness. Perhaps Ellison-like? Stallmanesque? I'm just not sure... To envision the entirity of a multi-billion dollar industry as a function of your own specific needs is just... truly beyond words. Wait a moment - are you Bill Gates? I mean, he has billions of dollars of buying power, and could very well shape the path of the microprocessor market for a couple of years.

You'll forgive me if I don't think the death knell has been sounded for AMD's 64-bit CPUs because of your apparent lack of immediate need.

Re:You know.....

[Eccles]

I didn't say yield, I said the chips/wafer increases. Since the cost of manufacturing a wafer of a given size is pretty constant, more chips/wafer means cheaper chips if your yield does not decrease, or if it only decreases by a small enough amount.

die size for a given design does not scale with gate size as you think it might.

I agree that is an oversimplification. (Esp. since things like pads don't change in size.) Picking the one example I could find quickly on the internet, a decrease from .18 u to .13 u (.72 as wide) was expected to decrease an Athlon from 128 sq mm to 80 sq mm (.625 as big). So you would get 1.6 times as many chips/wafer if the yield is the same, less if the yield decreases.

(See this, near the end, for the numbers provided above.)

Re:You know.....

[Anonymous Coward]

Sure, but for some reason he was too lazy to come up with a correct example. He should be dinged for that.

Re:You know.....

[spektr]

All signs point to it being cheaper to buy a dual opteron over the not so long term simply for the power saving features at idle. I figure that the new machine would pay for itself in less than a year.

How do you calculate this? Even if you could save more than 200 watt consumption this would be less than 2000 KWh per year, which costs less than $200 in my country. The slowest Opteron available at this time costs more than that.

dual Athlon MP, which uses about the same amount of power at idle as at full utilization

I measure a 20 watt difference per CPU (MP 2000+) between idle and full utilization at my workstation. I wish it was more, but it is definitely a difference.

Re:You know.....

[randyest]

OK, you didn't say yield, but the OP did, and then I did when helping him understand that the actual relationship of yield to process geometry is opposite of what he claimed. And then you interjected "the greater number of chips/wafer almost always more than compensates" in a discussion about shrinking geometried increasing yield. What, exactly, was supposed lead me to think that you were not saying:

the greater number of chips/wafer almost always more than compensates [for yeild decrease due to increased likelihood of fatal defects]?

And, if you weren't trying to say that, what were you trying to say? That cost per die would go down? That doesn't necessarily follow from migrating to a more advanced process, either. In fact, 90nm is still usually more expensive for most designs than 0.13um and in some cases 0.15um.

Moreover, the cost of manufacturing a wafer of a given size are absolutely not anything like constant. Most of the cost of manufacturing is in the investment in the fab itself and the cost of the masks. Building more advanced fabs (or modding existing ones to make more advanced technologies) is much more expensive. Making masks for smaller geometry processes is much, much more expensive! Both of these costs are paid for by the cost of the dice made.

BTW, note that your example cited an expected rather than actual die size reduction. I'll bet the actual wasn't even close. Those numbers came from a marketdroid spewing self-aggrandizing estimates, not the guy who hastoimplement it (i.e, me).

Finally, most fabs are shifting to larger wafers along with the switch to 90nm, so we don't even have a constant wafer size here :)

Re:You know.....

[ePhil_One]

You can't sell an AGP video card for $30 MSRP

Why not? GeForce 64meg w. tv-out are retailing for about that now.

By "YOU" the poster means a manufacturer. Clearly, you can sell old hardware that is taking up space in your warehouse for $10, because the alternative is to pay to have them landfilled. But even then "YOU" don't do it, you find a guy who speicalized in out of date hardware whos got the deep discount channels in place and "HE" sells it for $30. The original poster's point is spot on, it doesn't pay to manufacture a retail $30 video card

Re:You know.....

[Anonymous Coward]

quake 3?

Re:You know.....

[Hoser McMoose]

And just how many 386 chips do you think AMD or Intel makes?

Simple fact of the matter is that a company only produces the previous generation of processor for so many years after the new one comes out. AMD is simply saying that they will stop producing their 32-bit AthlonXP chips around the end of 2005, or roughly two and a half years after they brough out it's successor. This seems like a pretty reasonable time-line.

No one other than the odd-ball Slashdot editors is predicting that 32-bit processors will cease to function or anything like that. If you bother to read the article, you'll see that the Slashdot tag line for this discussion is totally incorrect.

Re:You know.....

[Predius]

Actually, quite a few. Both have even updated their 386 models within the last few years. 386's are still considered a workhorse cpu for the embededded market.

http://www.intel.com/design/intarch/intel386/ind ex .htm and look at the active listing pdf, you can still order new from Intel 386 and 486 cpus.

http://www.amd.com/us-en/ConnectivitySolutions/P ro ductInformation/0,,50_2330_8579,00.html AMD goes further back and is still offering 80186 derived cores.

Re:You know.....

[Predius]

Yup, after some work. (Harder to do when source isn't available.) There are software OpenGL renderers that are Mesa compliant and drop in place of a hardware implimentation. The rest is asking id to not use -mcpu=pentium when comping. If they refuse, I can emulate a pent instruction set via trapping.

Slower than death as I said, but I can do it.

Re:You know.....

[tomhudson]

poster wrote:

I'm sure you're getting a super board for that price too. What's it a 1/2 layer ECS fire-hazard? Perhaps something printed out by someone's inkjet printer?

After losing a week being screwed over by an MSI board w. the nvida2 mb chipset at 4 times the price (turns out that msi doesn't have a bios that works with the Barton 2500+ at 333, you have to up the voltage (voids warranty) and use kingston memory - but kingston says they no longer support this (brand-new) motherboard, because they've had too many problems with it), I'll take 4 of the cheapies.Oddly enough, they (the cheapies) run linux withut a hitch. I'm sitting in a room w. 3 boxes - 2 of which are just slightly better than the $99.00 cheapies I'm talking about - and they give less problems than the MSI and Asus boards in the rest of the place.

Oh, and just to show that it's not JUST MSI, my sister bought an asus w. all the trimmings, and the thing is pretty lame/sick/crappy.

At this point, I'm telling people to buy the generic pc-chips shit, and, if it craps out, you've saved enough to buy something twice as good in a year, anyway.

As far as producing processors under $40, intel still does a decent busines in 80186 chips as microcontrollers at a couple of bucks a pop. Once your fab is paid for, and the R&D, everything else is gravy.

Re:You know.....

[tomhudson]

I remember when 10mbps ethernet cards were $300.00. And when 8-bit sound cards (adlib) were $350.00. Now 100mbps ethernet cards are $12.00, and you can buy a 5-channel sound card for $20.00. Clearly, the manufacturers are still making a profit, or they'd stop production.

It's not like manufacturing processes have stood still. Ever-larger-scale integration has brought the component count down to almost nothing, and ever-growing markets have continued to bring economies of scale.

The same applies to all components, from CD-Roms ($25.00, down from $700.00 a 12 years ago) to hard disks to ram to everything else. I remember paying $70.00 for a mouse that didn't last as long as todays' el-cheapo $6.00 model.

Re:You know.....

[Anonymous Coward]

Agreed, 64 bit is great but at what price? 64 bit will entice me when I can get a dual processor with mobo and memory for under $500.

Sure don't need it otherwise....

Re:You know.....

[Anonymous Coward]

The other day Bill Gates mentioned in his discussion of the next Windows XP, (longhorn) that he would like to have 6 GHZ processors available, with 3 GHZ RAM, and perhaps a terabyte of storage. All this by 2006. BTW, I looked at your site, and are you really the gal in the "Kill Bill" movie?

Re:You know.....

[shepd]

>(turns out that msi doesn't have a bios that works with the Barton 2500+ at 333, you have to up the voltage (voids warranty) and use kingston memory - but kingston says they no longer support this (brand-new) motherboard, because they've had too many problems with it)

That's odd. I've sold a lot of those boards using PC2700 RAM without any problems at all. Perhaps kingston is supplying garbage memory. I don't know. I do this without making a single change to the BIOS, except to set the CPU to the proper speed.

You must use an AGP card for video to get this to work if you are talking about PC3200 RAM. AOpen, using the same chipset, warns that you cannot use such highspeed memory and expect stability with the onboard video.

It's all pointless anyways as the speed increases are infinitessimal compared to the heartache of an unstable system.

Ask anyone selling the low end stuff their return rates and compare it to the high end parts. There is a difference. I've seen it as a retailer for both end of the spectrum.

(BTW: Some PC Chips boards that don't include hacked and pirated BIOSes have been known to have bad IDE controllers that randomly corrupt data, due to their false advertising. I've been bitten by this a few years ago. Never again.)

PC Chips mode of business:

- Fake Cache
- BIOS Hacking (as above)
- "Customizable" chipsets (translate to: We use our inhouse garbage chipset with broken IDE support and remark it to whatever you like)
- Fake CPU speeds (*STILL!* K7SOM/K7SEM users beware)
- Paper thin PCBs (my experience)
- Website served from 56k modem (or so it seems), written by ESL students.
- No support after you buy it (If you get more than 1 working BIOS update you are so very lucky)
- Above is likely due to them PIRATING BIOSes from other boards (own experience, not unusual)
- Deceiving naming of products (That's a 666 Mhz VIA board)

There's also unverified rumours that the owners of PC Chips were into CPU remarking.

Think about that next time you buy PC Chips (or, *shudder* ASRock or ECS [same company, PC Chips has a need to change their name often, I wonder why!])

Oh, and before you think I'm just doing this for my business, I try to add more margin in for the junk parts as I know I'll just need to deal with returns right away.

Re:You know.....

[shepd]

>You can get a motherboard w. duron 2000 cpu for $100 CAN (about $75.US). Split the cost in 2 - half for mb, half for cpu, we are under the $40.00 price point already.

Yes, I know about those. They're a scam, there is no such thing as a duron 2000.

They are infact Duron 1.3 processors on PC Chips (aka "ECS") boards with BIOSes hacked to display "PRO 2000+". If you boot such a machine in linux it will disregard the incorrect BIOS info and show you the true speed.

You are right, though, the spilt is roughly half and half between the board and CPU.

The common names for such boards:

K7SOM+
K7SEM

Re:You know.....

[nametaken]

I won't plug my retail pc hardware co., but we carry ATI Xpert 2000 PRO cards, and GeForce2 MX400s for around $30. Less, in the case of the ATI card. If you can't sell these items for less, you may want to reconsider your business model... if it's not too late.

Re:You know.....

[Anonymous Coward]

ASRock arn't part of ECS group you know. ASRock are actually 100% owned by AsusTek. ASRock boards are EOL Asus products.

Re:You know.....

[ivan256]

I pay $0.29/kwh

Re:You know.....

[jaoswald]

$5 isn't just a different price point---the PIC is a different beast entirely. It is dramatically biased toward low transistor count and low usage of Si real estate. Try to find a PIC with more than 8k of RAM and you'll see what I mean. You can also get PICs in 8 pin packages. I don't know very much about the fab processes for either, but I'm guessing the layers of interconnect and critical fab requirements on a PIC are very few compared to an Athlon.

A modern main processor has tremendous need for I/O and on-chip memory. Tens of millions of transistors and hundreds of I/O pins, all of which have to be packaged and tested, and every transistor is a potential failure that reduces yield. The size of the chip is huge compared to a PIC, but even if you shrink the die, there is no way to reduce the test requirements, lead count, and number of fab steps to economize for low-cost production.

The PIC is a design meant to hit a low price point. The Athlon is a design meant to hit a performance target. Even Moore's law doesn't (by itself) change the latter to the former.

Re:You know.....

[centralizati0n]

In general, $20 CD-ROM drives don't get much support from their manufacturer.

Re:You know.....

[shepd]

>ASRock are actually 100% owned by AsusTek

*Were* owned by them. Don't go about saying that!

It even threatens legal action against Asrock customers who associate these motherboards with Asustek, in a surprising example of Chinese Walls.

I have to explain to enough people bamboozled by this sort of half-truth I don't need more people saying that!

>ASRock boards are EOL Asus products.

Now that is completely wrong.

Many AS Rock boards carry PC Chips parts, part numbers, and PC Chips designs. PC Chips and ECS are pretty much the same company (or at least sell identical products, down to everything but the color of PCB), however, ECS is the "high support" (as in they can speak English properly) version of PC Chips.

The quality of Asrock boards is entirely different from the quality of Asustek boards, the memo continues.

Re:You know.....

[kesuki]

200 watts in consumption saved is indeed possible,
consider the 45 watts of the extra cooling fans, the 80-120 watts consumed the second processor, and aditional capacitors and circuitry on the motherboard... we're already at a 165 watt saving, and i'm being somewhat conservative on additional power consumption/etc. We haven't even gotten to the cost in converting A/C to DC that is lost by the PSU.. so Utilized energy savings is going to run darn close to 200 watts, at a minimum... Also 200 watts in savings, yields 1765 kwh/year in electric consumption, not the obviously rounded up figure of 2000... at $0.35 per killowatt hour that comes to $617 in annual power consumption savings.
Even if you're paying the rock bottem rate of $0.22 per killowatt hour you're saving $388 in electic consumption per year.
$200 in savings? You're paying 10 cents a killowatt hour of electricity??? Where do you get your electricity from??? a personal fission reactor??? Nuclear power is the only form of power who's short term costs can break the 10/cent per killowatt barrier nowadays... and the long term costs of waste disposal means that selling at that price would cause a crisis point whereby the money needed to maintain radiocative waste depositories exceeded the power industrys ability to fund such ventures...

Re:You know.....

[truesaer]

But is this stuff actually being MADE anymore? I'm pretty sure AMD hasn't made any Durons in a fair while. This is just the channel clearing itself out.

Re:You know.....

[Sivaram_Velauthapill]

By that line of thinking, we can run everything on the ENIAC too. Yes, the very first computer can theoretically run everything--albeit a bit slower :)

I think a better criteria is to count which applications CAN'T run on a 386. I would say most won't. Not because they do anything special, but because they aren't compiled for the 386. Programs may use MMX/3dNow instructions...

Sivaram Velauthapillai

Re:You know.....

[kamukwam]

You're paying 10 cents a killowatt hour of electricity??? Where do you get your electricity from??? a personal fission reactor???

Maybe you forgot that there are more countries on this globe than only the US. Or do you know all available electricity prices from all over the world???

Re:You know.....

[Anonymous Coward]

In the UK, I pay - at current exhcange rates, and the dollar is high currently - 12.8 cents.

So for me, that saving would be $256 (a nice power-of-two figure).

160 is less than the current bottom-of-the-line Athlon64 CPU here.... never mind the new mobo/memory.

So, I guess it depends where you live...

Re:You know.....

[Anonymous Coward]

To envision the entirity of a multi-billion dollar industry as a function of your own specific needs is just... truly beyond words.

What more is the market than many people each enacting on their own selfish needs? Go fuck yourself.

Re:You know.....

[rainer_d]

tomhudson wrote:

>> poster wrote:
>>
>> I'm sure you're getting a super board for that
>> price too. What's it a 1/2 layer ECS fire-hazard?
>> Perhaps something printed out by someone's inkjet
>> printer?

> After losing a week being screwed over by an MSI
> board w. the nvida2 mb chipset at 4 times the
> price

Well, that's why in the end, I settled for a Tyan board. High price, but at least I can get definitive lists of RAM-chips and CPUs that are supported.
And it works very nicely.
I have little sentiments for people who buy 30 bucks K7S5A boards in 50 bucks cases+PSU and wonder why it bombs out now and then.
Of course you can buy them in dozens for the same price as a Tyan Tiger + "Expensive Case+PSU" combo, but the question is: do you want to get things done on your box or is building, modding, breaking and un-breaking it its only right to exist ?

Rainer

Re:You know.....

[JamesP]

once a processor goes below about $40, it seems like it's no longer worth producing;

Well, you're really a genious... Do you know how much does it cost to make a chip (i.e. just the silicon + packaging)?

Once it goes below $40 dollars profits are almost ZERO...

Re:You know.....

[Anonymous Coward]

HOLY. FREAKING. SHIT.

Where do you live?! I pay less than 10% of that in the US. Have a Look ($0.0267/kwh if you don't want to bother RTFL)

At $0.29/kwh, it's cost effective to build your own cogeneration facility and sell power back to your local grid!!! At the very least, MOVE. You're getting screwed ROYALLY.

Some other NYS Power provider rates.. which are considered to be some of the highest in the US:

ConEd: $0.185/kwh

EnergyEast (NYSEG): $0.1267/kwh

NiMo: $0.1206/kwh

RG&E: $0.1151/kwh

data source here

-AC

Re:You know.....

[Anonymous Coward]

but it won't be remotely playable, especially if network play is involved. no, you can't do it.

Re:You know.....

[chamenos]

I have little sentiments for people who buy 30 bucks K7S5A boards in 50 bucks cases+PSU and wonder why it bombs out now and then.

its probably due to the crappy PSU. i've been using a ECS K7S5A board for 2 years now, and its been rock stable and problem free, not to mention i got it at a bargain. bios updates are readily available from the ecs website, and they came often enough.

Re:You know.....

[Anonymous Coward]

Never heard of OEM I guess huh?

Or pricewatch.com for that matter

Re:You know.....

[Predius]

I think even better is to sum up. (Programs don't NEED to use MMX/3dnow/SSE/etc, they just speed things up). Its not that 32bit machines are incapible of functioning or doing whats required, and it's not that 64bit machines provided any new functionality that cannot be replicated with older systems, the ONLY aspect that obsoletes older systems is the speed at which they accomplish the task.

So, based on that, Intel's 32bit hardware seems to be chugging right along, so who cares if they don't have 64bit on the desktop? It hasn't been outgunned speed wise, therefor it's not obsolete.

Re:You know.....

[rifter]

"You can't sell an AGP video card for $30 MSRP "

>>Why not? GeForce 64meg w. tv-out are retailing for about that now.

By "YOU" the poster means a manufacturer. Clearly, you can sell old hardware that is taking up space in your warehouse for $10, because the alternative is to pay to have them landfilled. But even then "YOU" don't do it, you find a guy who speicalized in out of date hardware whos got the deep discount channels in place and "HE" sells it for $30. The original poster's point is spot on, it doesn't pay to manufacture a retail $30 video card

Unfortunately, YOU FAIL IT. Have you even *been* to a computer store recently? Do you look at prices on the internet? There have been $30 agp video cards for years now, and every year they get better. No, for the most part, they are not old models that once sold for high dollars, but in fact are cards specifically manufactured for this space. It used to be that they were dominated by chipsets that no longer try for performance, like SIS, S3, etc, but there were also ATI boards and a few others. Now there are bargain masement board manufacturers using old Nvidia chipsets to make boards for this space, like the one the poster mentioned which is brand new with a warranty for the princely sum of $30 at Fry's.

Re:You know.....

[tomhudson]

poster wrote:

In general, $20 CD-ROM drives don't get much support from their manufacturer.

So, at that price, who gives a shit? Either it works, or it doesn't. If it doesn't work, you return it to the retailer. If it works, you keep it. It's not like you're going to have to update the firmware (unless you bought an LG and got hit w. their "new implementation" of the FLUSH_CACHE command (besides, linux drivers allow slower burners to work reliably at more than their rated speed :-)

Re:You know.....

[Eccles]

OK, you didn't say yield, but the OP did, and then I did when helping him understand that the actual relationship of yield to process geometry is opposite of what he claimed.

"Yield" as used by CPU manufacturers is not necessarily what a layman means when he uses it. In the business, it's the ratio of good chips to total chips. But I could say "one wafer yields 50 good chips on average", and most of us would understand it. So while you are technically correct in CPU tech terms, the original poster was really concerned with what matters to Athlon pricing, which is more closely related to chips/wafer than to yield as you use it.

I'll grant you my MS in VLSI is more than 10 years old, so I haven't looked that closely at the effects of process changes on yield and die sizes. I was under the impression the changes in yield aren't that dramatic because they are countered by improvements in the technology. Otherwise yields for P4s would be incredibly bad compared to the 6502 yields of 20 years ago.

As for pricing differences of making masks, running fabs, etc., there's certainly a tradeoff of higher initial costs and amortization costs. But manufacturers are making those tradeoffs -- including using larger wafers -- because they are expected to pay off.

BTW, note that your example cited an expected rather than actual die size reduction. I'll bet the actual wasn't even close.

If you have an actual as opposed to expected die size reduction for very similar chips of different wire width, I'd be interested in seeing the actual numbers. Ditto yield numbers.

Re:You know.....

[ivan256]

I don't have my bill in front of me, but I live in eastern mass. The base regulated rate set about 10 years back is $0.11, and there's a fixed percentage that they're allowed to add each year that adds another $0.10 and there's some taxes, delivery fees... It all adds up. The electric company is Keyspan, which you may notice is at the top of the list you linked to.

Yes, I know. It's rediculous, but the cost of living here isn't the second highest in the country for no reason.

Re:You know.....

[tomhudson]

poster wrote:

Once it goes below $40 dollars profits are almost ZERO...

So why are they able to churn out tons of microcontrollers for less than 410 a pop (remember, these are, in many cases, modified versions of cpus that were found in desktop machines, with improvements ranging from moving more components onto the chip (serial i/o, for example), to ruggedization).

Simple - they're making it up in volume, and the processes used are no longer cutting-edge, so yields are near 100%.

Re:You know.....

[Loki_1929]

"What more is the market than many people each enacting on their own selfish needs? Go fuck yourself."

If the posting account belongs to a union of 300,000 consumers who are in the market for a new PC within the next two years, then you're absolutely correct. That being said, no multi-billion dollar market depends on the immediate needs of a particular individual. As for the colorful language you trot out at the end, I can only assume that your lack of a well-thought-out argument necessitated an obscenity drawing attention away from it.

Re:You know.....

[mmortal03]

While I agree that they are no longer worth PRODUCING at some point, retailers still hold stock of these processors if they haven't been sold. These retailers would rather make SOMETHING on what they have already paid for, then not sell them at all.

But then there is a counterpoint to this. A couple years ago, when Pentium 4 processors had surpassed Pentium III sales, there were stores online not willing to drop their prices on the forced-to-be-expensive-by-Intel Pentium III's of 1 GHz and higher. It got to the point where even though they were not selling any of the III's, the retailers were not willing to lose any more money on them, and it got to where these Pentium III's were not even worth buying for UPGRADING (in other words, it was cheaper to buy a new mobo, ram and P4 proc).

Now I do know that Intel themselves placed higher prices on these procs, SPECIFICALLY to drive the newer Pentium 4's into the marketplace, but I really do wonder if these stores ever sold their stock of PIIIs, or if they are still stubbornly leaving them on the shelves.

Re:You know.....

[Sivaram_Velauthapill]

But your argument basically implies that technological improvements don't mean anything. MMX/SSE/etc ARE improvements. Multimedia applications for example (like games) benefit greatly from them. Architecture differences DO impact the outcome. Doing things in parallel is different. Speed matters but it encompasses many different things.

With your logic, you can reduce everything into a simple counting machine. If everything can be run on 386, why stop there? How come you don't say the 286 is good enough? A 286 with a math co-processor and a few other things are just like a 386. Why stop at 386?

Sivaram Velauthapillai

Re:You know.....

[syukton]

You're right that when a new video card hits the market and is top-of-the-line with all the bells and whistles, it won't cost $30. But to say it will never cost $30 is foolish. I believe the parent post was saying he didn't mind the technological push because the sooner something new comes along, the sooner what's new now will become passe and cost $50 instead of $200. You can get an AthlonXP 2100+ CPU w/ motherboard for less than $80 shipped. You can get geforce2 and geforce3 cards for $40 and less. If you're willing to be just a bit behind the cutting edge, it works out to be rather cost effective.

As far as a CPU for $40: this one costs $39. Yes it's a Socket A 900mhz cpu, but it only costs $39. No shipping either, get 'em while the gettin's good! (And the 900 was a good overclocker, as I recall. might be able to bump that baby up to 1.2 before it gives any fuss) Duron 1.3 GHz for $34, free shipping.... Breaking the $40 price point is a AthlonMP 1.2GHz for $44 shipped....

three perfectly nice cpus to put in a motherboard. perfectly nice.

Re:You know.....

[CmdrTHAC0]

Would the Athlon Powersaving HOWTO be of interest to you? Assuming that CPU isn't too buggy....

Measurements (with an ammeter; sorry, I'm cheap :P) for AthlonXP 2400+ system (Soyo KT333 Dragon Lite, 512MB PC2700 RAM, ATI Radeon 7500 64MB DDR video, 80GB Seagate Barracuda IV hdd):

STPGNT = 0.51 A
HLT = 0.76 A
Full utilization = 0.91 A

This on 110V US AC. I can't tell you what the phase angle is, though. Coincidentally, the difference between STPGNT and HLT is what my 17-inch LCD monitor consumes at 100% brightness.

It makes my ancient UPS (salvaged from a 486 machine) is happy....

Re:You know.....

[ivan256]

All bets are off with dual CPUs. The power saving features that matter aren't supported in an SMP configuration.

hell...YES.

[Anonymous Coward]

of COURSE 32 bit processors will be obsolete...if your 64 bit processor runs the same as your 32 bit one and has 64 bit instructions why would you still use a 32 bit ? im still waiting for a 64-bit processor in a mini-itx format tho.

Finally

[Popadopolis]

Finally the more advanced processors will become commonplace, if it is not just a rumor. Even if it is a rumor, it is bound to happen eventually.

Re:Finally

[DoraLives]

Finally the more advanced processors will become commonplace

Now just how long have people been saying this?

Re:Finally

[Popadopolis]

Well, it has been true in a sense. Our common processors today are faster than those of 15 years ago, but that isnt saying much in and of itself. The change happens, but usually slower than expected.

bah

[fjordboy]

you young-ins and your 32 bit processors...I'm using 16 bit and I have no plans on upgrading now. You'll be back...

Re:bah

[Grant_Watson]

I'm using 16 bit and I have no plans on upgrading now.

You use 16-bit processors? Back in my day we used 8-bit processors, and we liked 'em -- and that was after we came home from school, which we walked to in the snow, up hill both ways.

You should listen to Grandpa Pete tell about how he had to read his own ones and zeros! 16-bit? You have it easy.

Re:bah

[Anonymous Coward]

Not me. But my mother, now - she wrote numerical simulations of rocket trajectories in octal. She thought it was great when she got an assembler.

Back in the day, my mother was hot stuff. She had a "Q" clearance (the level needed for access to atomic secrets).

Memory? A "tray" of memory was 100 words by 36 bits. Tubes. 36 bits because it gave you enough precision (10 digits fixed point, 8 digits floating) to subtract the masses of different isotopes and see how much mass turned into energy in the process.

Re:bah

[TheOldFart]

16 bits? My word processor runs on highly compressed carbon encapsulated by cellulose. The number of bits and bytes depends on how tense I am at the time of using it.

No big deal

[Anonymous Coward]

They will just stop making them u dont have to throw away anything...

A prediction is just a prediction.

[the_mutha]

Very interesting to see this kind of statement off AMD. IMHO though 32-bit systems will never die. However, for those systems that require 32 bit operation, 32 bit emulation from 64 bit systems will suffice.

Processors are always just getting ridiculously fast, no point in developping in a realm doomed for obsolecense, if 64 bit systems are able to attend to it adequately.

Then again, there are processors for mobile phones and other mobile devices, that won't require a 64 bit processor so soon... but then again, I suppose AMD is not in that market.

Re:A prediction is just a prediction.

[Anonymous Coward]

I think you missed the point.

They are talking about normal PC processors.

Not your PDA or Pocket calculator.

AMD is in the PDA market with a few products already out. (Low power CPUs)

Re:A prediction is just a prediction.

[Chibi Merrow]

Actually AMD makes plenty of embedded chips that use both 32-bit x86 architecture or their own custom tailored architectures for whatever they're used for. So yes they are in the mobile market and actually it's a big part of their business strategy.

As far as 32-bit emulation on 64-bit chips... Intel had to eat crow and release a software emulator for the Itanium because of how lousy its hardware emulation was.
The beauty (and only thing beautiful about it) of the x86-64 architecture is that it lets 32-bit code run side-by-side with 64-bit code with no performance loss. There's no emulation or anything of that sort. And once they get their yields up, there will be no reason a Hammer based CPU couldn't be priced similarly to current 32-bit Athlons, and it will still offer superior 32-bit performance compared to the current Athlon without the 64-bit portions of the chip ever being touched.

Yes, 32-bit only CPUs will still be quite useful in the embedded market (much like the Z80 and M68k CPUs live on) but the way things are going we may be running 32-bit apps on 64-bit CPUs in the not too distant future (kinda like running 16-bit Windows 3.x on a 386)

Re:A prediction is just a prediction.

[certron]

If you want my prediction, I think AMD is announcing this because they want Intel to tip their hand. AMD can always just keep selling the 32-bit machines, they can change their projections, and this is by no means the final word on 32-bit x86...

Obvious

[Geccoman]

With the prices rapidly dropping, and the performance (for the most part) seeming to be worth the money spent, this seems like the obvious direction. As far as replacing 32 bit machines, its just a matter of time, anyway.

Learn to write

[Anonymous Coward]

Repeat after me:
It's = It is

Re:Learn to write

[Bagels]

See sig for Strong Bad's take on this :) (he does it in the form of a nice little song)

Re:Learn to write

[Kenja]

It's is it is is it?

Re:Learn to write

[Theobon]

it's can also reffer to possession. Where it owns something you would say "that is it's object" ~Never trust an AC for grammar advice.

Re:Learn to write

[Waffle Iron]

it's can also reffer to possession. Where it owns something you would say "that is it's object"

No, you would write "that is its object". The word "its" is a possessive pronoun like "his" or "hers". You do not use "'s" to make pronouns possessive.

Re:Learn to write

[Bombcar]

Are you sure?

Because we don't use:

him's
her's
etc....

Re:Learn to write

[Anonymous Coward]

Mod parent up s/he gets a gold star for knowing basic english and pointing it out to stupid people who think they know everything.

Re:Learn to write

[Ianoo]

WRONG, learn to write.

Its = possessive
It's = it is

Never trust a logged in user for grammar advice ('cept me!).

Re:Learn to write

[Wolfrider]

--Oh it is, is it?

It's... (Start Monty Python theme song)

Good news/bad news for AMD

[burgburgburg]

Good news: I agree in the quickly coming obsolescence of 32-bit chips.

Bad news: I'm getting a dual 1.8Ghz PowerMac G5, baby! Yeah!

I have to go lay down now.

Re:Good news/bad news for AMD

[e40]

Funnier when you find out your G5 is not a 64-bit machine! Ha, ha (in your best Nelson voice).

Every time I post this, I get modded down by the Mac fanatics. So, go ahead, I'm expecting it.

Re:Good news/bad news for AMD

[bnenning]

your G5 is not a 64-bit machine!

Yes, it is. The OS just doesn't support its full functionality yet. Likewise, a dual processor Xeon running Windows 98 is still a dual processor machine. If you said "A G5 running Mac OS X 10.3 is not a true 64-bit platform", then you'd have a case.

Re:Good news/bad news for AMD

[Anonymous Coward]

Yeah, right...because IBM wouldn't know a thing about what makes an architecture 64bit. As if.

Re:Good news/bad news for AMD

[e40]

That's the point. If the hardware is 64-bit and the software is 32-bit, what do you call it? 32-bit, obviously.

"Doesn't support it yet" is very, very kind. There is no roadmap that shows when they will have a 64-bit OS.

Re:Good news/bad news for AMD

[Anonymous Coward]

You're retarded. No one is talking about 32-bit oses. Is XP Pro a 64bit operating system? No. So unless you bought a workstation and it came with the 64bit version of XP, you don't have jack either.

The fucking thread is about processors. G5, 64-bit processor. The roadmap that shows when they will have a 64-bit OS is next year. How do I know this? Common sense. The OS isn't even utilizing the power it has available to it, and it still owns xp.

Re:Good news/bad news for AMD

[Anonymous Coward]

You get modded down because you make irrelevant points and fallacious arguments. It is a 64-bit machine. The physical machine, the processor. The OS may not support 64-bit at this time, but that has nothing to do with the *machines* potential.

What you're saying is that a stock BMW is not capable of reaching a speed higher than 129 mph, because at that point a governor will kick in, and bring its speed down 10 mph. Bullshit. The bimmer is capable of reaching higher speeds; other factors (governor) preclude it from doing so.

Same with the G5, it is capable of 64-bit mode, but the software isn't there yet.

Anyway, fuck you, you're another person who thinks you're always right, and that anyone who disagrees with you must be blinded by brand loyalty....

You should be modded down, your thoughts are not worth the energy spent to read them.

Re:Good news/bad news for AMD

[Anonymous Coward]

"Yes, it is. The OS just doesn't support its full functionality yet. Likewise, a dual processor Xeon running Windows 98 is still a dual processor machine. If you said "A G5 running Mac OS X 10.3 is not a true 64-bit platform", then you'd have a case."

Ah hell, I'll jump in this one just for the sake of ticking off Mac Users with no sense of humour:

You enjoy your 64bit processor with 32 bit OS. I'll enjoy my dual Opteron 2.2ghz(which by the way, I built with all the bells and whistles for 800 bucks less than the dual G5 listed on Apples site) with Suse 64. Yeah, your procs are faster than Pentium 4s, but not faster than my Opterons, and especially not faster than my opterons with my 64bit OS biatches!

Mac Users can be pretty smart, when they're not trying to talk shit about the ULTIMATELY UNQUESTIONINGLY SUPERIOR PC platform.

Hehehe. Hey, your face is red as an Apple! Hehehe!

Re:Good news/bad news for AMD

[Anonymous Coward]

Well, I hate to tell you, but your G5 with OS-10 will spend about 15% of its time executing EMULATED 68K code!

Seems the folks at Apple never did finish a true port, and large chucks of the OS date back from the OS 4-9 days.

Re:Good news/bad news for AMD

[Anonymous Coward]

Is the G5 hardware really capable of 64 bits? I own a mac se/30. The hardware supposedly was the first 68k mac that supported 32 bits, but in reality the firmware had some 16-bit code.

Re:Good news/bad news for AMD

[Anonymous Coward]

You're a fucking moron. If you have a 64-bit processor but can't use but 32-bits of it, what's the fucking use saying you have a 64-bit processor? Get it now Mac fanboy?

Re:Good news/bad news for AMD

[Anonymous Coward]

There's no roadmap showing when LINUX will be 64-bit either! Hahahahahahahahaha! And the GPL is unconstitutional and designed to destroy UNIX!! You fools must die.

Re:Good news/bad news for AMD

[Anonymous Coward]

Please give the keyboard back to your Daddy. The grownups want to talk now.

Massive gains in cooling tech?

[PierceLabs]

Unless AMD or someone else has a massive gain with respect to being able to cool these monster CPUs along with shrinking the die so that they are suitable in general consumer electronics, I don't think anyone is going to stop producing smaller/cooler CPUs. While the 64 bit chips are great and all, I just can't imagine seeing one in a phone (for example) or even a PDA in the timeframe that they suggest.

Maybe they meant to include that they won't go into these markets which limits their desire to produce low footprint, low heat chips.

Re:Massive gains in cooling tech?

[Anonymous Coward]

Slashdot headline is (once again) misleading. AMD predicts that THEY will stopping manufacturing 32bit PC prosessors (Athlons) by the end of 2005.

AMD has already low-power chip for PDAs and it won't be 64bit anytime soon.

Re:Massive gains in cooling tech?

[Anonymous Coward]

64 bit processors aren't necessarily hot. some of the the mips processors run pretty cool.

Re:Massive gains in cooling tech?

[Naffer]

There's been plenty of gain in cooling technology. It came out a bit less then a hundred years ago and it's called a refrigeration. Several companies offer phase-change cooling systems for CPUs. I wouldn't be surprised if we have to go that way in the future.

Re:Massive gains in cooling tech?

[sydbarrett74]

You're absolutely right. 32b processors will continue to enjoy a healthy uptake in the embedded market. In fact, many embedded companies are still selling 8b and 16b processors en masse.

Re:Massive gains in cooling tech?

[The_K4]

There's one disadvantage to most of these: condensation. While this can be controlled (and somethimes even used) it's also very bad when your processor gets wet!

Re:Massive gains in cooling tech?

[tloh]

Let us not forget the good folks at Transmeta who are attacking this problem from the software perspective. More efficient code will make any processor run cooler, regardless if it is 64 or 32. However AMD decides to go about it, they probably don't intend these new chips to be used in situations where cooling comes at a premium. In those kinds of circumstances, PDAs for example, better software is a simpler solution.

Re:Massive gains in cooling tech?

[Anonymous Coward]

When they shrink the Opteron to 90nm in the spring of 2004, the chips will use only 45 watts of power unlike Intel's upcoming Prescott which will use 90-105 watts.

Re:Massive gains in cooling tech?

[ComputerSlicer23]

I'm not sure precisely what you mean about "consumer electronics". You can always underclock a chip to the point that it won't generate that much heat. I've heard of people doing that with really an "old" 1.8Ghz, and underclocked it (like you would with a laptop), way, way, way down, and used it as a part in a no moving parts router. (I've got a machine that is a powersupply, and a CPU fan away from a no moving parts CPU). I just haven't taken the time to find a 150Watt powersupply, and underclock the 1.4Ghz processor down to 200Mhz which should be way more then fast enough to handle my dinky routing needs.

Kirby

Re:Massive gains in cooling tech?

[Anonymous Coward]

Actually, the headline (and story in infoworld, and possibly the actual statement by AMD) is even more misleading than that, it's AMD's CPG division that thinks it will stop producing 32-bit machines. I'm just guessing the PCS division (which does the low power MIPS and Geode x86 CPUs) won't abandon 32-bit so quickly.

Re:Massive gains in cooling tech?

[PierceLabs]

If you're going through all this trouble to down clock a processor, wouldn't it make more sense to just use something that runs at a lower clock? Its like putting a Farrari engine in a car that is only ever intended to move at 35 MPH - sure you could do it, but wouldn't it make sense to just get an engine more suited to the task.

Re:Massive gains in cooling tech?

[pmz]

Unless AMD or someone else has a massive gain with respect to being able to cool these monster CPUs...

For quite a long time, now, Sun, for example, has been cooling it's CPUs using the full-size case fans and ducts to guide the air over their generous but modest heatsinks. Even the Blade 2000 with dual UltraSPARC IIIs just has a case fan and a duct. Take a look at the PowerMac G5; it's the same thing.

On the lower-end, Sun's UltraSPARC IIi and IIIi CPUs are given little heat sinks and teeny fans.

It seems to be only the PeeCee crowd that wants to attach a half-pound chunk of aluminum to their CPUs with dual fans, etc. I wonder if it's related to phallic underachievement, you know, the computer geek equivilent of those super-long slender dragsters with the 3000HP at the base or perhaps like a 300-foot-tall five-million-pound-thrust rocket whose only job is to lift three 200lb men and cargo into orbit...

Re:Massive gains in cooling tech?

[Anonymous Coward]

The adolescent processors will grow pubic hairs which will interfere with the cooling of the core. This is what makes them useless.

Aw, fuck, obsolecence

Re:Massive gains in cooling tech?

[ComputerSlicer23]

Depends on volume. Intel is famous for at various points in their history for producing the same chip, or making very minor adjustments to it, and selling it as 4 different chips. Because it's cheaper to run 1 line of 4 times the volume, then it is to run 4 production lines.

You only have to do Q/A once. Every Engineering tweak you figure out directly applies to all of the chips. You don't have to worry quite as much about not having enough production capacity to make type A chips, because type A has a lot more demand then was originally predicted.

It's my understanding that overclocking sites list which CPU runs of old Celeron's were just sold under a lower speed CPU and could safely be overclocked well outside of the range Intel labelled it as.

There's a reason you get a huge discount when you buy in volume, for similar reasons, it's much cheaper to produce in huge volumes. That's how Intel became the 800lbs gorrila in chip making.

Kirby

Re:Massive gains in cooling tech?

[Rob Simpson]

Well, if no makes such a processor that will fit a modern motherboard socket, then underclocking is the only option. Of course, there's also VIA CPUs, if you either have an old socket 370 board or want to buy one of their embedded boards. Presumably, the reason the AMD cpu was used was because they had an old system sitting around and decided to make it fanless.

Re:Massive gains in cooling tech?

[IM6100]

Some chip vendors are still selling millions of 4-bit processors per year. You'd better be ready to order at least a half million or so for them to mask your code, of course.

Re:Massive gains in cooling tech?

[trolleri]

Under-clock/volt - let us say - an athlon cpu at 1.4GHz,1.70V to ~400MHz,1.2V and you land on a maximum cpu heat dissipation of about 15W.

Now pick a CPU from
http://users.erols.com/chare/elec.htm
with a similar heat dissipation and let us have a match!
How about that juicy Cyrix M1-PR90+ (80MHz)? ;-)

Re:Massive gains in cooling tech?

[bhtooefr]

They make the Geodes? Holy shit! I thought the Geodes were Cyrix derivatives! (In case you didn't know, Cyrix blows chunks - Cyrix 3 (aka VIA C3/Eden) doesn't count, as it's a WinChip, which sucks, but not NEARLY as badly)

Re:Massive gains in cooling tech?

[Anonymous Coward]

> I wonder if it's related to phallic underachievement

Or it might be because your customer is going to stick a 2003 CPU into his vintage AT case that he bought for $58 in 1989 and has absolutely terrible airflow characteristics. Which is why Intel is pushing the BTX standard, etc.

Nah, forget that, the small penis explaination makes much more sense.

Re:Massive gains in cooling tech?

[drinkypoo]

The low-power CPUs are usually low-power in both senses of the word; they may use less power, but they also do less per clock cycle. Your non-low-power CPUs like the Athlon XP or P4 simply do more per clock cycle than, say, VIA EPIA, which isn't even superscalar.

The Athlon XP does more per cycle than the P4; the P4 does more per cycle than the C3. Think of the C3 as a 1.5 liter toyota motor, the Athlon XP as a V8, and the P4 as a turbo straight six, if you must apply automotive metaphors. The C3 is small and efficient, but it simply cannot do much. The Athlon is big and maybe inefficient, but it can do a lot. The P4 is also big and maybe inefficient, but it can do a lot too - However it gets its power from RPMs and not from displacement.

If you are going to be running a lot of processes at once, it makes the most sense to have a P4 or Athlon XP, not a C3 or an EDEN; Especially with P4's hyperthreading, if you are in a position to use it. Regardless the significantly superscalar nature of a non-budget CPU is often well worth it.

Re:Massive gains in cooling tech?

[Hoser McMoose]

It is a Cyrix derivative, though a few generations removes. Cyrix got bought out by National Semi a handful of years back ('98 if my memory is working correctly). After two years of mismanagement, they were mostly sold off to VIA, all except for the Geode division. NS kept on developing the Geode and selling it into set-top boxes with little to no success.

Then a couple months ago, they sold that division off to AMD. I suspect that the transfer of people, technology and whatnot is still a work-in-progress, but practically speaking, the Geode is now an AMD product.

To the best of my knowledge AMD plans on doing a fair bit of overhauling of the design, but to keep the basic philosophy (ie low-power, low-cost, heavily integrated x86 chip).

Re:Massive gains in cooling tech?

[jafac]

My dual G5's CPU and cooling takes up approximately half the interior volume of the case.
While it's a very slick-looking case, and very easy to get into, and very easy to do simple things like add RAM or a hard drive - there's a lot of other tasks (like adding more than one hard drive) that are nigh-impossible. Tasks that Apple has deemed, you really oughtn't to be messing with. But I digress. The point is; the dual G5's are frickin huge.
iMac? Non-trivial - but those Apple guys are pretty sharp.
Laptop? I don't think so. They'll need some help from IBM (ie. a *different* chip! most likely binary compatability, with a different capability)
PDA/Cellphone? ya right.

Re:Massive gains in cooling tech?

[drinkypoo]

Once upon a time, internal combustion engines tended to be air-cooled, because it was much simpler. As they became more powerful the need to dissipate additional heat led to water cooling becoming the norm. Computers are now not entirely unlikely to to go water cooling as a matter of course. How long will it be before someone sets out to aggressively create a set of standards (heh heh) for water cooling equipment? This could lead to retail CPUs with water fittings instead of fans and heat sinks, or so I would like to hope. If you got creative with your packaging you could come up with a sealed liquid cooling rig, but that I don't see happening. Processors moved to a larger module rather than the smallest possible package size which will reasonably be swappable by the user (as they are currently sold), and they moved away again - for a reason. That being, it didn't really provide any real value to anyone. Jerking around with the slot type CPUs was a big waste of time. I suppose it was all a plot to put out chips with lots of cache more cheaply, but it didn't end up being all that cheap for anyone.

I don't think AMD is going to stop selling their 32 bit processors entirely. I'd think they'd turn it into an "embedded" product which was probably about half the speed and a quarter the power consumption of their hottest shit, and not make it any faster. That way, you are still motivated to buy the 64 bit stuff, and they can move everyone further away from 32 bit, but they can probably make a lot of headway against intel's ARM-derived embedded processors. They currently sell assorted low-power x86 processors (Geode), so it's no stretch to guess that (at least) parts of the K7 core will make it into such a product soon enough.

Re:Massive gains in cooling tech?

[truesaer]

They were obviously referring to desktop chips. AMD has a separate division for embedded devices. And incidentally, the opterons are cooler than Athlons I believe.

Re:Massive gains in cooling tech?

[Anonymous Coward]

"t seems to be only the PeeCee crowd that wants to attach a half-pound chunk of aluminum to their CPUs with dual fans, etc. I wonder if it's related to phallic underachievement, you know, the computer geek equivilent of those super-long slender dragsters with the 3000HP at the base or perhaps like a 300-foot-tall five-million-pound-thrust rocket whose only job is to lift three 200lb men and cargo into orbit..."

I'm so sick of hearing this. Look, my cock is the size of Mount Everest and is it such a crime to want a PC that is too?

Geeze...

adaptability

[DenOfEarth]

As a user of open source, I think this shouldn't really be a problem at all, should it? I mean, once gcc can compile 64-bit code, than we should simply be able to recompile all of our current apps for these new processors, shouldn't we? I'd be happy if someone out there could point me out as not being in the know...

Re:adaptability

[RevMike]

As a user of open source, I think this shouldn't really be a problem at all, should it? I mean, once gcc can compile 64-bit code, than we should simply be able to recompile all of our current apps for these new processors, shouldn't we? I'd be happy if someone out there could point me out as not being in the know...

The Linux-on-Alpha project already did all the heavy lifting needed to run 64 bit almost ten years ago. Linux and *BSD is already running 64 bit on PowerPC. Virtually any package you can download that has an active support community is already 64 bit ready.

Re:adaptability

[Neon Spiral Injector]

That pretty much is true, along with the fact that GNU/Linux has been running on other 64-bit platforms for many years now.

The biggest problem I'm having building a custom install for my AMD64 machines is the fact they have 32-bit compatiblity. On the Alpha is was easy, the lib directories contained 64-bit libaries, because the machines were 64-bit, period. But with the AMD64 the lib dirs are still supposed to contain the 32-bit libs, with the 64-bit versions installed in a lib64 variant. This causes problems because almost no libary packages are setup to compile twice once in 32-bit and then over again in 64-bit mode with a simple ./configure.

Re:adaptability

[Anonymous Coward]

I am not sure what you mean :
1) Gcc 3.3 actually compiles to AMD64 (i.e. Athlon 64 and Opteron)

2) Gcc 3.3 generates pretty fast code too, as you can see on spec.org where IBM submitted results obtained with an Opteron, Suse Linux and GCC 3.3 : the baseline is above 1100 whereas Apple said that a Pentium 4 at 3.06GHz only achieves 880 with GCC 3.3. So it seems that the Opteron is a better processor for people who use GCC.

3) I write this text on my Athlon 64 running Mandrake 9.2 RC1 for AMD64. I can tell you that there is litlle left to adapt because it works pretty well. In fact you could hardly tell the system is the AMD64 version...

4) running 32 bit programs works fine : I have to use Java 1.4.2 for Linux x86 since the AMD64 version is not available. To do that, you just run the 32bit Linux program and it works at native speed.

As a conclusion, the Athlon 64 is good but it is still expensive (I paid 450 euros for my processor).

Re:adaptability

[jwsd]

A C pointer will grow from 32-bit to 64-bit. This will wreak havoc in most non-trivial C/C++ apps.

Re:adaptability

[Eric Sharkey]

On the Alpha is was easy [] because the machines were 64-bit, period.

This is not strictly true. I believe Linux on Alpha is purely 64-bit, but other operating systems (such as OpenVMS) allow programs to be compiled to use either 32-bit or 64-bit virtual memory spaces. Link

Re:adaptability

[budgenator]

SUSE LINUX 9.0
Full 64-Bit Power for Home Users - Only with SUSE LINUX 9.0 for AMD64 see SuSE for more details. It's a bit pricey at US$119.95.

Re:adaptability

[lederhosen]

No, most distributions will use both 32 and 64
bit libs, Debian does so on SPARC and is going
to do it on AMD64 as well.

Re:adaptability

[DenOfEarth]

I guess I was just wondering how much work it would take to change all the software that I like to use to a 64-bit platform. I do a fair bit of compiling in my day-to-day operation, so I can see how a well written C or fortran code could be compiled for either platform without much hassle, I was just wondering if there were some things that required a bit more finagling to get working...that's all

Re:adaptability

[DenOfEarth]

right, this is what I'm talking about, but would this even screw things up if you compiled the entire system for 64bit? I also wonder if the speed gain from upgrading to a 64bit processor won't actually really happen until most codes you want to run fast are written specifically for a 64bit architecture.

Re:adaptability

[eatdave13]

Only if they make assumptions about the size of a pointer.

Even if the program is sticking pointers into ints there still shouldn't be a problem since int will be 64 bits anyway.

In C++ it really shouldn't be a problem with the much stronger type checking.

The only place I see it being a problem is in over-optimized programs or assembler routines.

Re:adaptability

[Anonymous Coward]

I guess I was just wondering how much work it would take to change all the software that I like to use to a 64-bit platform.

To switch to AMD64, none - your existing 32-bit code will run on it. To recompile for 64 bits, well... if the programmer has done something stupid like rely on the size of a machine word, then you might encounter hard-to-fix bugs, but software that sloppy is probably pretty buggy in the first place.

Re:adaptability

[AnotherBlackHat]

I mean, once gcc can compile 64-bit code, than we should simply be able to recompile all of our current apps for these new processors, shouldn't we?

Yeah... Right.

Of course we should, but are you willing to bet that nobody ever assumed that "long" was a 4 byte quantity when they were allocating memory?
That everybody used ~0x4f instead of 0xffffffb0?
That (n>>24) must fit in a byte?

-- this is not a .sig

Re:adaptability

[Anonymous Coward]

> something stupid like rely on the size of a machine word ... but software that sloppy is probably pretty buggy in the first place

A lot of GNU software had that problem, until Linux was ported to the Alpha and other 64bit platforms.

Re:adaptability

[Anonymous Coward]

Kewl, I want some!

Soo.. when can we expect a 32-bit port of the 64-bit version?

Re:adaptability

[randombit]

Of course we should, but are you willing to bet that nobody ever assumed that "long" was a 4 byte quantity when they were allocating memory?
That everybody used ~0x4f instead of 0xffffffb0?

Sure they have. Except hopefully by now somebody has tried to use their code on Alpha, or SPARC64, or PPC64, or MIPS64, or any of the other 64-bit CPUs out there, and complained that it didn't work. Either they fixed it (no problem, then), or they didn't (in which case they're both lazy coders and bad maintainers, and so it'd be best to stay away from their stuff anyway).

Re:adaptability

[hughk]

The word length was 64-bits but the address space could be either 32 or 64-bits. Earliest OpenVMS/AXP implementaions came from the VAX so it was logical to start with a 32-bit address space. However Digital's Unix went 64-bit pure quickly as did Linux.

If you have source code, adapting for sizeof * = long long rather than just long wasn't a major problem but it could be tedious.

Re:adaptability

[Anonymous Coward]

Nowadays, if there is no need to optimize for memory and/or speed, then it is likely coded in Java/C#/VB/Perl/Python/Shell scripts. I doubt 64-bit will benefit those programs very much. The fact that you have to go for C/C++ means you have to care about pointers and bits and bytes. More importantly, you have to care about memory size. Doubling the size of pointers and integers can have dramatic impact on the memory used by such programs. Most memory intensive enterprise softwares are database related. If a server needs 4 Gig memory, then it is likely a database server. Database softwares can have millions of pointers in memory because they are typically organized as tree structures. Doubling the size of pointers can needlessly double the memory usage of such programs.

Re:adaptability

[eatdave13]

That is true. I may be wrong, but I thought the parent was talking about actual compilation/runtime errors.

I hope the editors realize...

[AKAImBatman]

I'm not in a hurry to ditch any of my 32-bit machines, so long as I get them replaced by 2038.

I hope the editors realize that 32bit processors CAN process 64 bit numbers. In Java, for example, the date is handled by a 64bit number that tells the number of milliseconds since Jan 1, 1970. Amazingly enough, that clock won't run out for another few billion years.

Oh, and most Unixes have fixed the time problem. The real issue is getting the programs to switch over to the new APIs.

Re:I hope the editors realize...

[the_mutha]

No, the REAL issue is memory space. 32 bit just won't cut it anymore for large database servers and the like, regardless of the movement for clustering.

The thing is, in the desktop it will take longer for machines to require more than 4/8 GB of memory.

Re:I hope the editors realize...

[AKAImBatman]

Yes, that's the REAL issue for switching. The REAL issue for time is simply to get programs to switch over to using 64 bit time APIs. Nothing to do with 64/32 bit processors. Sorry if I was confusing on that point.

Re:I hope the editors realize...

[Quarters]

Machines need >8GB of memory to handle the 64bit number clock code? Crap!

Re:I hope the editors realize...

[mindstrm]

Not the same thing, sorry

a 32 bit machine working on 64 bit numbers involves multiple iterations of the same units... obviously a computer can work on arbitrary sized bit fields.. it can do 900 bit math, too.....

but at it's heart, it's doing 32 bit stuff.

Java might do this, but most unix clocks are 32 bit, and roll over in 2038.

Re:I hope the editors realize...

[Anonymous Coward]

I found one application that seemed to have no problem with 2038. It's death year was 2108. It also used 32-bit times from 1970, but it declared them unsigned.

Re:I hope the editors realize...

[ryanw]

No, the REAL issue is memory space. 32 bit just won't cut it anymore for large database servers and the like, regardless of the movement for clustering. The thing is, in the desktop it will take longer for machines to require more than 4/8 GB of memory.

Hum, seems like several companies have handled the memory issue. How the heck are they getting away with 64GIGS of memory on some of these 32bit servers?

Re:I hope the editors realize...

[Anonymous Coward]

8-bit processors can process 64-bit integers... add with carry and all that... In fact, you can write routines on 8-bit processors to do 256-bit arithmetic or even higher!

The thing is whether or not they can do it in a "reasonable" amount of time :)

Re:I hope the editors realize...

[BagOBones]

Lots of fun little tricks like paging and such..

Tests have shown that 64 bit addressing is faster than 32bit/paged methods.

Re:I hope the editors realize...

[AKAImBatman]

> Hum, seems like several companies have handled the
> memory issue. How the heck are they getting away with
> 64GIGS of memory on some of these 32bit servers?

It's called segmented memory. Remember the old DOS days when programs couldn't address more than 64K of memory without adding special "Large Model" addressing? Well it's baaaackkk! No program can access more than 4 gigs of RAM without being compiled specifically to bank switch. Thus, you can have multiple programs that can all run with up to 4 gigs of RAM, but none of those programs will be able to process more data than that. 64 bit processors make this easy by allowing flat addressing of >4Gigs. Just like how 32bit processors allowed addressing of >1024K (minus the ~300K that IBM "reserved").

Re:I hope the editors realize...

[Anonymous Coward]

Yeah but most 64bit chips have 128bit wide registers, they just can't use them for addresses. v-9 sparc for example.

A number of 32bit machines have 64bit wide registers, again can't use them for addressing but can store and manipulate data in them.

Re:I hope the editors realize...

[AKAImBatman]

a 32 bit machine working on 64 bit numbers involves multiple iterations of the same units... obviously a computer can work on arbitrary sized bit fields.. it can do 900 bit math, too.....

but at it's heart, it's doing 32 bit stuff.

Not true. Most modern processors have silicon dedicated to processing 64 bit numbers. In fact, most processors can do 128 bit numbers thanks to SMID instructions (e.g. SSE, 3DNow, Altivec, etc.)

The current definition of a 64 bit processor is one that can use 64 bit memory addressing.

Re:I hope the editors realize...

[damballah]

output from my box: Tue Jan 19 03:14:01 2038 Tue Jan 19 03:14:02 2038 Tue Jan 19 03:14:03 2038 Tue Jan 19 03:14:04 2038 Tue Jan 19 03:14:05 2038 Tue Jan 19 03:14:06 2038 Tue Jan 19 03:14:07 2038 Fri Dec 13 20:45:52 1901 Fri Dec 13 20:45:52 1901 Fri Dec 13 20:45:52 1901 Linux t 2.4.21-0.13mdk i686 unknown unknown GNU/Linux

Re:I hope the editors realize...

[AKAImBatman]

Welcome to Linux hell. You might try a Solaris or FreeBSD installation.

*ducks*

Re:I hope the editors realize...

[pmz]

The thing is, in the desktop it will take longer for machines to require more than 4/8 GB of memory.

Recently, it seems gaming consoles lag PCs by about five to seven years (PS2: 300MHz MIPS + 32MB RAM; XBox: ~700MHz + 64MB RAM; GC 400MHz? + 48MB RAM).

It would be pretty cool if, in 2012 or so, we had 2.5GHz+ consoles with 4GB+ of RAM (or even SMP if that Cell marketing fluff is true).

Re:I hope the editors realize...

[Short Circuit]

At the time, that reserved memory went to good use. Video memory and system applications were expected to go there. (This was before the days of on-board video memory.)

Re:I hope the editors realize...

[AKAImBatman]

> (This was before the days of on-board video memory.)

Incorrect. EGA and VGA cards had onboard memory that was *mapped* to the A000 and B000 ranges. That memory was reserved because IBM didn't want to deal with direct I/O port communications in software. The only part that sucked was that they used so much memory. I mean really, did they need to whole B000 range for text mode graphics? And why couldn't we use the same memory for graphics and text?

Of course, some memory extenders would try to detect what video mode you were in and make use of the extra memory. :-)

Re:I hope the editors realize...

[sql*kitten]

The thing is, in the desktop it will take longer for machines to require more than 4/8 GB of memory.

Yes and no. There are plenty of scientific/engineering/financial apps now that can take advantage of as much memory as you can give them on the desktop. Sun sells a dual-processor workstation with 8Gb of memory for about $6000. People are using them right now - you can't (yet) buy a PC that can do that at any price.

Re:I hope the editors realize...

[EinarH]

Several smaller companies have started to deliver workstations based on AMD Opteron with up to 16 GB ECC RAM.

For example Boxx; www.boxxtech.com

Re:I hope the editors realize...

[Cyno]

The current definition of a 64 bit processor is one that can use 64 bit memory addressing.

So.. a processor that can understand 64-bit numbers.

Current CPUs do math, much like old CPUs. We're smart enough to get them to do 64-bit math even though they wouldn't understand it very well. Just like we got them to do decimal based math with binary numbers. I think its an Alan Turing thing.

Re:I hope the editors realize...

[IM6100]

You bring up nostalgic memories of trying to squeeze out that extra 16K of RAM in DOS...

The good old days. My father's first IBM-PC had 64K on the motherboard. Four rows of 16Kx1 DRAM chips. Only the first row was soldered in.

Re:I hope the editors realize...

[AKAImBatman]

>> The current definition of a 64 bit processor is one that
>> can use 64 bit memory addressing.
>
>So.. a processor that can understand 64-bit numbers.

Do you have any idea what "64 bit memory addressing" means? It doesn't mean math. It means "more pigeon holes to put stuff in". There are two things stopping modern processors from going to 64 bit addressing:

1. Comatibility
2. More complex and expensive memory controllers, motherboards, bus designs, etc.

The silicon on the chip is the least of our concerns.

Re:I hope the editors realize...

[bhtooefr]

It's an IBM PowerPC in the GameCube. Keep in mind, the PS1 used a 33MHz MIPS (1MB RAM?), and could kick 133MHz/16MB RAM PC ass when it was launched. The PS2 competed against 800MHz/128MB RAM PCs in it's day. Consoles are actually GAINING ground on PCs again. (keep in mind, the NES was a 6502@1MHz versus the Apple II line with 65(C)02@1MHz CPUs)

Re:I hope the editors realize...

[AKAImBatman]

Dare I mention EMM386 and Himem.sys? Or the LIM (Lotus, Intel, Microsoft) specification? Or the 286 mode and triple faulting? Or -

...

You know, now that I think about it, it's a wonder I didn't hate computers back then.

Re:I hope the editors realize...

[drinkypoo]

Am I missing something? I thought that the Athlon 64 and Opteron were currently limited to 48 bits of address space, for packaging reasons? As far as I can tell, a computer is n-bit when it has a word size, data path, and address space of n bits.

Re:I hope the editors realize...

[Anonymous Coward]

The thing is, in the desktop it will take longer for machines to require more than 4/8 GB of memory.
Not the way I write code....

Re:I hope the editors realize...

[G-funk]

No, the REAL issue is memory space. 32 bit just won't cut it anymore for large database servers and the like, regardless of the movement for clustering.

Every intel chip since the PPRO can handle 64gb of ram

Re:I hope the editors realize...

[AKAImBatman]

> I thought that the Athlon 64 and Opteron were
> currently limited to 48 bits of address space, for
> packaging reasons?

The design can address 64 bits. It's physical ability to support memory is limited by the motherboard. For example, the Intel Pentium architecture can address 4 gigs, but consumer motherboards tend to have maximums of 512MB to 2Gigs.

Re:I hope the editors realize...

[Anonymous Coward]

Current AMD64 chips can only support 48-bit addressing, no matter which motherboard you use.

Hopefully people have wised up and won't write software that uses those extra 16 bits for something. (See old Mac 24-bit software).

Re:I hope the editors realize...

[AKAImBatman]

Actually, the processors in home consoles have traditionally sucked. What makes them useful (and impressive) is the video hardware that does all the work. Let's take the Super Nintendo as an example. This machine could barely manage a few megahertz, yet it was scaling/rotating/blitting out hi-res graphics with millions of colors. How did it do it? Well, there was Sprite based hardware that could take a section of memory and generate a video overlay signal directly off that memory. The processor simply had to say "move it here" or "scale it by this much" and the hardware did it. There was no "blitting" as we think of it on PCs.

Of course, this meant the hardware was limited to the maximum number of sprites that could be handled in hardware at a given time. In fact, You may have noticed an artifact of this approach when too many sprites got on the screen and all the characters started flickering. Still, the results were impressive and PCs couldn't hold a candle.

Now, when 3D graphics came along, PCs (for the first time ever) ended up using the same video hardware as consoles. Consoles still had some advantages in the area of special optimizations and platform consistency, but the performance gap was narrowing.

What do we have today? PCs dressed up like consoles. (i.e. XBox)

Re:I hope the editors realize...

[Anonymous Coward]

processors CAN process 64 bit numbers. In Java, for example,

OH, SHUT THE FUCK UP!@!@!@!@!@!@!@!@!@!

Yes, you're really a genious for pointing that out...

Re:I hope the editors realize...

[Nexx]

And those of us who wanted to run Falcon 3.0 (required 600k free conventional memory!) installed QEMM.

Shortly afterwards, a bunch of programs started putting themselves into protected mode. I remember writing a program to see how much conventional memory I can consume before things started going hairy.

Re:I hope the editors realize...

[Anonymous Coward]

Ah, for the days of QEMM, Stacker, and PC Tools for Windows 2.0 (still a better GUI than windows xp, despite being over a decade old), all running on a 486 with 8 megs of ram and a 450 meg disk.

I still remember the hard choices I had to make, disk-wise. Ultima VII, or Word 2.0? What a mindbender!

Not that things have changed all that much. I still have to pick and choose what I put on my 20 gig disk. And word hasn't really improved all that much (for me) since version 2.0.

Re:I hope the editors realize...

[Anonymous Coward]

OH, SHUT THE **** UP!@!@!@!@!@!@!@!@!@!

Yes, you're really a genious for pointing that out..

Awww, aren't you cute? Little trolly-wolly, oopsy-poopsy. Run along now, and tell your parents I said Hi!

Re:I hope the editors realize...

[bill_mcgonigle]

Sun sells a dual-processor workstation with 8Gb of memory for about $6000. People are using them right now - you can't (yet) buy a PC that can do that at any price.

That's not such a bad deal - a G5 with 8GB will run you about $5700. Probably a better option than the Sun for a desktop PC, but $3200 of that is RAM - those big sticks are still ludicrously expensive.

Of course, I'm just waiting for the reply claiming that you can get a Gateway for $699 that can do the same. ;)

Re:I hope the editors realize...

[hplasm]

Eek! A Miscrofoft Coderr...bzzt!

Re:I hope the editors realize...

[Anonymous Coward]

Have you tried compiling GNOME lately?

Re:I hope the editors realize...

[Rakarra]

I still remember the hard choices I had to make, disk-wise. Ultima VII, or Word 2.0? What a mindbender!

I remember Ultima VII was more than just a pain when it came to disk. It was incompatible with any memory manager, including himem.sys. You needed a seperate boot disk just for it so it could do all its own management.

Re:I hope the editors realize...

[hplasm]

Still waiting for my Math Coprocessor to arrive before I try that one.. :+|

Ohh yea?

[Naffer]

a tactic to try to capture the 64/32 bit standard leaving Intel in the rear

I was sure that Intel had announced plans to integrate it's 64bit "Yamhill" extensions into the Prescott chip which should be launched early next year. I think AMD is trying to wish it away.

Re:Ohh yea?

[paitre]

Nope.
Besides, the Yamhill extensions are supposedly AMD64 compatible...even Intel doesn't want to piss of the Beast of Redmond.

Re:Ohh yea?

[Loki_1929]

"I was sure that Intel had announced plans to integrate it's 64bit "Yamhill" extensions into the Prescott chip which should be launched early next year."

They haven't, and they wont, unless the Athlon64s begin to eat away at too much of Intel's market share. Intel's problem is that if it releases 64-bit extensions in Prescott, it will be forced to do so for the Xeon line as well. The problem with this is that it would send 10 years of research and development, along with countless Billions of dollars down the drain. Itanium relies on one thing and one thing only: a need for 64-bit processing. The few Itaniums that are selling will be the only ones sold if Intel's customers can get 64-bits on a Xeon.

Unless Intel is forced to, it's simply not going to do "64-bits on the cheap". Intel has nothing to gain by announcing or implementing 64-bit extensions now; we don't even know if the Athlon64s are going to sell. On the other hand, they have everything to lose if they do open up Prescott to 64-bit quickly. What Intel is probably doing now is scrambling to come up with ways to sell Itaniums to its customers once 64-bit Xeons are available. Itanium's miserable sales to date show that it has enough trouble just selling it as-is. Take away the one advantage it does have, and they may as well take it off the table now.

Aside from that, we have absolutely no idea how well Prescott's 64-bit extensions perform. We have no idea if they've been perfected, nor what kind of problems Intel has with them. We have no idea if bugs exist in the instructions, nor what sort of benefit we'd see from them. They could very well be based off of AMD's own 64-bit extensions (Intel has licensed the technology), which would make it rather embarassing to release them to the public. It would also constitute a huge and embarassing shift in position on the topic of 64-bits for home; something Intel has publicly stated is not something the market currently needs.

I'm not saying Intel wouldn't possibly announce 64-bit instructions in the near future - it could be tommorrow for all I know. All I'm saying is that everything points to Intel keeping the 64-bit extensions under wraps for as long as absolutely possible. As of right now, the only 64-bit instruction sets we know of (when it comes to Intel/AMD), are EPIC for Itanium and the ones from AMD. It's doubtful that Intel has secretly developed a whole brand new instruction set just for the P4, and it's even more doubtful that they somehow rigged the P4 to use Itanium's 64-bit instruction set. Thus, we're left with Intel banking on limited adoption of AMD's 64-bit CPUs, which does not appear to be the case thus far.

P.S. All signs point to Intel releasing the first handfull of Precotts around Feb of 2004, with volume closer to early Q2. Solving voltage leaks that push your operating temp to extremes isn't something you throw a bandaid on; it's something you design around. I just hope, for Intel's sake, that they aren't rushing Prescott out the door as they have on other chips (P3 1.13GHz).

Re:Ohh yea?

[AKnightCowboy]

I was sure that Intel had announced plans to integrate it's 64bit "Yamhill" extensions into the Prescott chip which should be launched early next year. I think AMD is trying to wish it away.

You know, I haven't met anyone recently that has gone out of their way to buy an Intel processor. They may have been stuck with it in their desktop, but everyone I know that has built a system puts an AMD Athlon in it. I guess it could be price, but I think a little bit of it has to do with giving Intel the old "fuck you" for selling us massively overpriced chips all these years. It's ridiculous that they expected mid-range chips to go for $300-$400 just a few years ago while AMD was selling the same processor power for a third of that.

Re:Ohh yea?

[Naffer]

Is the Itanium's (Itanic?) 64 bit setup faster or better then AMDs? For that matter, is WindowsXP 64bit compatible with BOTH 64bit chips or only with the AMD ones?

Re:Ohh yea?

[Loki_1929]

" Is the Itanium's (Itanic?) 64 bit setup faster or better then AMDs?"

Most of the performance comparisons you'll find would be at Ace's. The big problem with Itanium (aside from the price), is the hit you take when you try to do something with 32-bit code. Itanium requires heavy optimization on all fronts before it begins to perform reasonably well. Opteron will run whatever the hell you throw at it, and it'll run it quickly. Aside from that, why drop $10,000 on an Itanium when a faster Opteron can be had for $1,000? AMD's 64-bit platform was never really designed to compete with Itanium anyway. The Athlon64 competes with the higher-end P4s and low-end Xeons, where it offers excellent performance and 64-bit technology. The AthlonFX competes with low-end to high-end Xeons in workstations and gaming machines (think EE) where it, again, offers comparable or higher performance while sporting 64-bit instructions that allow for things like more addressable memory. The Opteron competes with mid to high-end Xeon single and multi-processor workstations and servers. Opteron also competes somewhat with the low-end Itanium market; those who want/need 64-bits but can't really afford Itanium's price tag. Once you get into the higher end of Itanium's spectrum, there's plenty of competition from Sun, IBM, and others. Opteron might do better here once AMD starts throwing multiple cores on each chip, but probably won't even try for these types of customers (except with clusters) with this generation of chips(K8/K9).

"For that matter, is WindowsXP 64bit compatible with BOTH 64bit chips or only with the AMD ones?"

My understanding is such that M$'s 64-bit Windows is still in development for both chips, but that the Itanium version is set to be released sooner. In any event, they're supposedly specifically optimizing 64-bit Windows for the AMD 64-bit instruction set, which should give us some outstanding performance boosts. Look for the 64-bit Windows for AMD64 to be released to the public some time around Q3 '04.

Re:Ohh yea?

[Anonymous Coward]

What fucking mess.

Too bad I'm missing out.

http://www.apple.com/powermac/

Re:Ohh yea?

[bhtooefr]

Actually, I am currently in the process of going out of my way to buy an overpriced Intel CPU and an overpriced RadiSys board. It's called a Pentium M 1.3GHz and a RadiSys LS855. I want the low heat output/power consumption, and the high performance per watt/MHz. Look at benchmarks of Pentium M laptops versus Pentium 4 or Athlon XP M laptops. I want a friggin Pentium M desktop!

Re:Ohh yea?

[Hoser McMoose]

The latest and greatest Itanium2 (running at 1.5GHz and with 6MB of L3 cache) is sometimes faster than AMD's Opteron and sometimes slower when both are running natively compiled code. In generally they are pretty close. This is actually quite a coup for AMD, since the Opteron is a much smaller and cheaper chip. The fact that they're able to throw down with the big boys (Intel's Itanium2 and IBM's Power4+) is pretty darn impressive.

For WindowsXP, right now the only 64-bit version of Windows XP is for Intel's Itanium instruction set (IA-64). Microsoft does plan on producing a version of WinXP for AMD's 64-bit instruction set (AMD64, aka x86-64), but it has been delayed multiple times (typical Microsoft) and now isn't expected until mid-2004.

I want to see 64bit software compared

[Preach the Good Word]

I read several of the reviews, and all stuck with 32 bit code for the comparition between the Intel P4 and the AMD Athalon 64. Linux runs on the Atahlon in 64 bit mode, wouldn't be hard to compile PovRay and Doom on a 64 bit compiler and see if anything changes. Thats just for an easy test.

Many real world (science?) applications benifit from 64 bit processors, find some (presumably running on UltraSparc, PPC, Alpha, or such) and port them over to see how the 64bit abilities of this chip compares to the other existing chips.

I run open source OSes, and open Source applications. I don't care about 32 bit performance because I'm fairly sure that if I did have an Athalon 64 I wouldn't run 32 bit code very often. I can choose between many chips, compatable instruction sets to me means gcc (or other compiler) has an output for them. 32 bit x86 compatiabily is nice for the few times I have to run something 32 bit (normally in Wine) and that doesn't happen very often.

Re:I want to see 64bit software compared

[Anonymous Coward]

ATHLON. No second A... Argh...

Re:I want to see 64bit software compared

[Gothmolly]

It's Athlon, not Athalon, or Atahlon. You have an extra A, A-hole. You can't spell the word 'benfit' (sic) either, and Sparc is actually "SPARC" (an acronym).

Perhaps you're a poorly written trollbot?

Re:I want to see 64bit software compared

[Hoser McMoose]

There are some benchmarks available at the Spec website. You have to do a little bit of digging, but AMD has submitted a lot of results with identical systems running 32-bit and 64-bit Linux.

On average, running AMD64 code will buy you about a 5% boost in performance over running IA-32 code on the Opteron/Athlon64. Of course, some applications will see a much larger improvement (some apps are more than twice as fast), while others will actually be slower (64-bit pointers = more data to load from memory and fewer pointers that can be stored in cache). Note that this actually has just about nothing to do with 32 vs. 64-bit for most cases, but rather that AMD64 doubled the number of general purpose registers (16 vs. 8 in IA-32). Considering that the lack of registers was one of the last major downsides to the x86 instruction set (as compared to something like PowerPC, MIPS, SPARC, etc.), this was a very good idea. Of course, even 16 visible registers is rather low, most other ISAs have 32+ visible registers.

Agreed.

[OtakuHawk]

I plan on having an Old PII or Cyrix still availible, to run old 16-32 bit games/programs. It's too much of a hassle to get them to run on XP/2000.

Re:Agreed.

[BagOBones]

Not a Game Boy but you can get them on Gamecube
http://www.gameshark.com/gamecube/articles/431995p 1.html

Why buy 32 now, though

[RevMike]

I'm not in a hurry to ditch any of my 32-bit machines, so long as I get them replaced by 2038

True, but it is not reasonable to think that two years from now someone would choose to buy a new desktop machine that is reasonably powerful without it being 64 bit. By that time the price points for 64 bit will be down substantially.

I do believe that 32bit will have a somewhat longer life in the laptop market, as well as the low power/small footprint niche of the Mini-ITX form factor.

Re:Why buy 32 now, though

[Lawrence_Bird]

I think AMD is right on the mark. There is no rational reason why given a choice between 64/32 and 32 you would want, or should, choose the 32. By this time next year prices of the lower end Opterons should be approaching $100 as production ramps up, which is certainly competitive with anything in the only 32bit world.

Re:Why buy 32 now, though

[budgenator]

I'm in the market for a new machine to replace my 700Mhz athlon and have been seriously looking at AMD64, mobo, case, PS, 2 gig ram come in at about the same as I paid for my 286 system many years ago ($1250, it amazes me how 'puter have stayed at about the same price), an opteron for about $1550 well dressed

Re:Why buy 32 now, though

[ThaReetLad]

Except that there are already Athlon 64 laptops available, and next year AMD plans to produce 64 bit chips which dissapate just 30 watts of heat, compared to the 70-90 of todays chips. These will make mobile 64 bit computing a serious possibility, even for those on a budget. Compare this to Intels plans for 100 Watts of heat waste from their next range of chips and you begin to see that AMD's future is very bright.

Obligatory Bill Gates quote...

[techstar25]

"No one will ever need more than 32 bits of processing."

Re:Obligatory Bill Gates quote...

[Anonymous Coward]

Yea, I guess you're right. The Digital Equipment PDP-10's were 36 bits. That sure didn't get them anywhere.

Go AMD!

[geeveees]

I have always been an AMD fan, and have never bought an Intel cpu in my life. Today I am glad that I didn't give those Intel ***** my money:

I got an MCE error ("Bank 2: 940040000000017a") on my linux box today, I want to look it up, but I have to subscribe to their online support thingy (http://idl.books24x7.com/toc.asp?bookid=6290)... Fine, I go ahead but it turns out that in the 'trial' subscribtion they replace all valuable info with "xxxxxx". GRRR that gets under my skin!

And did you know AMD is sending free amd64 arch manuals (4 nice books) freely to anyone in US and Canada? Did you?

Vote with the wallet!

Re:Go AMD!

[psykocrime]

And did you know AMD is sending free amd64 arch manuals (4 nice books) freely to anyone in US and Canada? Did you?

Yep, mine came in yesterday!

In Intel's defense however... I could have (almost) sworn that (at one time at least) they had freely available PDF's of all their manuals available for download. Maybe that's no the case anymore, or maybe my memory is playing tricks with me... but I sure thought they used to be available...

Re:Go AMD!

[AKAImBatman]

> And did you know AMD is sending free amd64 arch
> manuals (4 nice books) freely to anyone in US and
> Canada? Did you?

Did you know that Intel had free manuals years before the AMD64 was introduced? Did you!? And did you know that they are NICE manuals that are carefully packed, unlike my AMD64 manuals which are flimsy and arrived with water damage? Did you!?

Re:Go AMD!

[geeveees]

Did you know that I had this problem *now*, not years ago. Somehow I don't think they'll send a free manual all the way to europe when their damn site isn't even free!

Re:Go AMD!

[happyfrogcow]

water damage? c'mon now, they just shipped you free water, and you're complaining?! sheesh.

(sarcasm... shhhh)

Re:Go AMD!

[AKAImBatman]

> c'mon now, they just shipped you free water, and you're
> complaining?!

Two words: Wall Drug

Re:Go AMD!

[xkenny13]

And did you know AMD is sending free amd64 arch manuals (4 nice books) freely to anyone in US and Canada? Did you?

Yep, mine came in yesterday!

Well, I didn't. Where does one sign up for the free manuals? Got a link handy?

Muchas Gracias...

Re:Go AMD!

[haruchai]

Here is the link to get those AMD64 manuals:
http://www.amd.com/us-en/Processors/DevelopWithAMD /0,,30_2252_875_4622,00.html
If you're located outside the US and Canada, you can download them from here:
http://www.amd.com/us-en/Processors/DevelopWithAMD /0,,30_2252_739_7044,00.html

Re:Go AMD!

[Anonymous Coward]

Hmmm...www.intel.com, Support & Downloads, Processors, an IA-32 processor, manuals...there it is!

What happens when I type the name of the book into Google search...first link...there it is!

Maybe Intel is keeping this information from Europeans? There is a link for international downloads on their site but maybe it doesn't work.

Re:Go AMD!

[psykocrime]

Well, I didn't. Where does one sign up for the free manuals? Got a link handy?

http://www.amd.com/us-en/Processors/DevelopWithAMD /0,,30_2252_875_4622,00.html

16-bit? 16-bit?

[burgburgburg]

You're soft. Why, in my day, we sent rockets to the moon with 8-bit processors, and we considered ourselves lucky to have them.

16-bit? Why don't you just go lay down on a feather bed and let servants peel your grapes for you? Harumph.

Re:16-bit? 16-bit?

[MAXOMENOS]

You're soft. Why, in my day, we sent rockets to the moon with 8-bit processors, and we considered ourselves lucky to have them.

Rate this comment "+1, True and Ironic"

Re:16-bit? 16-bit?

[pvt_medic]

And didn they just come out with some ethernet cards for the commodore 64. Which way are we going 64 bits or 8 bits.

Re:16-bit? 16-bit?

[Anonymous Coward]

What's ironic is that NASA was in better shape back then! It's a joke now.

Re:16-bit? 16-bit?

[Anonymous Coward]

Big deal!

In my day we had 1 bit processors. But it wasn't even a bit. You see, we didn't even have binary, all we had were zeros!

Re:16-bit? 16-bit?

[JordanH]

Ironic maybe, but not true. I don't know for a fact, but I would guess that very few or no 8-bit processors were involved with sending men to the moon. The 8008, which was the original 8-bit microprocessor, came out in late 1972. I doubt that any of these were used by NASA during the Apollo program, which ended in late 1972.

There might have been some odd 8-bit machine that I don't know about that was used in NASA, but I would expect the bulk of the computing work for the Apollo program was done by Minicomputers and Mainframes which had basic addressable memory sizes of 12 or more bits.

Re:16-bit? 16-bit?

[Waffle Iron]

Looks like the Apollo guidance computers were 15-bits. From a random Google hit:

Each computer had two types of memory, erasable and fixed. The fixed memory contained the programs, constants and landmark coordinates using 36,864 terms or words, each of 15 bits length. That came to a grand total of 74 kilobytes of memory. The erasable memory, which was used to store variable data used in calculations or as registers for logic operations, had only 2,048 15-bit terms.

I remember reading elsewhere that the 36 Kwords of ROM were hard coded by hand threading the bit patterns with tiny wires and magnetic cores, and then they were sealed in a block of epoxy. Turnaround time to fix bugs took weeks.

Re:16-bit? 16-bit?

[SoSueMe]

In my day we had 1 bit processors. But it wasn't even a bit. You see, we didn't even have binary, all we had were zeros!
That would explain the relative value of most A/C posts.

Re:16-bit? 16-bit?

[Jaysyn]

Damn, I feel really spoiled now.

Jaysyn

Re:16-bit? 16-bit?

[shut_up_man]

Listen you cocky young whippersnapper, in MY day we broke the Hun's codes with 1-bit difference engines the size of buses, running on vaccuum tubes infested with poisonous moths bigger than your hand. We entered our programs with sledgehammers, completely from memory, during air raids, in the dark, while we were completely drunk!

And you tell the kids of the day that, and they don't believe you...

Re:16-bit? 16-bit?

[frobisch]

The board computers (two identical, one in CM and one in LM) on the apollo had 5760 logical gates and the word-size was 16 bit.
There are two interesting pages about computers in manned and unmanned spaceflight:
Computers in Spaceflight the NASA Experience
and a german page
Computer in der Raumfahrt

... maybe not for embedded processors though

[questionlp]

I doubt if we will see all 32-bit processors cease to be produced considering that the overhead required may not be feasible for embedded or real-time environments. For a lot of purposes, mostly ones where power consumption is critical, 32-bit ARM, MIPS, PowerPC and other embedded processors are more than enough... considering that a lot of those applications don't really need to address anything beyond 4GB of memory, nevermind having 4GB of physical and virtual memory in the first place.

Even if 32-bit processors will cease to exist in the desktop, notebook, workstation and server markets in 3 years, I bet that there will still be a lot of 32-bit only applications will still be made and supported for a long time (look at 16-bit Windows applications that are still in use today).

a word from the Processor Growers Association

[G4from128k]

I wonder what applications will drive the adoption of 64-bit computers? Besides playing the latest games, most real-world applications seem to run fine on older 32-bit processors, even sub-GHz processors. AMD's prediction is self serving.

That said, I have my eye on a new dual G5, so I guess I've bought into the hype that size (word size) matters.

Re:a word from the Processor Growers Association

[MysteriousPreacher]

I'd have thought that memory hungry applications are the ones are going to help give 64 bit processors a push. 4GB of RAM just isn't as much as it used to seem.

Re:a word from the Processor Growers Association

[Anonymous Coward]

We have 5 app servers running 4 gigs of RAM right now and we'd love to be able to give them more RAM. Why do we need so much damn memory? The whole system is written in VB for christ's sake and I'm not kidding or trolling or flamebaiting ; I'm dead serious. 400 megs worth of compiled VB running on 5 app servers. Sometimes it makes me sick just coming into work.

Re:a word from the Processor Growers Association

[WoTG]

I think 64 bit will take root at the servers and quickly force 64 bit processors out to the workstations. Even now, 4 GB of RAM in a server is a constraint in many businesses, and it will be much more pronounced in only a couple years.

Given this situation, Microsoft is bound to polish and market 64 bit Windows Server (which will soon be available for x86-64) because they know that 64 bit Linux distros are out or will soon be out. MS's place in the server world is far from secure, so it's in their interest to be as competitive as possible in this regard.

So, with servers (and high-end workstations) moving to 64 bit really soon, and the fact that one of x86's strengths is in volume manufacturing, the natural step (especially for AMD) is to move all chips to 64 bit in a fairly short time frame. Besides, the incremental cost per CPU is minimal, AMD claims somewhere around 10% die space. And, it takes more money to design and manufacture separate 32bit and 64bit chips than it does to sell a single product line.

I guess I don't think that any "killer app" is really required above and beyond what is out there already. Big fat database, email, and application servers that could use more than 4 GB of memory NOW.

Re:a word from the Processor Growers Association

[G4from128k]

Servers and workstations have been available in 64-bit versions for more than a decade with Alpha, R4000, SPARC, and other high-end machines. Computers with more than 4 GB of RAM are common in enterprise systems. Intensive applications have already moved to 64-bit long ago. We don't need new processors to address this issue (although cheaper processors are always appreciated by the CFO). This move was directly driven by the specialized applications used in enterprises and workstations.

Since I don't see a need for the average consumer or corporate PC user to host a 1,000 user SAP instance or do real-time fluid dynamics simulations of nuclear explosions, it seems less likely that they will "need" 64-bit computing. You don't need more than 4 GB of RAM to get e-mail, browse the web, write a memo, or edit a 5-megapixel image (frankly, 128 MB should be more than enough, unless your OS & apps are bloatware).

That said, one could argue that people don't need massive 4-wheel-drive SUVs, but that doesn't stop them from buying them. And if, as you suggest, the incremental cost of 64-bit is neglible, then the move will occur anyway.

Perhaps you are right, 64-bit is inevitable. If so, I predict a massive windfall for distributed computing projects like SETI@Home when home PC users find their CPU utilization in the single digits. Either that, or the folks in Redmond better start dreaming up new ways to soak up all that extra RAM and CPU cycles.

Re:a word from the Processor Growers Association

[TFloore]

Multimedia will encourage 64-bit CPUs without much of a problem. It isn't required, by any means, but it makes things a lot simpler.

Take a 10gig MiniDV file off your camcorder and edit it on your computer. You as a user don't care how it is accessed or used, but the programmer will be a *lot* happier to just memory-map the entire file, and let the OS only actually load those pieces needed as they are needed.

It isn't required by any means, but it makes a lot of things simpler.

Any time you mess with larger-than-2gig data files, you want a 64-bit cpu.

Oh, and...

Perhaps you are right, 64-bit is inevitable. If so, I predict a massive windfall for distributed computing projects like SETI@Home when home PC users find their CPU utilization in the single digits.

64-bit doesn't really have anything to do with CPU utilization. A blazingly-fast 32-bit CPU, and a blazingly-fast 64-bit CPU, are both blazingly fast. Both are still overkill for just reading email, browsing static web pages, and writing memos. 64-bit doesn't necessarily mean fast, it just means (usually) a wider memory bus.

Re:a word from the Processor Growers Association

[pueywei]

However, I would think that 64bit implementations of distributed computing clients would be much faster than the current 32bit clients.

hrmmmm I don't have to worry about bits.

[zaqattack911]

I'm betting by 2005 my 32bit cpu well become self-aware and upgrade itself.

Re:hrmmmm I don't have to worry about bits.

[Shazow]

Spontaneously, right?

- shazow

Re:hrmmmm I don't have to worry about bits.

[Anonymous Coward]

Since you are already self aware (don't argue) how are you doing on your upgrading process from neurons to positrons?

Sorry, just had to ;)

Re:hrmmmm I don't have to worry about bits.

[sharkey]

Perhaps it'll hit puberty, and its bits will get bigger.

Re:hrmmmm I don't have to worry about bits.

[Anonymous Coward]

Hasn't worked for you, though, has it?

Re:hrmmmm I don't have to worry about bits.

[ForestGrump]

No, only AMD procs become self-aware and evolve into 64 bit.

-Grump

Re:hrmmmm I don't have to worry about bits.

[pipingguy]

I'm betting by 2005 my 32bit cpu well become self-aware and upgrade itself.

"Hi, I'm Chippy, your processor assistant. Would you like me to upgrade myself to 64bit at minimal cost? Please select your preference below:

Yes / No (Self Destruct Option)"

Re:hrmmmm I don't have to worry about bits.

[Anonymous Coward]

I'm betting by 2005 my 32bit cpu well become self-aware and upgrade itself.

"Upgrade" as in "Replace"? You mean it's going to commit suicide?

What kinds of programs are you running there anyway?

32-bit processors and 2038?

[Anonymous Coward]

Is the poster suggesting that 32-bit processors won't be able to use 64-bit variables to store the date?

So.. 32-bit processors will be useless past 2038? Is he serious?

What a moron...

Re:32-bit processors and 2038?

[Anonymous Coward]

They won't be useless, but 2038 is 35 years from now. Take a look now at what computers from 35 years ago look like. Now think again.

Stupid me!

[spidergoat2]

Once I predicted the end of 16 bit chips. Then I found one in my PDA...and my cell phone....and my clock radio....and my DVD player.....and......

Re:Stupid me!

[Anonymous Coward]

I bought a cheap PDA a couple of years ago and found it had a Z80.

Re:Stupid me!

[freeweed]

And 8 bit chips still outsell any other category, by a huge margin.

AMD isn't claiming NO 32 bit chips will be around, they're talking the home desktop market. I'm pretty sure no one still sells 16 bit desktops these days :)

Re:Stupid me!

[rocket97]

My PDA is powered by a Dorito.

Great...

[mikesab]

35 more years of the Y2K38-bug-apocalypse-is-coming hype.

2038 is a problem?

[fullofangst]

I'm not in a hurry to ditch any of my 32-bit machines, so long as I get them replaced by 2038 By 2038 I suspect what we know as 'computers' will not be measured by something so simplistic as bits ... :)

Re:2038 is a problem?

[Anonymous Coward]

They said the same thing 35 years ago, in 1968. Bits are fundamental to information theory.

Re:2038 is a problem?

[Rob Simpson]

Plus, I plan on dying around then anyway, so if it is a problem, it's Somebody Else's.

Re:2038 is a problem?

[ThaReetLad]

Quite right. By 2038 computing power will be measured in qubits

maybe.

maybe....

[mwilli]

I don't think that they are predicting that all 32-bit processors will be phased out by then, but will probably not be manufactured anymore. It's the next logical step in the progression of microprocessors. 32-bit will phase out just like the 16-bit, whether it will be in 2 years or no we will have to see, but my comp is good for at least 5 more years as is.

Re:maybe....

[bhtooefr]

Oh, they'll still be manufactured. Cheapo toy PDAs and games run on 4-bit CPUs, graphing calculators run on 8-bit Z80s, older Palm OS PDAs run on 68000-based 16-bit Dragonball CPUs (Palm JUST finished switching to 32-bit ARM-class CPUs), and 32-bit will live on for years to come. 8-bit CPUs are the most common CPUs today. They were introduced DECADES ago, and phased out in PCs a mere 10 years ago (when Apple discontinued the Apple //e), and they're still extremely popular. 50 years from now, my watch will have a NetBurst-based Watchium 4 in it. Socket 478 watch CPU with a huge heatsink, anyone?

hmmm

[tomstdenis]

end of x86 processors maybe. Last I checked the 32-bit ARM cores were not only speedy but took quite a bit less power than the average x86.

Tom

Re:hmmm

[Detritus]

HP's latest hand-held calculators use ARM cores and are powered by AAA batteries.

Re:hmmm

[Anonymous Coward]

There's nothing like having HP use a 75MHz ARM processor only to emulate a 4MHz Saturn processor. Talk about a waste of power. Seeing it do 1000! is kind of neat, though.

How Long Until 128-bits?

[Nom du Keyboard]

Draw the line. In little more than 25 years we've gone from the 4004 (4-bits) through 8-bits (8008 and successors) to 16-bits, 32-bits, and now 64-bit microprocessors. Why stop now. We already have 128-bit and 256-bit wide data and memory busses. How long until a true 128-bit microprocessor?

Re:How Long Until 128-bits?

[tomstdenis]

Most of it is hype. The biggest problem aside from lack of address space is the fact that the x86 instruction set sucks and was never meant to run this fast. That's why you get a 10x or more difference for power consumption of ARM and x86 cores.

128-bit ALU's won't be useful at all, ever. Aside for things like bignum math co-processors you don't need it. Heck for the most part 32-bit ALU's are over kill. For example, when I hit submit on this form it will prolly strlen() the buffer. Chances are I won't write more than 64KB so why do you need 32-bits to count that?

Tom

Re:How Long Until 128-bits?

[Short Circuit]

You're forgetting about floating-point numbers. ALUs already process more bits than are contained in the data types, just so they can be more accurate. The more bits you have in a floating-point piece of data, the more decimal places you can accurately represent.

This is very important in physics simulations, which is important to scientists and game developers. Having a floating-point physics model has several advantages (large/detailed maps, accurate motion calculations).

Re:How Long Until 128-bits?

[tomstdenis]

Generally this is true but no so much for games as SIMD is [e.g. SSE/3DNOW]

You could take the ARM route and just make things modular. E.g. a ARM7TDMI is based on v4 of the ARM instruction set. An ARM11 device would be based on v5 which sports floating point [and simd later on afaik].

The point I was trying to make is that the x86 instruction set is what will die, not 32-bit ALU's.

Tom

Re:How Long Until 128-bits?

[Anonymous Coward]

Graphic processors (GPUs) already use wide data paths, but mainly for SIMD. The problem with ever-wider datapaths is that you don't really need them, as each extra bit gives you 2x the address space and magnitude of numbers of computations... it's an exponential growth.

Meanwhile, the physical resources required to engineer the wider datapaths require bigger and bigger chips, which is undesirable for any number of reasons.

If you don't need it, there's no point going to wider sizes, especially (in the case of general purpose processors) you need to retrofit existing software.

Think about this: we went from 4-bits to 16-bits in under 10 years, but 32-bit computing has been with us for almost 20 years, and it still has a lot of headroom left.

Re:How Long Until 128-bits?

[NerveGas]

ARM cores and P3/P4/Athlon/Opteron cores are so incredibly different that comparing them is simply stupid.

Your ARM core has what, 36,000 transistors? Try adding on a floating point unit, for starters. Then add in MMX, SSE, and SSE2 instruction sets. Then see how much power it consumes.

ARMs are great when you need a low-power, low-speed, integer-only processer. Unfortunately, when you need more muscle, an ARM just won't cut it.

steve

Re:How Long Until 128-bits?

[jonbryce]

Doesn't one of the games consoles already have it?

Re:How Long Until 128-bits?

[tomstdenis]

Dude, shut the fuck up you stupid fucking idiot and fucking read the point .... fuck...

The point is the ARM ====>instruction set===== is vastly superior. I've heard rumors that Intel toyed with the idea of a Ghz ARM core. So it's totally possible to add 512KB of cache, an FPU and such. What you will end up with is a core that still takes less power, has a richer ALU and generally cooler.

Let's take some comparisons. x86 has all of 4 registers. You can muster up 3 more if you don't like debugging. The SIMD stuff is mostly just adhoc added on bs. ARM has IIRC 16 registers [1 for PC, 1 for SP].

The x86 set favours memory operands. ARM favours bundling [e.g. you can rotate/shift inside an instruction].

The ARM instruction set also sports generic multipliers [e.g. not just to one pair of registers]. It's multiplier is very fast [comparable to that of the Athlon, way faster than the P4].

blah blah blah.

Tom

Re:How Long Until 128-bits?

[Wesley Felter]

Address space demand grows at about 1 bit per year. Today's biggest computers have 1TB of memory (40 bits), so they'll exhaust 64 bits around 2027. Desktop computers might not need to make the switch until around 2037.

Re:How Long Until 128-bits?

[Scott Wood]

Dude, shut the fuck up...

My, we're feeling mature today.

The point is the ARM ====>instruction set===== is vastly superior.

It's a bit better then x86, but it's not that wonderful, and it's not the reason that ARMs are so low powered.

I've heard rumors that Intel toyed with the idea of a Ghz ARM core.

It's quite likely; they've got 733MHz XScales now. However, don't be fooled into thinking that they perform anywhere near what a superscalar, out-of-order 733 MHz x86 CPU would achieve. Things like that are where the bulk of the power consumption comes from.

x86 has all of 4 registers. You can muster up 3 more if you don't like debugging.

8, not 4. One is for the stack, but you counted that for ARM as well. One (not three) more is for debugging, and is not necessary if you have sufficient debugging information outside the code.

ARM favours bundling [e.g. you can rotate/shift inside an instruction]

Yeah, I'm sure that's conducive to high-speed implementations.

The ARM instruction set also sports generic multipliers [e.g. not just to one pair of registers]

So does x86 these days. It's been there since at least the 386 (maybe earlier); try to keep up.

blah blah blah.

Indeed.

I'm no particular fan of the x86 instruction set, but you could at least damn it for the things it is guilty of.

Re:How Long Until 128-bits?

[Scott Wood]

Heck for the most part 32-bit ALU's are over kill. For example, when I hit submit on this form it will prolly strlen() the buffer. Chances are I won't write more than 64KB so why do you need 32-bits to count that?

The odds are pretty low, but not zero. Thus, the code needs to handle that case (especially since it's most likely generic code that is not written specifically for Slashdot), and to do that, it can either use 32-bit integers, or implement bignums.

Guess which is faster?

Re:How Long Until 128-bits?

[reflective recursion]

I'm not sure 64-bits will ever be exhausted. If it does, I'm sure it won't be for at least 50 years or more. Disk space and RAM are growing, but the thing is, software need is growing ever so slowly. We pretty much have the gaming issues resolved (seperate "GPU"). Databases probably use the most resources, and I really don't see the need for more than 64-bits addressing here. If you have a database that can grow to the size of 1 exabyte (nevermind an entire _16_ exabytes), I'd say you have enourmous redundancy in the database itself. I can't even fathom how one would go about collecting that amount of information. I'm not sure high resolution graphics of the entire surface of each and every planet in our solar system at a scale of 1 pixel per 1mm would fill a database of that size. We are talking _enourmous_ amounts of data here.

And, of course, if you have an extremely specialized database like the above, you would simply seperate the database into more manageable sections over multiple machines, etc. No need for more than 64-bit here, either.

Re:How Long Until 128-bits?

[Hoser McMoose]

How long? A LONG time I would guess.

First off, the definition of a bit-width has changed somewhat from the days of 4, 8 and even 16-bit chips. Back then you were normally talking about the size of your registers and the width of your data bus. Now we're mainly talking about the number of bits you can directly address in a flat memory space (ie the size of your pointers). So the change hasn't been quite as dramatic as you suggest (though it certainly has been dramatic).

However, with 64-bit chips, we can now directly address a LOT of memory, 10^19 bits worth. We're really going to have to make some fairly major changes to our current microprocessor paradigm before we get to that point. Processors keep getting exponentially faster, but most other components in the system are only improving in speed linearly (or, at best, at a much slower exponential rate). If we keep this up, but the time we're dealing with 10^19 bits worth of data, the speed of I/O systems will be so out of whack with the speed of the processors that we won't get anything done.

32 bits alive and well.

[Lumpy]

Hell 16 bit processors are alive and well!

most industrial PC104 form factor PC's are running 386 processors. many embedded systems, the stuff that does real work and critical applications are 16 bit or lower processors.

Maybe AMD is dropping the lower bit processors, but Intel sure won't as long as there is a demand for them.

Re:32 bits alive and well.

[njdj]

Hell 16 bit processors are alive and well!

most industrial PC104 form factor PC's are running 386 processors.

The 386 is a 32-bit processor.
(There was a later variant called the 386SX that used a 16-bit bus, but it wasn't popular, and anyway the CPU was still 32-bit).

Re:32 bits alive and well.

[RevMike]

Maybe AMD is dropping the lower bit processors, but Intel sure won't as long as there is a demand for them.

More than likely they'll leave them to the niche market. Via, for instance, concentrates on the lower power-small footprint-minimal cooling market.

Re:32 bits alive and well.

[swillden]

There was a later variant called the 386SX that used a 16-bit bus, but it wasn't popular

In fact, it was commonly called the "three eighty-six sux."

I always wondered if someone at Intel didn't do that intentionally.

Re:32 bits alive and well.

[red floyd]

Intel blew it with the 386SX. They should have made it 286 pin-compatible.

The chip that really sucked was the 486SL. We developed a board using it, and couldn't find anyone at Intel who would admit to having worked on it.

Re:32 bits alive and well.

[apharov]

Actually current PC/104 and PC/104+ systems are commonly based on the Geode processor that was originally developed by Cyrix for low-end laptops. Geode design was then bought by National Semiconductors and most recently by AMD.

Most recent versions of the Geode chip are roughly equivalent to low-end Pentiums, so it's really a far cry from 386s. New developments in the field are PC/104 CPU modules using the most recent VIA processors, which are quite a bit more powerful than the Geodes are.

Ah, teenagers. always wanting the trendy machine

[Pac]

You expend too much, youngster. My 128K 8 bit Apple II still works fine - I have Visicalc for my spreadsheet needs and the CPM card allows me to use the wonderful WordStar, the king of the word processors. Who needs anything else?

How many?

[Agent R]

There are how many machines out there that run on a 32-bit architecture? All I can say is that that the hardware/software industries should offer interim solutions to allow everyone the time to upgrade.

It was annoying enough watching critical software that worked in Windoze 98, end up being in-operable in Windows XP with no solution around this.

Re:How many?

[Anonymous Coward]

"There are how many machines out there that run on a 32-bit architecture?"
Lots... and?
"All I can say is that that the hardware/software industries should offer interim solutions to allow everyone the time to upgrade"
There is, it's called 32-bit backwards compatabilty. The Athlon 64 runs 32-bit code natively. Information on it is available all over AMD's site, /., or any hardware site on the net. If your programs became inoperable between 95 and XP it was because of the windows API not because of hardware. Before you complain, rtfm.

Why do we need 64 bit?

[Leroy_Brown242]

I have yet to find anyone willing to give me a straight answer my lesser mind can understand.

Why do I need 64 bit computing?

Re:Why do we need 64 bit?

[randyest]

More memory (>4GB) and more computations done per clock cycle (i.e. more perfomance per GHz).

If this doesn't excite you, then you don't need 64-but computing. Many people don't, you may be one of them -- that's OK, just please sit quietly and wait until the rest of us come up with enough memory-hogging apps to fix that for ya ;)

Re:Why do we need 64 bit?

[RevMike]

Why do I need 64 bit computing?

You probably don't.

However, the cost of a 64 bit machine will be about the same as a 32 bit machine, so why buy an old technology when you can buy a newer one that will last longer.

Re:Why do we need 64 bit?

[Anonymous Coward]

You will appreciate 64-bit computing when you peform functions that are math-intensive like rendering CGI, gaming, databases, CAD/CAE, etc. If you do not do any of these things then you don't need to care about 64-bit CPUs. But eventually, they will be very cheap and the only kind of CPU you can get.

Re:Why do we need 64 bit?

[Anonymous Coward]

You most likely don't! However, there are a many scientific applications that require huge amounts of memory, and this seems to be the advantage of 64 bit machines. Using 64 bits huge amounts of memory can be accessed ( I heard something like every atom in the known universe, but I don't know if this is true maybe someone can comment but I do know that 2^64 = 1.84e19 is a whole hell of a lot ) Therefore, processes would not be limited to 2 gigs as they are now, but none of this solves the memory bandwidth problem. Cause trying to fit terrabytes of RAM into a machine will raise issues with space and delays due to distance from the CPU and a whole host of other problems! So I'm curious to know what they are planning to do to make utilization of such power efficient and affordable. But as far as what the need is I think it is a memory thing man! We developers are always looking for more to store our huge games and volumetric data sets!

Re:Why do we need 64 bit?

[Barlo_Mung_42]

Think back a few years. Why did you need 32bit?
Same reason.

Re:Why do we need 64 bit?

[Anonymous Coward]

Because you'll want to buy a computer game that requires more than 4GB of RAM. Think about how much memory (including any swap space--that's part of the virtual address space, too) many games already use. And since all /. play games all day, this would be your answer.

However, in all likelihood, you probably don't need 64-bit computing. It's just for people like me who actually need to do computations with wider integers (which it speeds up nicely), or need to address large data sets in memory. In a few years, you'll come along for the ride just because all the software will have moved on, but you won't need it.

re: Why do we need 64 bit?

[tuc]

The other day I was writing a simple app to modify some of the bit flags in MPEG movie files. Some movie files exceed 2GB or 4GB in length. The code would have been pretty easy to write if I could have addressed the entire file, but I had to implement a sliding-window thing which was kind of a pain.

Re:Why do we need 64 bit?

[Anonymous Coward]

However, the cost of a 64 bit machine will be about the same as a 32 bit machine, so why buy an old technology when you can buy a newer one that will last longer.

Nope, there's 32 bits more there to sell people. 64-bit chips will be more expesive - at least at first - then you might see a slight price drop. But I bet you anything it'll be like DVD's - suddenly you'll be pining for the day when you could buy a perfectly good 32-bit chip for $150. Now the cheapest on the shelf is a $300 64-bit chip that you really don't need for email and a bit of browsing...

Re:Why do we need 64 bit?

[Detritus]

Virtual address space.

There are interesting things that can be done with very large virtual address spaces, such as mapping large filesystems and databases into memory, and sharing a single virtual address space among large numbers of distributed systems. There are systems that depend on each object having a permanent and unique address.

Re:Why do we need 64 bit?

[Anonymous Coward]

No, we don't need 64 bit but the extra registers in the new CPU is surely nice!

Re:Why do we need 64 bit?

[Anonymous Coward]

The more 64 bit processors you buy, the cheaper they get for those people looking to do real work

Re:Why do we need 64 bit?

[hemanman]

You needed 32-bit so you could get the wonders of virtual memory.

Imagine swapping out your least used programs in memory, what wonderfull "speed" enhancements!

Designing an OS to only run programs in memory is insane and obsolete! It would run far to fast for the average user....

-H

I can see it

[steveha]

AMD has said that next year they will be shipping Opteron chips that only dissipate 30 Watts. An opteron runs 32-bit code faster than an Athlon, and totally owns if you run it in 64-bit mode. If you could buy a chip like that, for the same price as an Athlon, why would you want an Athlon?

(If there was a problem getting a good motherboard for the Opteron, that would be a good reason to still want Athlons, but there isn't a problem. There are plenty of good motherboards for Opteron and Athlon64 already.)

steveha

Hardware won't push 32 bit limitaitons alone

[eidolons]

There's no point in a hardware company proclaiming the end of 32bit chips. It will be the demand of software that will push current processor capacities. Currently, we have no software on the horizon that would push mainstream markets to push out 32bit. Word Processors certainly aren't going to require 64bit anytime soon.

The only market I can see are games. Is that going to convince people to curtail their upgrading budget and shell out $2000 just to play doom3 and half-life2 and all those games at the best possible frame-rate? Maybe, but 2005 seems soon.

Re:Hardware won't push 32 bit limitaitons alone

[Short Circuit]

I recall hearing that Unreal Tournament developers were repeatedly running into issues with available memory when designing maps and other game content.

Re:Hardware won't push 32 bit limitaitons alone

[Anonymous Coward]

Any game that required it's customers to have > 2GB of RAM would sell like shit, even on a 64bit platform.

Re:Hardware won't push 32 bit limitaitons alone

[ThaReetLad]

You're confused. They don't require you to have > 2Gb of physical RAM, they need you to have > 2Gb of virtual address space. So that's 256 Mb RAM + 1.75 Gb free disk space. At worst.

Actually a lot of that 2Gb could be taken up with stuff that hasn't been loaded yet, or with stuff that might never be loaded.

Amazing

[Kelz]

If they are so great at predicting future trends, why didn't they realize what the mainstream computer user wants?

They, IMHO, lost the Intel/AMD war because they tried to convince people that (truthfully or not) their processors, though with less MHz or GHz, were faster than Intel's. This may have been true but the average user just wants it to look good on paper. Likely the only reason they are turning a profit is because of the relative expensiveness of the new Intel processors compared to AMDs.

Re:Amazing

[tukkayoot]

Right. Bigger numbers look good on paper, hence the model numbers.

Re:Amazing

[Loki_1929]

"This may have been true but the average user just wants it to look good on paper."

(Bob) "Hey, Bill. I just got a new computer!"

(Bill) "I was thinking about getting a new one myself. Did you get one of those new Pentium 5 ones?"

(Bob) "Nope, one of the tech guys at work was telling me how those are only 32 bits, so I got one of these Athlon ones. They're 64 bits."

In the battle of bigger vs better, AMD has 64, Intel has 32. AMD introduced the model numbers for precisely the reason you raised - the average user has no idea what actually influences performance. The new AMD CPUs are set to ramp up in clock frequency very quickly, which will cause the model numbers to shoot up fast as well. What you'll end up with is Average Joe Consumer looking at Athlon64 4400+ & Pentium(?) 4Ghz. AMD is likely to pump up the "Their's is 32, our's is 64!" marketing, and regardless of whether the consumer has any clue whatsoever what that means, it makes AMD look better on paper.

Re:Amazing

[Zathrus]

This may have been true but the average user just wants it to look good on paper

Ok, so go ask Average User how fast the CPU in the HP Pavilion 3000+ is. Odds are they'll say 3.0 GHz, which isn't true but is proof of AMD's success in "looking good on paper".

The issue that AMD has long had is poor motherboards. Via had a long, long time with poor chipsets/drivers which lead to crashes (this pretty much ended with the KT133A, but it's popped up every now and then since). They also had issues with MS not including the drivers for the chipsets with the OS (which is a death knell, especially for something like a motherboard -- the boards worked without the drivers, but they were dog slow). They also had some thermal problems, which were wonderfully overhyped by the hypemasters at Tom's Hardware (no, I won't provide a link -- if you don't know it you're better off).

Nowadays those issues are in the past. Nvidia has been producing rock solid motherboards for over a year now. Via has finally worked out its issues as well. Via even has chipset support in XP (and Win2k/ME IIRC). Anyone who spouts heat issues is an idiot -- Intel chips now have higher power consumption and heat dissipation than AMD does at the same effective processor speed.

AMD's had issues breaking into corporate PCs though, and still does. Most PCs sold for corporate use are Intel only. They've also had problems breaking into the notebook arena, and they're making a slow go of it in both areas. AMD has long had the enthusiast market, particularly the value-oriented gamers, but it's by no means a lock, and it's really not a very large market.

Re:Amazing

[VoraciousGorak]

"Relative" expensiveness? Intel's response to AMD's FX-51 was the Pentium 4 Extreme Edition. Anyone care to quote a price? I'm thinking it was somewhere around $940... for just the CPU and retail heatsink/fan.

$940 for a CPU is a bit more than "relatively" espensive. o.O

Re:Amazing

[Rtsbasic]

Aren't you forgetting SiS? I've got 2 systems here both using SiS 730(s) chipsets, never had a problem with them.

Re:Amazing

[Kref1]

The average user wants to know how many pentiums a new computer has. Intel's marketing department is very good at getting the pentium name out there and the average to below average users out there are more concerned with getting a pentium4 than some AMD processor.

Re:Amazing

[Tablizer]

I agree. It is a marketing ploy to create a demand that does not really exist yet. To use a worn-out cliche:

1. Prophet end of 32
2. Profit

Re:Amazing

[Zathrus]

SIS isn't very active in the AMD motherboard front -- it's pretty much all Via and Nvidia. They do have some chipsets, sure, but they're not as fast as the other two are, nor do they have as many features.

Not even sure that SIS offers a AMD64 MB for that matter.

Re:Amazing

[Anonymous Coward]

Perhaps .. But Apple is good at saying shit like that

Re:Amazing

[mystdrkn]

If what I hear regularly is true, AMD will have trouble in the future as well. My wife has repeatedly run into business professors who tell their students that there is only one processor manufactuer, Intel. These same professors cannot/will not take a clue. If AMD is brought up in class, it's existence is flat out denied. Of course if suits are in control and buy into marketing from Intel, what do you expect?

Re:Amazing

[Loki_1929]

"1. Prophet end of 32"

Actually, this whole thing is about what someone at AMD said regarding AMD's discontinuation of 32-bit chips by around the end of 2005. Another AMD person said he thought it would be more like 2007 before AMD completely stopped making 32-bit chips. Neither of them talked about the end of 32-bit anything; merely pointed out where they thought their product line would go based on customer demand. With 64-bit chips being about the same price as 32-bit chips, why would you want to buy the one based on older technology?

There was no prophesy here - just roadmap speculation.

Re:Amazing

[mcrbids]

Ok, so go ask Average User how fast the CPU in the HP Pavilion 3000+ is. Odds are they'll say 3.0 GHz, which isn't true but is proof of AMD's success in "looking good on paper".

So, I have an Athlon XP 2000+. I have no qualms about describing it as a "2 Ghz Athlon" because:

• Intel has made clockspeed "the" indicator of processor performance,
  
• AMD has provided an *accurate* method of comparing real-world performance of their Athlons to Intel's clockspeed, and
  
• The chip says "2000" on it. Wouldn't that mean 2 Ghz?

I know full well and understand that it actually runs at 1.66 whatever Ghz. But truth be told, I really don't care. 2 Ghz describes it well enough, and since my 2 Ghz compares quite well against an Intel P4 2Ghz, it's "good enough".

And it doesn't look good just on paper. AMD's "2 Ghz" generally performs very close to, and usually better than, Intel's "2 Ghz". I don't apologize for my Athlon's "2 Ghz" not being "real", as I might have for a Via C3 or perhaps the Cx 6x86. (anybody remember those?)

Re:Amazing

[Anonymous Coward]

this pretty much ended with the KT133A

As someone with one of these, I can confirm the problems are still here.

Re:Amazing

[suitti]

Ok, so go ask Average User how fast the CPU in the HP Pavilion 3000+ is. Odds are they'll say 3.0 GHz, which isn't true but is proof of AMD's success in "looking good on paper".

On the other hand, when the average user asks me what the AMD model names mean, I tell them that my AMD 1800+ is more than twice as fast as the 1.6 Ghz Intel P4 at work. This P4 has 133 MHz RAM, which is less than half the bandwidth available to my 1800.

My points is that AMD's suggestive model numbers do not appear to be misleading. If anything, they're conservative. Yes, I understand that single numbers are a poor guide to performance. I also understand that the market wants single numbers. AMD is using Intel's numbers (MHz), and relating it to real performance in a conservative way.

For me, this builds trust. I still recall Intel's trust busting FDIV bug handling fiasco. Really, tell me again about how I don't care if my math is inaccurate. Why should I buy from them again?

Back to the comparison. My company bought cheap P4's off the shelf. What I did was buy a good motherboard. My guess is that I spent $20 more on it than was spent on the P4. I was able to do this because I put the box together myself. Had I picked up a "system", it wouldn't be $20, it would be $500 more, to get the same performance, as near as I can tell.

My company has more than 30,000 seats. They could easily assembly PCs themselves and save money. In fact, there are many things they could do to save money. The moral here is that big companies aren't necessarily very bright.

Re:Amazing

[Zathrus]

If anything, they're conservative

By and large, and they're considerably cheaper as well.

That said, IIRC, even the P4 1.6GHz had a memory bandwidth greater than what PC133 memory could supply. The fact that the motherboard doesn't offer it is another issue. The P4 1.6 hails back to when Intel was pushing RAMBUS though, which is a whole different debacle.

My company has more than 30,000 seats. They could easily assembly PCs themselves and save money. In fact, there are many things they could do to save money. The moral here is that big companies aren't necessarily very bright.

No, you just don't understand business yet. Yes, they have 30k computers. What's their primary line of business? Unless it's making computers it makes no sense at all for them to hire a hundred people to assemble the PCs and do support on them -- because once you put together your own PCs you can no longer get hardware or software support from a single source. The amount of time and money that would be spent handling the constant failures (and in a sample set of 30,000 there's always going to be a hundred or so that need service) far exceeds whatever money could be recouped by buying parts on the cheap. You also have to factor in the additional HR load (particularly bad if PC assembly and maintainence is nowhere close to your line of business -- how do the HR people know who is and isn't a good candidate?) and similar issues (like space for the additional employees, warehousing of parts, inventory management, etc).

I've built every PC I've owned except the very first which was well over a decade ago. And even that system got the video, memory, and CPU upgraded before I replaced it. That's fine for me though -- I'm confident in being able to build and maintain the computers for my household. But I'll never do it for a friend or family member again. Doing free tech support for myself or wife is one thing. Doing it for someone else is another.

Re:Amazing

[Anonymous Coward]

There is one flaw in your statement... You indicate that the model numbering used by AMD is accurate, when in fact, it is part of the reason their name in the industry has taken a huge hit over the past 12 months. No one can trust AMD any more. Their 3000+ and even 3200+ were out performed on tests when compared to the 2.4 P4 HT, not to mention the slaughtering at the hands of the 3.0 Ghz version. Many companies (besides Intel) as well as large PC magazines and websites have asked AMD to be more realistic and stop trying to con consumers...

Talking from inside the industry, AMD have taken notice of this, and you will probably see a shift in model numbering in the future. Oh, and I forgot to mention that the 2600+ will outperform the 2800+ on some applications due to its higher clock speed.

Intel understand that Ghz does not mean performance, their Pentium M (centrino) models demonstrate this. But you can't claim a processor's performance based on other companies products. Do car companies say that their engine produces 200hp, but it's a 300+ because their car is ligher than the oposition models that have 300hp??? come on...

Ghz is a way to measure a processor so that it can be comparable to it's stablemates. You know, that the difference between a 2.4P4 and a 2.6P4, is pretty much the difference between a 2.6P4 and a 2.8P4. Benchmarks show this, whereas AMD are all over the place. Lable processors by their speed, then list their performance upgrades (higher instructions per cycle, HT and so on) and use benchmarks to back it up.

Don't try to confuse consumers, cause it'll backlash on you... as has happened to AMD...

"People who think they know computers buy AMD,
People who know computers buy Intel"

32-bit processors will still be around

[dtjohnson]

I think that AMD is just saying that *they* will stop making 32-bit processors by 2005 but not that 32-bit processors will be dead by then. It is reasonable for AMD to end production of the 32-bit processors because AMD does not have a lot of manufacturing capacity and they will want to make their 64-bit processors rather than 32-bit processors with the capacity that they do have. Also, AMD's 64-bit processors are also better 32-bit processors than many of the 32-bit processors that have been sold in the past.

The 32-bit processors will obviously be around for a long time yet but they just will not be made by AMD. Intel will keep making them and probably other companies such as VIA and that chinese 'red storm' company (can't recall the name of it) will make them for many years to come. The old 16-bit 286 processors are still made today, even though Intel stopped making them years ago.

Re:32-bit processors will still be around

[Popadopolis]

I think the point is that AMD will stop making them, if not other companies as well. AMD is essentially saying that they are kicking off the phase-out. These things are always slow to happen and dates set by seemingly anyone in the computer business are almost always wrong anyway.

Re:32-bit processors will still be around

[phalse phace]

You're probably referring to Culturecom Holdings Ltd's V-Dragon chip as mentioned in this Slashdot story.

Re:32-bit processors will still be around

[Jeff DeMaagd]

I thought Intel was still making a lot of those old processors for the embedded market. I suppose that since other companies have clear rights to sell those chips so it might be those competitors that I'm thinking of, but Intel still sells to that market to some extent.

Re:32-bit processors will still be around

[dbIII]

I think that AMD is just saying that *they* will stop making 32-bit processors by 2005

And why not? Their 64 bit chips are quite happy pretending to be a Xeon. The big costs have been in development, it doesn't cost them much more to put a 64 bit chip on silicon than it does with their 32 bit chips.

The old 16-bit 286 processors are still made today

You can get a 286 on sapphire, just the thing to put on your gold plated communications sattelite (both materials make sense in the conditions).

Re:32-bit processors will still be around

[juhaz]

The big costs have been in development, it doesn't cost them much more to put a 64 bit chip on silicon than it does with their 32 bit chips.

Perhaps not generally, however the Opterons and Athlon64 chips are big. Way bigger than AMD's "sweet spot" for silicon is, so they actually do cost more than XP's to churn out.

This might change as they move into 90nm.

Moot point

[Anonymous Coward]

What this really says is that all AMD processors will support AMD64, which by definition supports 32-bit instructions.

So they're going to stop making their older 32-bit-only designs, which they would do anyway if they had new 32-bit-only designs.

They're just turning it into 64-bit hype to try and push their "advantage".

Is there any chance that ....

[Anonymous Coward]

AMD is predicting that it's sales of

Is there any chance that CowboyNeal will learn to spell "its" by 2005?

IPv6

[RevMike]

Finally the more advanced processors will become commonplace, if it is not just a rumor. Even if it is a rumor, it is bound to happen eventually.

64 bit PCs will be here long before IPv6 makes significant inroads in replacing IPv4. (Ducks out of way of ensuing flame war)

Re:IPv6

[sik0fewl]

Looks like your ensuing flame war isn't going as well as expected..

Anyway, maybe all 64bit PCs should be forced to ship with only IPv6 and ban IPv4 from being installed on 64bit computers. While this may slow down the adoption of 64bit computers it will definitely speed up the adoption of IPv6 by having it fully in place by the year 2038 :)

Re:IPv6

[tomhudson]

poster wrote:

While this may slow down the adoption of 64bit computers it will definitely speed up the adoption of IPv6 by having it fully in place by the year 2038

Do you really believe we'll still be using 64-bit cpus in 2038? We already have 256-bit gpus, and we've had 64-bit game consoles for a while. 64-bit cpus are just another stop in the road at this point.

Look at history. in 1971, the Intel 4004 (the grand-daddy of todays' x86 cpus) was 4 bits. So:

4004 (4 bits) -> amd opteron (64 bits) in 32 years
= 4 bits -> 8 bits -> 16 bits -> 32 bits -> 64 bits

We've gone form a half-byte to 8 bytes. So, why won't we see the same over the next 3-1/2 decades?

64 -> 128 -> 256 -> 512 -> 1024 bits
= 8 bytes -> 16 bytes -> 32 bytes -> 64 bytes -> 128 bytes

Of course, the gpus in yur game console will then be doing 8192-bits at a time (1kbyte).

Re:IPv6

[amorsen]

The neat thing about protocols is that it is reasonably cheap to support several of them. IPv4 will stay around for a long long time. You can still find IPX in places. That does not make IPv4 a failure.

Re:IPv6

[Anonymous Coward]

Please correct me if I'm wrong, but when you say "We've gone form a half-byte to 8 bytes" isn't that a bit (cough) misleading? Isn't the whole point of moving from 4 to 8 to 16 to 32 to 64 to be able to hold bigger numbers in a single byte?

Re:IPv6

[IWannaBeAnAC]

Because the advantages dimishish. the progression from 4 bits -> 8,16,32,64 is not exponential, but an exponential of an exponential. The more 'natural' progression would be word sizes increasing by some constant number of bits every generation, which would imply the range of arithmetic (and memory sizes) increasing exponentially. But it is technically easier to have word sizes that are themselves a power of 2. We have now reached the limit where 32 bits is barely enough to address the amount of memory desirable even for a small machine, but with 64 bits increases that tremendously. If it took 30 years to get to 64 bits, it should take another 30 years to get to 128.

Of course, this is completely independent of things like memory bandwith tricks (I/O with multiple words at a time) and vector processing (operating on multiple words at a time). But that is orthogonal to the underlying word size. '256-bit GPU' is just a marketing term, that has nothing to do with the amount of memory it can address or the number of bits of precision of the arithmetic, which is what is really meant by the CPU word size.

Re:IPv6

[CaptKilljoy]

Do you really believe we'll still be using 64-bit cpus in 2038? We already have 256-bit gpus, and we've had 64-bit game consoles for a while. Uh, GPUs are 256 bits in the same way that 8 C64's glued together are 64-bits, which is to say, not in the slightest. It's just marketing BS.

Re:IPv6

[Sunlighter]

'256-bit GPU' is just a marketing term, that has nothing to do with the amount of memory it can address or the number of bits of precision of the arithmetic...

Actually, what they mean by this is that the GPU can grab 256 bits in one bus cycle.

GPUs also use SIMD, which probably means that you can add 16 16-bit integers to 16 more 16-bit integers with signed saturation, or something like that, in the GPU.

You're right that the GPU can't address 2^256 bytes of memory and it doesn't use 256-bit integers or floats. But '256-bit GPU' is more than just a marketing term.

Since so many graphics modes are 32 bits per pixel now (and even more if you include a depth buffer, stencil buffer, etc.), you're probably talking about 8 pixels per bus cycle. This is of great advantage for GPUs.

Re:IPv6

[tomhudson]

No, it's also about manipulating more data with each cycle. Imagine doing an XOR between two 1kbyte-wide registers, as opposed to 32-bit wide registers, as just 1 example.

There actually was a card that was sold by (IIRC IBM) a few years ago that would do 1kbit-wide math, for banking hardware.

Re:IPv6

[tomhudson]

poster wrote:

Uh, GPUs are 256 bits in the same way that 8 C64's glued together are 64-bits, which is to say, not in the slightest. It's just marketing BS.

You might want to research the hypercube from Thinking Machines. It was an early parallel-processor machine w. 65,536 8-bit cpus "glued together". Very cheap solution (at the time) for doing air-flow modeling, etc.

Re:IPv6

[CaptKilljoy]

Uh, Thinking Machines' Connection Machine CM-1 and CM-2 were composed of up to 65536 *1-bit* processors which could be grouped together in various ways for processing. Even then, grouping 32 of these bit slices would have been in no way competitive with an equivalent 32-bit microprocessor

(Some advice: If you're going to be snide, make sure you know what you're talking about first.)

Re:IPv6

[Anonymous Coward]

...And they'll be here right in time for Duke Nukem forever!

Re:IPv6

[IWannaBeAnAC]

No. Bus width is completely different to the word size of the CPU.

Ever since the Pentium, the memory bus width on x86 CPU's has been at least 64 bits. Would you say that that qualifies a Pentium to be a 64 bit machine?

Re:IPv6

[tomhudson]

The CM-5 was based on a SPARC cpu. These were never 1-bit. The one I was referring to, however, was a one-off profiled in Scientific American. The Ones you were referring to had 1-bit cpus and 64-bit fpus. So we're both right :-)

Disregard that last

[Bagels]

-_- stupid sig didn't update right (you'll still get a funny, just not the relevant one)... now it should work.

Re:Disregard that last

[Anonymous Coward]

Your sigs aren't copied to your comment when you post.

Re:Disregard that last

[bhtooefr]

No, they're copied to every comment you ever wrote when you change it.

Re:Ah, teenagers. always wanting the trendy machin

[DrEldarion]

128K? Wow, what a luxury. I'm still running on 64K.

Superprocessors

[Popadopolis]

Does anyone know the point at which, to make computers any faster you would need superconducters?

Re:Superprocessors

[Cyclopedian]

Two conditions exist to require superconducting processors for higher performance:

1) You notice that it takes more than two seconds for that program window to close after clicking the close button.
2) When the temperature of the CPU is at 30 degrees Kelvin.

*THEN* at that point, you get superconducting processors.

Be on the lookout for the new line of Prometia coolers dubbed K class. Sure to bring that pesky electron resistance to nothing.

-Cyc

Re:Superprocessors

[Popadopolis]

Actually, there is a perovskite that is a superconductor at 125 Kelvin. Perovskites are crystal structures that are considered "high-temperature superconductors." My conversion may be a little off, but I believe that 125 Kelvin is about -1 Ferignheight (spelling may be off too).

Re:Superprocessors

[digital bath]

125 Kelvin = -234.67 degrees Fahrenheit... gotta love google :)

Re:Superprocessors

[Popadopolis]

Ah, forgot about the x10^-2 part. Note to self: remember to look at the entire result before jumping to conclusions...

Re:Keysh-OWNED

[Anonymous Coward]

What good are niggaz anyway? We'll give him a good send off with some rope and a tree limb. We gon have a real time lynchin'!!!

that's fine and dandy but...

[ambienceman]

how does this relate to the advancement of internet pr0n? better and more reactive holograms?

Re:that's fine and dandy but...

[ambienceman]

hmmm...maybe i'll stick with 32 for a bit longer.

2038 is not a problem

[Anonymous Coward]

Just redefine time_t to be 64 bits and off you go. gcc has been able to handle 64 bit ints for many years now, even on 32 bit processors.

Sure, it's not terribly efficient at it, but it's not like you need to do enormous numbercrunching with time_t's. If you do, well then 64 bit is certainly for you.

Consider all the years that 8-bit computers were doing 16-bit calculations...

Re:2038 is not a problem

[Linker3000]

I've just run the '2038 test' perl code on my Redhat 9 box and it rolls back to 1901 - I'm in panic mode now! I understand I have to find a retired COBOL programmer and offer them lots of money to fix it - is that right?

Re:Ah, teenagers. always wanting the trendy machin

[RevMike]

I have Visicalc for my spreadsheet needs and the CPM card allows me to use the wonderful WordStar, the king of the word processors.

What is sad is that I'm only 30 and I can tell you that Ctrl-B reformated a paragraph in WordStar. I usually used wsn, however, to edit my C programs, which I then compiled with my lattice C compiler. Those were the days when men were real men and clocks ran at 4.7MHz.

Power consumption

[yerricde]

if your 64 bit processor runs the same as your 32 bit one and has 64 bit instructions why would you still use a 32 bit ?

I'm assuming that a 32-bit processor draws less current than a 64-bit processor, which becomes important for handheld devices.

Re:Power consumption

[tarquin_fim_bim]

Would you be asuming that, for a similar amount work done, or just having the thing switched on.

Re:Power consumption

[yerricde]

Would you be as[s]uming that [64-bit processors generally run on more current], for a similar amount work done, or just having the thing switched on[?]

Both. A 64-bit ALU takes at least twice the gates of a 32-bit ALU (actually slightly more to handle long-distance carries). Also remember that 64-bit code and data structures tend to take more space than their 32-bit equivalents, overflowing the cache and requiring a faster memory bus to compensate, which in turn drains more power. Any real advantage that more transistors give (Moore's Law) will generally be eaten up by software rushed to market without optimization (Gates's Law). Finally, I'm assuming that because of their bigger die size, 64-bit CPUs drain more power than 32-bit CPUs even in HLT state.

Re:Power consumption

[Hoser McMoose]

FWIW at 130nm fab process (ie current processors), AMD figures that their 64-bit chips have roughly a 5% larger die-size than an otherwise identical 32-bit chip. When you shrink to a 90nm fab process (next year's chips) that number shrinks, and when you start switching to a 65nm fab process (at about the end of 2005) the difference is totally negligible.

Power consumption for 32 vs. 64-bit chips might be of a minor issue for embedded chips, but for AMD's desktop chips it's already a non-issue. AMD's Opteron and IBM PowerPC 970 chips (both 64-bit) already consume less power than similarly-performing Intel's Pentium4 and Xeon chips (32-bit), simply due to different designs.

Right, and here comes the future how I see it...

[Anonymous Coward]

1st) 32Bit
2nd) 64Bit
3rd) 32Bit again but without a noisy fan

I believe that some day people will look for low noise levels instead of CPU power they don't need to watch some DIVX. Games will be played on Nintendo consoles.

Not to Worry....

[turgid]

I'm sure our good friends at intel will have a $200 itanic for the home and desktop market by then that will only dissipate 250W of heat or something. So don't all go rushing out to by these newfangled AMD thingies at once...

Itantium Heartburn

[Nom du Keyboard]

The company [AMD] expects to ship between 50 and 100 million 64-bit processors over the next three years

I can't think of anything more likely to give Intel heartburn than this.

MCE

[sgasch]

I'm sure you already found this out by reading the intel manual but MCE = machine check exception. It's a hardware generated exception like a NMI. I think MCE happens when a parity error is detected on the memory bus so my first suspician would be your motherboard or memory. But I also seem to remember other ways for that exception to be raised (unfortunately my good books are on loan to a coworker).

Not that I'd recommend it, but you can handle one of these exceptions and effectively ignore it. Of course this raises the possibility of data corruption. A little linux hacking should do the trick. BTW there's a reg key to disable the bugcheck in NT and as far as I know Win9x ignores the exception altogether (excellent plan there).

Good luck tracking this down.

Re:MCE

[Naffer]

So it's a simple memory error. I'd say that the best route to take would be to just sit tight and see if it happens again. Big sticks of memory that lack ECC or a suddenly undervolted northbridge can cause em. Check out memtest86 and test your RAM for stability.
I'm not a Unix user, but thats what all us uneducated Windows PC users do when we overclock or have memory errors.

Re:MCE

[Anonymous Coward]

Could you tell me what the key in NT is to disable a bugcheck, and does it work in newer versions of NT (2k, XP)

Post Is Misleading

[jetkust]

AMD is only speaking of the PC industry #1. 32 bit processors will start fading out in the same way 2x cdrom drives did, or 1 gig hard drives. They will keep making 32 bit processors, but when 64 bit processors reach a certain price customers will have no reason to keep buying 32 bit. Plus none of it mentions anything about throwing away existing 32-bit machines. This "end of 32-bit Processors" crap is just the usual slashdot propaganda.

WordStar?

[Tumbleweed]

You punk - you should stick with native Apple ][ word processors, like Magic Windows ][.

Punk.

Oh yeah, get rid of all that extra RAM - you don't need anything more than 48K! :)

Re:WordStar?

[bhtooefr]

Desolder 16K from the mobo AND yank out the 80 col card? Yeah right.

Re:WordStar?

[Tumbleweed]

Oh, you've got one of them newfangled //e's, then, not a ][.

And what's an 80 col card? :)

Re:WordStar?

[bhtooefr]

80 column card. Some were available for the ][/][+ systems. Basically, the first ones added 1K of VRAM. The VRAM was allocated so that the even-numbered columns (if the first column is 0) were in the normal 1K of text VRAM, and the odd-numbered columns were in the additional 1K of text VRAM (note that later //e systems used this 1K VRAM for a double-hires video mode in addition to the 80-col mode). Most 80-col cards had 64K RAM instead of 1K, so that the system could have 128K RAM.

Re:WordStar?

[Tumbleweed]

Please to note the smiley face after that. It was a joke. I had a //e, myself, and eventually upgraded it with a 512K RAM card - it was crazy fast after I loaded the floppy image (all, what, 142K of it?) into RAM. Fun times...

Re:WordStar?

[bhtooefr]

You were lucky. I had a cheapo clone //e with a 6502A (2MHz! Woohoo!), a legit //c with all of the RAM already there (w/o expansion - it was the first model), and a Laser 128EX without an expansion card installed (it had a //e slot hanging off the side)

Why?

[t_allardyce]

UNIX calculates time in seconds since its birthday, GMT 00:00:00, Thursday, January 1, 1970 C.E. The industry-wide practice is to use a 32-bit variable for this number (32-bit signed time_t)

I'm probably missing something but why did they use a signed number? they could have got an extra 70 years out of it!

Re:Why?

[Anonymous Coward]

What? And break compatibility with versions of Unix from Before the Christian Era? You've got to be kidding. It may seem dated to some, but Julius Caesar was a pretty mean hacker for his day. In fact, current Unices still depend on some of that code. Why do you think legions of programmers still use editor number 6?

Re:Why?

[jonbryce]

To deal with dates before 1970?

The date of birth field in most places will still require this.

Re:Why?

[t_allardyce]

oooh, but that would only go back to 1901? surely there were people older than 70 back in the 70's?

Re:Why?

[Detritus]

See The Development of the C Language by Dennis M. Ritchie. UNIX predates C by a number of years.

Early versions of C did not have all of the integral types that we are now used to having. I remember using an old C compiler that only had "char", "int", "unsigned" and "long". The only 32-bit type was long, which was signed.

Re:Why?

[Lochin Rabbar]

No system time only affects an application in as far as it might question the system for the current time. D.O.B. fields in databases are just bits in files. The difference between the two can be calculated irrespective of what the arbitrary start date of system time was.

How can you say that?

[Orien]

IMHO though 32-bit systems will never die.

How can you say that with a strait face? Doesn't that sound just a little bit like "no one will ever need more than 64k of ram"? ALL technology has built-in obsolescence. It probably won't be in three years like AMD wants us to believe, but 32 bit systems will eventually be obsolete, just like the horse drawn buggy you will only be able to find them in museums and in the basements of fanatic collectors. I feel like writing a cron job that will remind me to look you up in 2023 and remind you of that statement you made.

Re:How can you say that?

[CaptKilljoy]

How can you say that with a strait(sic) face?

Because he actually knows what he's talking about. Here's a hint: the market for 8-bit processors is absolutely enormous even today. There's even a small but significant market for 4-bit processors. Do a little research.

(I can't believe the parent post is modded +3, Interesting.)

Re:How can you say that?

[Hoser McMoose]

Sure there is a market for 8-bit processors now, but will there be in 10 years? 50 years? There was a market for vacuum tube-based computers after transistors made them totally obsolete, but you sure as hell don't find anyone making vacuum tube based compared anymore (actually, knowing the Slashdot community, there probably is some wierdo out there doing just that, but I digress :> ).

The market for 8-bit chips is already starting to disapear. Why? Not because they are no longer useful, but because the difference in price between an 8-bit and a 32-bit chip these days is negligible, and if you can standardize ALL of your development on a single chip, so much the better. Why bother have one $2 8-bit chip to do one task, a $2 16-bit chip to do another task and a $2 32-bit chip to do a third task when you can do all three using $2 32-bit chips.

This is the same reason why AMD plans to stop making 32-bit desktop processors in the not-too-distant future. Already the difference in price between making the 64-bit Athlon64/Opteron and a theoretical 32-bit version of the same chip is very small (5% difference in die size according to many previous AMD documents, so probably about a 1-3% difference in total cost of production). That number will fall over the next two years to the point where it's totally pointless for them to bother making 32-bit chips anymore.

32-bit chips aren't going to disapear after 2005, but sooner or later they will. Hell, eventually ALL processors as we know them today are likely to be replaced, probably by something that none of us can even guess about today.

Re:How can you say that?

[Talez]

Yeah... well... 16-bit processors will never die.

God forbid the day Motorola stops making 68K chips.

Re:How can you say that?

[CaptKilljoy]

but because the difference in price between an 8-bit and a 32-bit chip these days is negligible

Lets say there are 250 million pocket calculators (or CD players or microwaves or whatever) out there in the world. Every penny shaved off the cost of the processor and support electronics would save the manufacturer 2.5 million dollars. Do you call that negligible?

8-bit microcontrollers may go away but it won't be in the lifetime of anyone here.

Yes, but,

[gillbates]

The real question is how long will it be before the BIOS is 64 bit protected mode?

Probably never.

I still write in 16 bit assembly because the BIOS still runs in 16 bit mode. It would be nice if AMD broke backward compatibility for once and started off with 64 bit firmware so I could at least write 64 bit assembly. The mixed-mode stuff (16 and 32 bit) that I would have to do for OS programming is getting ridiculous:

• I could write in 32 bit mode, if I wanted to write drivers for every single conceivable piece of hardware out there. While this would be ideal, it is far from possible. And since the processor starts in 16 bit mode, I'd still have to write at least a little real mode assembly.
• I could still use 32 bit mode if I wanted to use a call gate to call the 16 bit code of the BIOS, or:
• I can write 16 bit code, call the BIOS directly, and only have to worry about the 64k and 1M memory limitations.

I don't like any of these solutions, but it's a lot easier to fit kernel modules in 64k than it is to write call gates for BIOS services. The reason why I like using the BIOS is because it is standard across differing computers - I don't have to write a different driver for every single video card and hard disk controller that I might come across. Plus, if it uses the BIOS, I can be reasonably certain that it will run on an arbitrary PC; I don't have to do any hardware probing or detection.

Well, it's a pipe dream, I guess.

Those of us who like to program their own hardware took a serious hit when the 32 bit OS became the standard. We either ended up jumping through hoops to use the 16 bit BIOS from protected mode, or we just decided not to use more than 1 megabyte of the machine's memory. If they had installed 32 bit BIOS's when the 32 bit processors came out, we would never have had these problems.

But no, we still have a 16 bit BIOS because the manufacturers are afraid that some fool might want to run DOS on their 3GHz Pentium 4 with 1 GB of RAM....

Re:Yes, but,

[dmayle]

I don't know why I'm doing your work for you, but you should try educating yourself.
Check out BIOS32 services. It's a 32-bit entrance for BIOS services, and it's an industry standard...
PDF link

Re:Yes, but,

[Anonymous Coward]

I didn't know anyone even used BIOS calls for real work any more. It seems to me you should invest in a real time OS. There are a number of ones that would provide you all the 32 bit hardware drives you need. Some aren't much bigger than a BIOS image.

Re:Yes, but,

[Svartalf]

"I still write in 16 bit assembly because the BIOS still runs in 16 bit mode. It would be nice if AMD broke backward compatibility for once and started off with 64 bit firmware so I could at least write 64 bit assembly. The mixed-mode stuff (16 and 32 bit) that I would have to do for OS programming is getting ridiculous..."

Uh... You might want to check out the architechture spec on the AMD64 IA. It's not got ANY 16-bit or 8-bit backwards x86 compatibility.

Re:Yes, but,

[Anonymous Coward]

> manufacturers are afraid that some fool might want to run DOS on their 3GHz Pentium 4

Using DOS is intergal to the setup of many machines (with Norton Ghost, for example). It's also needed to boot NetWare and Windows 98 (which still gets deployed in certain corps).

Besides, Intel is starting to push protected mode firmware (EFI) that's still Real Mode BIOS compatible.

Re:Yes, but,

[Anonymous Coward]

Actually it does have all the 16-bit stuff, but it's disabled when in 64-bit mode. The systems still boot in Real Mode and are fully compatible with all 8086 and 80286 operating systems.

Re:Yes, but,

[the_greywolf]

Uh... You might want to check out the architechture spec on the AMD64 IA. It's not got ANY 16-bit or 8-bit backwards x86 compatibility.

yeah, i've got the AMD programming manuals buried here somewhere. in the 64-bit "long mode", 16-bit operations are illegal. however, it initializes in 16-bit mode, and all 16-bit instructions are valid instructions at that point.

16-bit instructions only become illegal in 64-bit "long mode." please make sure you remember to clarify that point in the future.

Re:Yes, but,

[jack torrence]

I don't plan on giving up DOS. I think Win32 compliant means DOS32 compliant also (or can at least be forced to be so). Why do I have to do what everyone else does? So as long as the binaries are still there I am in business. No?

Re:Yes, but,

[Anonymous Coward]

And the BIOS32 service to put text on the screen and change video modes is...? Oh wait -- there isn't any. You must use the crappy old 16-bit INT 10h.

And the BIOS32 service to access the hard drives is...? Oh wait -- there isn't any. You must use the crappy old 16-bit INT 13h.

And the BIOS32 service to determine memory size and layout is...? Oh wait -- there isn't any. You must use the crappy old 16-bit INT 15h.

And the BIOS32 service to read from the keyboard is...? Oh wait -- there isn't any. You must use the crappy old 16-bit INT 16h.

PnP and PCI are the only BIOS32 services I've seen. Educate yourself, dumbass.

Re:Yes, but,

[RogerWilco]

But no, we still have a 16 bit BIOS because the manufacturers are afraid that some fool might want to run DOS on their 3GHz Pentium 4 with 1 GB of RAM....

My dad just bought a brand new P4 2.4 HT /w 512 Mb and 80 Gb HD. He bought an MSI instead of an Asus motherboard just so he could still use both his 3.5" and 5.25" floppies. Asus only supported 1 floppy drive.
Then when he installed PC-DOS 2000 it freaked out on him. Didn't know what to do with the new 80 Gb HD. Oh and he had a problem assigning partitions, as the max size was 2 Gb. and the alphabet only has 26 letters...
I suggested he should install DOS 3.3 and see if that would fare any better.

;-)

Well, of course

[ottffssent]

The announcement is pure marketing, hyping their 64-bit chips.

The reality is that old processors get replaced by new ones, and once high-end chips end up as low-end chips - nobody buys K6s anymore. AMD's just saying that they'll continue phasing out old processors, leaving only 64-bit-capable ones.

AMD's current lineup spans K7-era chips at the low to mid range and K8s at the top. In two years, they'll have K8s at the low to mid range and K9s (or whatever) at the top.

AMD has a competitive 32-bit chip, which will move down the product line just as their previous 32-bit chips have. That it does 64-bit too is an added bonus that will get carried along with it until all of AMD's chips are 64-bit capable. It's not some grand scheme, just the result of how the industry works.

In my day...

[Shazow]

You're soft. Why, in my day, we sent rockets to the moon with 8-bit processors, and we considered ourselves lucky to have them.

Let me guess. Against gravity both ways? In the cold void of space? 384,000km each way?

More bits, same problems, today.

Embedded Market

[tombou]

Even when the desktop market moves on completely to 64 there will still be the embedded market that can use the 32 bit procs for processor intensive applications.

The new via eden is attractive ...

http://biz.yahoo.com/bw/031014/145512_1.html
ht tp://www.via.com.tw/en/Products/eden_n.jsp

Agreed.

[yerricde]

I agree. I'll upgrade from 16-bit to 32-bit when my favorite Sega Genesis and Super NES games are ported to Game Boy Advance.

it could be dead in AMDs eyes

[MagicBox]

But I doubt it very much. I think this is a PR stunt from AMD, who is trying to get more people interested in their 64bit architecture. That's fine, but the prediction seems to be more realistic for AMD than the industry in general. I predict in 6 months AMD will find another excuse to take their claim back

Perhaps...

[devphaeton]

...i'm not as geeky as i used to be, but i've currently got an XP 1800+ system with EIDE drives, and a TNT1(!) video card.

It replaces my 300mhz k6-II (which i still use almost daily).

Everyone and their dog can run out and buy the latest n greatest hardware. I personally feel that i'm set for about the next 4 years. By then, stuff like SATA and 64-bit processors will actually mean something.

Right now, there are few OSes and applications that can really use this hardware, including Apple. It's kinda pointless, IMHO

But hey, i can't wait till we've all got 64-bit hyperthreaded tools like rm ls cp ln grep mv......

Re:Perhaps...

[VoraciousGorak]

...i'm not as geeky as i used to be, but i've currently got an XP 1800+ system with EIDE drives, and a TNT1(!) video card.

I personally feel that i'm set for about the next 4 years.

Uh-ohhhz! Someone's not a gamer! ;)

64bit / 32bit same old x86 crap

[Anonymous Coward]

It great that after 10+ years of analysits predictions and hype were finally 2-3 years away from main stream 64 bit machines (ma & pa's $700 computer with 64bit processor). However, that processor will still be using the ancient and totaly absolete x86 architecture. A cpu where a large chunk of the core is spent making it backwards compatible instead of spending it on greater cache or making the processor more power effecient etc...
BAH...

John Titor?

[Aero Leviathan]

Oh.. so the 2038 problem is nearly fixed? So this John Titor guy is a fake, right?

Whew!

384,000km?

[billstewart]

Kilometers? In my day, when Americans went to the moon, we only had *miles*, and only had 238,000 of them....

Re:384,000km?

[YU Nicks NE Way]

Yeah, but in *my* day, we only had light-seconds, and about 1.5 of them.

So there!

Re:384,000km?

[red floyd]

You had light seconds? We had furlongs. 1,904,000 of them!

Re:384,000km?

[Talez]

Yeah? Yeah? You had furlongs?

Back in ancient egypt when the aliens were helping us to build a space elevator we only had cubits and we had to go 837,760,000 of them to reach the moon!

Re:384,000km?

[Anonymous Coward]

back in my days, we only had kilometers.

Re:384,000km?

[stfvon007]

You had units of measurement? Back in my day we just called it far away. Anyway the space program's computers have been slow to be upgraded. (laptops 5 years ago were better than the space shuttles computer and were probably better even before that) Though with the way modern computers are im not sure they would want them upgraded. The shuttles computers need to be reliable before they are fast. You wouldnt want the shuttle computer to have a blue screen of death come up.

Re:384,000km?

[Anonymous Coward]

in ancient egypt when the aliens were helping us

Hardly. Jaffa, kree! ;-)

Re:384,000km?

[stmpynode]

Bluescreen of death? Isn't that the sky?

Red storm rising (Cray & AMD)

[Space cowboy]

According to Computer Shopper, AMD and Cray are collabarating on a new chip interconnect method, which they claim runs 20x the speed of current solutions, called 'Red Storm' ...

Just a 'news in brief' item, so no real details...

Simon

Re:Red storm rising (Cray & AMD)

[parkanoid]

Here's a link with some details. Apparently it's a direct hypertransport tap between the CPUs. Interesting, I was thinking of something like this when I first heard about AMD64 using hypertransport.

Re:Red storm rising (Cray & AMD)

[ChaoticLimbs]

"Red Storm". "Black Widow". All so masculine and male-centric. Couldn't you have "Project Doily" or "Operation Gingham Curtain"?

Re:Red storm rising (Cray & AMD)

[FauxReal]

"Red Storm". "Black Widow". All so masculine and male-centric. Couldn't you have "Project Doily" or "Operation Gingham Curtain"?

Doesn't the term "Black Widow" refer to either the poisonous female spider and/or a woman who marries and then kills men for financial gain?

Re:Red storm rising (Cray & AMD)

[ChaoticLimbs]

Yes, it does. However, the use of this generalization of one spider's occasional behavior as a term to degrade women who choose to kill their spouses is insensitive. As everyone knows, men are insensitive more than women are. Therefore, Black Widow is a leftover of male hedgemony, and should be expunged.

Re:Bush is beloved in UK- Guardian

[Anonymous Coward]

I for one, welcome our new word inventationalizing overlord.

In other news...

[djupedal]

Sept. 21, 2010 - Lake Oswego, OR.

Sears Network Division Headquarters today announced a ban on all older 32 bit based computers on their internal network due to contributed slowdowns data throughput issues and overall drains on employee production.

WooHoo...

[velo_mike]

Tue Jan 19 03:14:01 2038 Tue Jan 19 03:14:02 2038 Tue Jan 19 03:14:03 2038 Tue Jan 19 03:14:04 2038 Tue Jan 19 03:14:05 2038 Tue Jan 19 03:14:06 2038 Tue Jan 19 03:14:07 2038 Tue Jan 19 03:14:07 2038 Tue Jan 19 03:14:07 2038 Tue Jan 19 03:14:07 2038 My Ultra 1 is ready for 2038 - provided it doesn't exceed my 16M ram and 2gb HD... D'Oh...

Re:Ah, teenagers. always wanting the trendy machin

[Anonymous Coward]

and... It will continue to run until 2040 ;)

----------
All Mac OS date and time utilities have correctly handled all issues related to the year 2000 since the introduction of the Macintosh. The original date and time utilities, introduced with the original Macintosh 128K in 1984, used a 32-bit value to store seconds, starting at 12: 00: 00 a.m., January 1, 1904. Since Apple used a 32-bit value to store seconds, this means that the last date represented in this 32-bit value is 6: 28: 15 a.m. on February 6, 2040.

obsolete 32-bit systems in less than 3 years

[pmz]

My six-year-old Sun Ultra workstation won't be obselete! What a relief!

Re:Ah, teenagers. always wanting the trendy machin

[rizzo]

Hey I used fredwriter until my senior year of high school (1994). I was the only person in school using the only Apple II in the school. Rest of the sell-outs used the Mac labs. Handing papers printed off of dot matrix when everyone else had laser/inkjet was awesome as well.

Re:You fail it sucka

[Anonymous Coward]

As a punishment, you should wrap duct-tape around your balls and rip it off in one go.

You call that punishment? Somebody doesn't know how to have a good time.

DO NOT DISPAIRE!!

[bsDaemon]

IPv6 will be out in time enough to boost the number of players in the Duke Nuk'em Forever online free for alls!!

Key point: Devs like 64-bit

[ErikTheRed]

I think that the key point of this article is that developers can now develop and test 32- and 64-bit apps on the same machine. With many high level languages (and even, to a certain extent, C/C++), it's fairly trivial to develop a version that compiles under both archs, especially if you're starting a new project (just have to watch your int & pointer sizes, etc). I think that a key attraction in CPU-intensive apps (games, multimedia creation/editing, scientific, etc.) will be the extra 8 general purpose registers available in 64-bit mode. They can produce order-of-magnitude performance increases for parameter passing, many inner-loops, etc.

Read...

[Short Circuit]

...this for info about AMD64 and 64-bit computing in general.

Re:Read...

[Anonymous Coward]

thanks for the link!

Re:Read...

[Anonymous Coward]

Mod parent off-topic.

Couple of common answers

[Anonymous Coward]

1. Why do I need 64 bit?
Even if you don't, you aren't the market. If servers and gamers will only upgrade to 64 bit next year, and 64 bit has enough performance boost to make it more effective per dollar, heavy users will adopt it, making it cheaper and more usable for the mass market.
2. Why in '05?
It's another upgrade cycle from now. By then, there will be enough support products and user base that most important apps will be available there. And, of course, AMD's fabs will be obsolete. 32 bit will represent a low margin business at that point.

You don't.

[ProtonMotiveForce]

It's just something to rally the AMD fanboys.

There is no reason (and there isn't some near term "future" you should buy for) to get a 64-bit processor over a 32-bit processor for almost any home computer user. I don't care if you're the geekiest power user out there running all kinds of fancy pants games, you don't need it - it won't help you.

64-bit CPU's won't be relevent until the average home PC has at least 2 gigs of memory. That's about 2-3 years off _at least_. Your fancy pants 64-bit processor you buy now "for the future" will be obsolete by then.

Re:You don't.

[Short Circuit]

You're missing the whole point. AMD64 solves the "chicken and egg" problem by bundling the egg in with standard offerings.

Re:You don't.

[Leroy_Brown242]

Your fancy pants 64-bit processor you buy now "for the future" will be obsolete by then.

It all comes down to buying what will do what I want for the best price. If an 64 bit computer will run my 32 bit apps, run them faster, and not run my credit card any more, then I'm going 64 bit. I don't care if my 64 bit computer is obsolete in 5 years, if it's cheaper and better now, then I know what i am buying.

Re:You don't.

[Kenja]

"AMD64 solves the "chicken and egg" problem by bundling the egg in with standard offerings."

And they charge you for an omlet with has browns, toas, sausage, bacon, ham, biscuts and gravy. For the price I could buy three chickens and two eggs.

Re:You don't.

[ProtonMotiveForce]

Now that makes sense. If it's faster and cheaper than a 32-bit processor then fine - it's a good buy.

But the fact that it's 64-bit shouldn't enter into a (rational) buying decision.

Re:You don't.

[Hoser McMoose]

There is no reason to get a 32-bit processor over a 64-bit one either though. As long as the price is the same, why worry?

Seriously people talk about not wanting 64-bit chips like there's some downside to it. Switching the hardware to 64-bits is essentially free. The cost in silicon is already minimal for a desktop processor with today's (130nm) production capabilities, and with the 90nm production it becomes negligible.

So, the easiest answer to the question of "Why 64-bit?" would be: Why not?

Embedded systems

[jhines]

Embedded systems, which are found all over, can still make use of a 32 bit cpu. I don't know how much custom work AMD does, but it would be a shame to throw away the designs.

Re:Embedded systems

[ziggy_zero]

I think it is implied that 32-bit CPU's will be dead for the desktop....not embedded systems. We can all agree that 32-bit will definitely still be quite useful, just not necessarily in the desktop, definitely not in the server, and maybe not even the laptop markets.

Re:Embedded systems

[juhaz]

Something like Athlon is probably overkill for just like every embedded market there is. It's damn fast but it's also (just like all the other fast desktop chips) power-hungry and runs damn hot.

AMD does make small chips for embedded systems... I don't see them trowing those out, this is just for the biggies, and even for them, why would they throw away the designs just because they don't make the things any more?

Well, of course it won't be in your phone

[sbma44]

If I'm completely wrong here, someone please correct me, but the only reason for going to 64 bit is to escape the limitations on addressable memory imposed by 32 bits. Until embedded devices come bundled with more than 4GB of RAM, there will be zero reason to use a 64 bit chip.

Or am I wrong? Someone help me out here. I see an awful lot of coverage of 64 bit processors on slashdot and other consumer/overclocking sites given the fact that I know *no one* who's got 4 gigs of ram in their box.

These things are applicable for huge servers, industrial 3d rendering and video editing. But only the most advanced applications of each need more than 4GB of memory. I just don't understand this at all -- when you get to needing this amount of memory, odds are you'll be out of the prosumer space anyway (eg, making a feature film), and can drop money on one of the existing 64 bit processor options.

Why is a 64 bit entry by a consumer chipmaker such a big deal? Sure, making cutting edge tech affordable is nice, but the 4GB barrier is not what's keeping the average Joe from making Jurassic Park.

Basically, I'd argue that the number of flops available to the average user is a bigger barrier to him or her attempting these huge applications than the amount of addressable memory is. Until AMD starts selling a rendering farm for $400, 64 bits will make very little difference for the people getting most excited about this stuff.

Although I will say that for the average university lab working with huge datasets, this could be a real boon.

Re:Well, of course it won't be in your phone

[Planesdragon]

If I'm completely wrong here, someone please correct me, but the only reason for going to 64 bit is to escape the limitations on addressable memory imposed by 32 bits. Until embedded devices come bundled with more than 4GB of RAM, there will be zero reason to use a 64 bit chip.

You're wrong.

A 64-bit chip, AFAIK (IANAEE), can process data in 64-bit "chunks." While this does pre-emptivley remove the 4 GB barrier (in 5 years or so, if current pace remains constant, 4 GB+ will be fairly common), it also has the nice side-effect of speeding up the processing of any properly-compiled program that handles data in chunks bigger than 4 bytes.

All other things being equal (which they never will be), a 64-bit processor will be faster than a 32-bit processor, simply by virtue of getting more out of each clock cycle, even if they address the same memory space.

(Another, off-the-top-of-my-head use for 64-bits: addressing fast hard drives with the same schema that is used to address memory.)

Re:Well, of course it won't be in your phone

[Hoser McMoose]

The larger addressable memory space is the primary advantage to using 64-bit chips. You also get 64-bit integer registers (generally speaking, though the addressable memory size and integer register size aren't always linked on all architectures, though if you have only one of the above I wouldn't really call it a 64-bit chip). For most applications, 64-bit registers don't get you anything, but there are a few situations where they do help.

As for 64-bit on the desktop, I think that now is the time to start the move. Installed memory on desktops tends to double every 18-24 months. Right now 1GB is the norm for a high-end desktop. That means that by mid to late 2005, 2GB will be the norm, and that's pretty much the max for 32-bit chips without resorting to all kinds of ugliness. Sure, you can still address 4GB of memory with a 32-bit chip, but that's your virtual address space. Having more than 2GB of memory on a 32-bit chip means that you need less virtual memory than physical memory, which is fine for a system like Linux, but not so good for BSD or Windows (due to differences in how they use virtual memory). You also start running into some issues of memory fragmentation. In short, a 64-bit processor becomes a real advantage any time you have 2GB of more memory, which as mentioned above, will likely become common place for new systems sometime in mid to late 2005.

So why move to 64-bit now if you don't need it for two years? Simple, software is much slower (and more expensive) to move to new architectures than hardware. AMD's 64-bit chips have been out for 6 months already, and they were at least 6 months late before that, yet we've only just recently started seeing the first versions of AMD64 Linux. WinXp for AMD64 still isn't year (won't be for 6+ months), and applications will take time after that. If you wait until the last minute to start shipping 64-bit chips, you won't have an operating system to use it for.

As for the Jurassic Park thing, even if I had the most powerful render farm in the world with nearly infinite resources, I still don't think that I would be making Jurassic Park. That sort of thing takes a bit more than just processing power! :>

Re:Well, of course it won't be in your phone

[Simonetta]

Why is a 64 bit entry by a consumer chipmaker such a big deal? Sure, making cutting edge tech affordable is nice, but the 4GB barrier is not what's keeping the average Joe from making Jurassic Park.

In my experience it's not the speed of my 1Gig Duron that is the biggest limiter of my productivity on a PC, it's the clucky and non-ergonomic software.

For example, Windows message boxes. So easy to code, but a true pain in the ass to deal with. You can't just click left for yes or OK and right for no. You got to take your hands off the keyboard and move the idiot mouse pointer to some tiny little place on the screen.

Another truely stupid concept is the idea (in Windows again, naturally) that positioning the mouse cursor to the exact point in your text where you want to insertion point to appear will ALWAYS highlight the entire text passage, either the word, phrase, or file name. It's driving me nuts, because I can't change anything in Windows that will make me more productive, while I have near infinite flexablity over trivial things like icon title font size.

It's almost enough to make me switch to Linux. Which I suppose will happen someday (if I can ever get over the deep fear and hatred of UNIX that seems to have been embedded into every college student who had to take one computer science class in the 1970s).

But anyway.

We need more better, more flexable, more configurable, and cheaper software far more than we need a 64 bit PC.

This is not a rant or troll, just an observation.

Thank you,

Re:Well, of course it won't be in your phone

[Anonymous Coward]

"For example, Windows message boxes. So easy to code, but a true pain in the ass to deal with. You can't just click left for yes or OK and right for no. You got to take your hands off the keyboard and move the idiot mouse pointer to some tiny little place on the screen."

Uh, ever try using the Tab key to select Yes or No, then hitting Enter? CTRL-TAB, ALT-TAB, etc? Don't need to use the damn mouse.

And then you bitch in the next sentence about positioning the mouse pointer... Don't use the stupid thing, use the arrow and page up/down keys to get the cursor where you want it. No mouse needed.

Smart business decision on AMD's part

[meanfriend]

It's no surprise that AMD is predicting that they will stop making 32-bit processors by the end of 2005 (which is still a full two years away)

It's a smart decision because the new Opteron CPUs are now performing very well against Intel's fastest 32-bit CPUs and AMD thinks they can continue to compete with Intel in the 64-bit arena.

Keep in mind that while AMD is immensely popular for lower cost boxes, the real profit margins are made at the high end. The new 64-bit AMD chips could go a long ways towards making AMD profitable and it's no surprise they want to push high end parts as soon as possible.

It seems to me that lately AMD has lost some momentum with the high end Athlon XPs not being able to keep pace with a high-end P4.

Part of the problem was inflicted by AMD themselves and their + performance rating scheme for the XP line of processors.

While AMD stated that the + rating was not meant to directly compare with any Intel processor, some remarkable coincidence always seemed to result in an Athlon XP XXXX+ benchmarking about the same as an XXXXGhz Pentium 4 yet cost ~20% less. Unfortunately, the PR ratings didnt scale linearly (especially at the top end) and they whole scheme now looks contrived. AMD would be wise to put it behind them ASAP.

You'll notice that the new Opterons now use model numbers that have very little basis for comparison with clock speed from either vendor. Only time will tell if this strategy sells any better...

stop calling the BIOS

[Anonymous Coward]

There is very little reason to call the BIOS any more. Unless you are writing your own substitute OS. In which case, stop bitchin'. It's not supposed to be easy.

Re:stop calling the BIOS

[gl4ss]

** There is very little reason to call the BIOS any more. Unless you are writing your own substitute OS. In which case, stop bitchin'. It's not supposed to be easy.**

isn't it obvious? gillbates=billgates!! his honing up for longhorn!!

Re:stop calling the BIOS

[Anonymous Coward]

I think he's talking about writing BIOS boot drivers that live on SCSI cards and such, these don't call in to the BIOS as much as the BIOS calls in to them! The card cannot set up the process mode and must expose a 16-bit interface for backward compatibility.

Limits of Underclocking

[Detritus]

Many processors use dynamic logic that prevents them from operating reliably below a specified clock rate. Check the spec sheet for your processor before you underclock it.

Re:Limits of Underclocking

[IM6100]

I once underclocked a 286 machine to about 500 KHz. It had a crystal block oscillator that made it run at the standard 6 MHz. I plugged in a slower oscillator. It took a HELL of a long time to get through the POST but it worked. There's a minimum speed that most chips can run, though.

My namesake, the IM6100 processor (the "PDP-8 on a chip" processor, I have tubes of 'em, write if you want to buy one) is all static CMOS and can run down to zero hertz. Well, not zero hertz, but you can use a hamster wheel with a hall-effect switch to drive the clock if you like.

Newton Says...

[BubbaTheBarbarian]

"The abacus will always be the standard of mathamatical reconing for the future..."

64 isn't fast enough

[Cyno]

I need at least 3+ Ghz 64-bit chips, for around $50 a piece before I'll be in a hurry to upgrade.

I want a 2x increase in performance over my 2Ghz 32-bit CPUs and some software to take advantage of those extra bits.

As long as AMD is going to continue disabling their desktop chips, preventing them from working in dual/quad processor SMP configurations I can wait. :)

Re:64 isn't fast enough

[ChaoticLimbs]

I agree. What we really need is a processor that scales well for a cheap price. How about a 64 bit MP capable processor design that stacks like legos? Water cooled multiprocessor lego goodness.

Really, Op systems used in 2038?

[Diakonoss]

How many people expect that widespread use of the operating system currently in use now will be in use in the year 2038 (35 years on the future)? How many systems running now are the same operating systems running since 1968? I thinkg we'll all have more current systems/software by then...

Re:Really, Op systems used in 2038?

[budgenator]

How many systems running now are the same operating systems running since 1968?

it would probably surprise you.

Old microprocessors never die

[TheLink]

"IMHO though 32-bit systems will never die"

Old microprocessors never die. They just end up embedded ;).

Hmmm. That's strange, I don't see something like this in those "Old xxx never die" lists. Looks like a new one :). So here's it for the lists:

Old microprocessor engineers never die. They just end up embedded.

``640K of memory should be enough for anybody.''

[The Tweaker]

Bill Gates has laid down the law!

``640K of memory should be enough for anybody.''

We shouldn't need any stinking 64 Bit Processors!

Re:THANUSES

[RigMonkey]

Wow. This guy must either really bored or really inept.

At least the Inanium is dead

[Animats]

We were all dreading having to program that turkey.

4 and 8 still the kings

[freeweed]

...Add to that 250 million 32-bit chips the much greater number of 16-bit processors, estimated at over one billion per year. Then add another billion eight-bit processors, and another billion four-bitters

From here.

32/16/8/4 bit processors aren't going anywhere any time soon, just off the desktop.

You think that is weird?

[rmdyer]

The WinNT/2k/XP Kernel keeps system time in 64 bits, but the beginning date is 1601!

FILETIME

The FILETIME structure is a 64-bit value representing the number of 100-nanosecond intervals since January 1, 1601 (UTC).

Why-o-why waste so many of those 100ns intervals?

+1

Re:You think that is weird?

[Anonymous Coward]

According to my highly-specialized scientific calculator, that scheme is good for another 58,000 years! Who cares if they started with the year 1, 1601 or 2001. If we're still using Windows and NTFS in the year 60K, we have much bigger problems tha mangled file dates.

x86 bashing

[Anonymous Coward]

x86 may be an ugly hack, but it's not going away anytime soon, because it's far too much work to replace and x86 CPU makers have become so good at working around the limitations and extending x86 without throwing away backward compatibility (like with AMD64). Today x86 is only an instruction set, the CPUs do things very differently behind the scenes.

Comparing modern high-performance x86 CPUs with ARM is pointless IMHO. Making low-power, low and medium performance processors is one thing but competing with P4 and Opteron is quite another.

Make an ARM with big caches, OOO execution with advanced branch predictors, register renaming, SIMD, multiple execution units and clock speeds in the multi-GHz range and you'll probably end up with the same kind of die size and power consumption as the x86 CPUs.

Re:x86 bashing

[tomstdenis]

That's just it. The ARM core doesn't need OOO, register renaming, etc... It actually has more than 4 registers!

As for all the other stuff.. well Quite frankly people are vastly spoiled. 3dNOW for instance has but one use [take a fucking guess]. I'd rather have an inferior ARM core taking one tenth of the power so I can word process, latex and compile stuff compared to my fully bloated Athlon XP-M which prolly takes 25W when idling.

Tom

Re:x86 bashing

[Scott Wood]

That's just it. The ARM core doesn't need OOO, register renaming, etc... It actually has more than 4 registers!

So why do Alpha, PPC, etc. have those things, when both have 31/32 registers?

The proper answer is, "ARM doesn't need them because it is targeted at low power applications, not high performance."

Re:x86 bashing

[Anonymous Coward]

Don't bother to try turning a head of a fanatic that has made his mind with a facts. It's not going to work.

Ha...Win 98

[charnov]

Wow, you put "Windows 98" and "critical software" in the same sentence. I think a sysadmin's head exploded somewhere.

Re:Ha...Win 98

[Agent R]

I wish I didn't. But there are companies that still utilize the older Windows OS to run the earlier software they really need to have.

AMD predicts it will stop producing 32-bit proces

[special_agent]

I reckon Digital Equipment believed the same thing
once the Alpha processor was starting to ship.

Wishful thinking I suppose.

AMD's commercials..

[ProtonMotiveForce]

look like 3rd rate Ron Popille infomercials. It's pretty funny (and somehow, sad).

They make it look like some kind of "news" show, and make all kinds of ridiculous claims about hoe 64-bit CPU's are "forward compatible". Ahahaha. Like Joe Schmoe will see any benefit to a 64-bit CPU. Even super-geeks don't usually have more than a gig of two of memory in their home computers, and these assholes are trying to sell Joe Schmoe a 64-bit "forward compatible" processor? What a joke.

ungrateful goldbricker!

[Thud457]

Goddamn, give me a $1.3e7 signing bonus and I'd at least have the couth to FAKE a positive attitude!

AMD's sort of smart

[LBArrettAnderson]

for saying this.. it's a pretty good marketing scheme to get people to believe their 32 bit cpu's will be out of date so soon in the future so rather they pay extra money for the 64 bit cpu's. i believe it'll be at least 5 years for 32 bit to be "out of date". but amd can say whatever they want to get people to pay more money

Re:AMD's sort of smart

[Anonymous Coward]

AMD have (unfortunately) based the last year of there marketing trying to convince people that their model numbering (2800+ etc) is equivalent or greater than the opposite Intel (2.8Ghz). This hasn't worked, and because on many situations they have fallen WAY short of the mark, they have lost a substaintial ammount of credibility. It will be a while before AMD says anything that people pay attention to, and by that time, Intel will have struck. They can't expect to lie to the consumer and simply asume no-one noticed.

"Those who don't know computers buy Macs,
Those who think they know computers buy AMD,
Those who know computers buy Intel"
- Industry quote

Re:``640K of memory should be enough for anybody.'

[Anonymous Coward]

You know, that's been old for quite some time now, so any time you feel the need to say something funny, please choose a new joke/regretful comment to wear down.

Re:Ah, teenagers. always wanting the trendy machin

[Anonymous Coward]

128K? Sweet.

Although switching those banks in and out make programming anything over 48K a real pain.

But of course!

[porky_pig_jr]

IPv6 is 128-bit addresses, so first we have to see the advent of 128bit PCs. Right?

Reading too much into a comment.

[Vellmont]

I think the original poster meant that 32 bit computing won't die in the same sense that 8 bit computing is still alive, in controllers, etc. The same is true for AMD. All AMD said is they don't see themselves producing 32 bit processors around the end of 2005. The Athlon64 is the replacement for the venerable Athlon, first introduced in 1999. Athlon was a replacement for the K6. AMD stopped producing the K6 a couple years after the introduction of the Athlon, why would they produce 32 bit Athlons a couple years after the introduction of the Athlon64? After the first process shrink it's about as expensive to produce the new higher-transistor chip with the new smaller process/feature size as it is to make the old chip with smaller transistor number, but larger process/feature size. Many people seem to think that AMD not producing 32 bit processors means the end of 32 bit computing. That's obviously ridiculous as it'll take many years before 64 bit OSs are the norm. Remember that the whole point of the Athlon64 over the Itanium is that the Athlon64 has very good 32 bit support to make a transition go far more easily.

How many 32-bitters does it take to run a fridge?

[EmbeddedJanitor]

How many 32-bit CPUs does it take to run a fridge?

None, it's a 4 or 8 bitters job.

Re:Ah, teenagers. always wanting the trendy machin

[bhtooefr]

Actually, it's an Apple II running DOS 3.3 & CP/M 2.2 apps. Both OSes are date-independent (no date stamps). ProDOS 8 WAS date-dependent, and is NOT y2k compliant (ProDOS 16/GSOS is y2k compliant, but that's IIgs, not //e).

Re:Ah, teenagers. always wanting the trendy machin

[bhtooefr]

The 65xx CPUs could address 64K w/o bankswitching. So, it's programming anything over 64K that's a pain (although that 64K includes VRAM - BTW, 80 column apps aren't a pain in BASIC, even though they use 65K RAM (2K text VRAM instead of 1K)).

Re:Obligatory Bill Gates misquote...

[red floyd]

How about this misquote:

64 bits should be enough for anybody

Heard that one before

[Alomex]

By depending on price cuts for Athlon-64 and Opteron, AMD is predicting that it's sales of 32-bit CPUs will fall off and obsolete 32-bit systems in less than 3 years.

I've heard that one before, in 1992 when DEC was working on its Alpha chip. It didn't happen. 64 bit addresses are wasteful in terms of cache lines. Hence the VLIW architecture of the itanium. If you end up with a processor where most instructions (with operands) are 16 bits wide but processed four at a time is that really a 64 bit CPU?

The real reason to go 64-bit

[Anonymous Coward]

I'm looking getting my first amd-64 machine just so I'll no longer get those pesky "terminal too wide" errors!

Persistent OSes...

[reflective recursion]

There have been numerous attempts at designing a persistent LispOS for 32-bit, x86, architecture. The biggest problem is Lisp data has to have typing tags associated with each object. As you can imagine, 32-bits shrinks down quite fast with more tags. A 32-bit address space with a mere 2-bits of tags leaves only 1 gig addressable (2^30). Enough for RAM, perhaps, but not nearly enough for the 130+ gig drives we have today. Perhaps once 64-bit computing it will be finally practical and perhaps possible. You can have a 16-bit tag with a possible 65k range of values (excessive, really) and a remaining 48-bits which can address a decent ~281 terabytes. With this method you could address every possible location in RAM _and_ on drive.

Re:Persistent OSes...

[voodoo1man]

Here's a very insightful Usenet post by Duane Rettig explaining the whole issue with tagging on 32 vs. 64 bit machines. It turns out smaller is better!

Re:Persistent OSes...

[reflective recursion]

It turns out smaller is better!

In what way? Smaller tagging, or smaller address space (64 vs 32)? The only thing I notice the article talking about is the design of Franz's Allegro CL fixnum tagging and how they use less bits in the design of that specific type. It is typical to have a cascading tagging system... where the fixnum usually is identified before other types. This allows fixnum to have a good range of values since fixnums are typically stored in the machine word itself. In other words, you could have a tagging system that used 32-bits, but fixnums would be identified with only 2-bits. Which leaves 62 bits (on a 64-bit arch) of fixnum storage.

Re:Persistent OSes...

[voodoo1man]

I thought the section "why not 8 bits for tag?" was pretty clear on that.

Re:Persistent OSes...

[reflective recursion]

You're pretty vague, there. You said "smaller is better"... quite a broad generalization there. They are talking tags for fixnum values only. Plus it's an implementation detail... I never said "more tags are better." Plus I'm not so sure what this has to do with moving from 32-bit to 64-bit arch.. which is what this topic is about.

Uh huh...

[ThisIsFred]

I'm not in a hurry to ditch any of my 32-bit machines, so long as I get them replaced by 2038.

I suppose you're still running a 16MHz 386 then.

The switch is probably going to happen faster than we can predict. I'm already eyeballing Athlon FX hardware thinking about my next PC upgrade. I don't need this power, but I will buy it anyway, because it'll be 2% faster for the stuff I do.

what the hell are you talking about?

[Anonymous Coward]

IA32 implies 8086 and 286 compatability, jackass.

Re:what the hell are you talking about?

[AvitarX]

Actually in the mode that runs 64 bit code it cannot run any legacy Real Mode code.

They stopped the backward compatability at that point.

You can still boot it in a such a way the it cannot run 64bit code and still use real mode apps.

filled with contradictions

[Anonymous Coward]

If you like programming your own hardware so much, why do you try to get the BIOS to do it for you?

If you owned your own operating system, why not just make it 16-bit real mode so you wouldn't take such a serious hit? Poor you!

You can, of course, use 32 bit (and presumably 64 bit) features in "16 bit assembly" (whatever than means). You will end up with overrides, of course.

The BIOS interfaces are the way they are due to history, not fear of failing a DOS test. The system boots the same way, memory is described the same way, option ROMS work the same way, power management, etc. Changing this means breaking everything for no good reason.

The BIOS will not be 64-bit protected mode because it has no reason to be and this causes no problem to anyone save for retards like yourself. BIOSes have 32-bit interfaces (and protect mode interfaces), as well.

Re:AMD predicts it will stop producing 32-bit proc

[mikefoley]

The difference here, with AMD's processors, is that they are x86's. Alpha was a totally different architecture. The AMD64's will run 32-bit code quite fast. You can still run your Windows 2000 box on an AMD64 in native 32-bit mode. Totally different than Alpha because you don't NEED a recompile.

I used to work for DEC and API (Alpha Processor Inc)
Yea, still out of work 2 years later.

bah, 8 bit CPU's are still out there..

[nurb432]

Perhaps for new personal desktops and servers they *might* have a point, but in the embedded market, 32 bit isnt even here yet.. let alone gone....

They better not hold their breath on desktops either..... 2005 is far too soon...

2038 I think not!

[Anonymous Coward]

Let's go over the math. First of all many clock types are signed 32 bit types on 32 bit platforms. This gives us:
% set max_32_int [expr {pow(2,31) - 1}]
2147483647.0
% #That's how many seconds we can store from the epoch.
% set sec 60; set min 60; set hours 24;
24
% expr {$sec * $min * $hours * 360}
31104000
% #that's one year
% expr {((($max_32_int / 360) / $hours) / $min) / $sec}
69.0420411201
% # 69 years from the epoch will fit
Now if we use an unsigned 32 bit int we would get twice as much time. For example:
% set max_32_u_int [expr {pow(2,32) - 1}]
4294967295.0
% expr {((($max_32_u_int / 360) / $hours) / $min) / $sec}
138.084082272

So, we could conceivably have 138 years of time since the epoch to work with, but because most systems use a signed 32 bit type we are stuck with the 2038 limit of some systems. I suspect that the usage of a signed type is so that some routines can return -1 on error. We could work around that by returning a value via a pointer.

Re:How many 32-bitters does it take to run a fridg

[bhtooefr]

http://www.lgappliances.com/cgi-bin/product.cgi??m =thing&cmd=display&idThing=19

One, running Windows 98 for internet access, and a 4 or 8-bit (maybe even no CPU) for temperature control.

WHAT THE FUCK?!

[greymond]

This just in - in 2005 AMD Predicts

AMD si teh w1n!!!! ALL YOUR CPU's ARE BELONG TO US!!!!

Re:WHAT THE FUCK?!

[Anonymous Coward]

I thought it was funny....

Re:Ah, teenagers. always wanting the trendy machin

[Wakkow]

Nonsense! All we need is Mercury delay lines where each bit is a pool of mercury with pulses causing waves in the mercury determining a bit is on or off. Ahhhhh the good ol' days.

64 bit cpus are not a magic y2038 bullet ...

[dougmc]

I'm not in a hurry to ditch any of my 32-bit machines, so long as I get them replaced by 2038.

64 bit cpus may indeed handle 64 bit numbers internally rather than 32 bit numbers, but that's hardly a magic bullet for the y2038 issue.

If your filesystem only allocates 4 bytes to a timestamp, it's going to break in 2038, 64 bit cpu or not. Any file formats or structures that only allocate 4 bytes to a time value will have the same problem -- and there is a LOT of them out there. And to make matters worse, if you change the format to allocate 8 bytes to the timestamps, then it's almost certainly not going to be compatible with old software anymore.

Also, porting things to use 64 bit cpus rather than 32 bit cpus isn't particularly easy. Yes, you can run in `32 bit mode' and they'll work fine, but many (mostly C) programs work under the assumption that integers are 4 bytes and so are pointers. In a 64 bit cpu, running in a 64 bit mode, this is not true. This really isn't a big problem, however, as the AMD 64 bit cpus can and do emulate a 32 bit cpu as needed.

And we don't need 64 bit cpus to fix the problem anyways -- we could use 2 32 bit ints to store the time stamp if we wanted to. It's a bit more work, but it could certainly be done, even with 32 bit cpus.

Workstation memory and "Moore's Law"

[linux11]

The doubling every 18 months seems to apply to workstation memory just as much as it applies to CPU speed. This shouldn't be a surprise since more applications tend to depend on memory for speed than raw CPU cycles. After all, if a section of code/data needs to be swapped from disk back into memory, all a faster processor can do is more NOOPs or context switch to a different process. So, while around 1980 a nice home computer would need about 64k to be "beefed up," now that about 15 cycles of 18 month periods have passed we are seeing beefed up workstations having around 2GB of memory or 64k*2^15. Next cycle is 4GB which maxes out the address space for 32 bits. We nearing the point where "power" users will start expecting workstations with over 4GB of memory and that definately calls for phasing out 32 bits on the desktop/laptop.

Re:Ah, teenagers. always wanting the trendy machin

[Wolfrider]

--Seriously, if you liked Wordstar - the ' joe ' package provides "jstar" in Linux. I use it for all of my editing that doesn't need a WP.

Addressable memspace!

[CrystalFalcon]

it still has a lot of headroom left

I'm not so sure about that. Addressable memspace (using a signed pointer, which for some reason seems to be the norm) is 2 gigabytes with a 32-bit proc. My last box had 512M of RAM, I just upgraded it to 1G. I predict that less than a year from now, people will realize there is a 2G ceiling and start being uncomfortable about it.

Machines with more than 1G are rather common today. Servers will less than 1G are rather uncommon, at least new ones.

Itanium wouldn't go away if Xeon had 64-bit add

[Glasswire]

The online trade press is (deliberately?) ignoring this , but on app benchmarks (not just HPC) Itanium is kicking butt.

Check out the Top Ten Performance TPC-C benchmarks for on OLTP and 5 of the top 10 systems are Itanium and ALL of the the top 3. An HP Superdome with 64 Itaniums running Oracle 10g was the first ever system to do OVER 1 MILLION transactions per minute.

NO ARCHITECTURE, Opteron, Xeon, Power, (Certainly not SPARC or MIPS) can touch that right now.

Re:Itanium wouldn't go away if Xeon had 64-bit add

[Loki_1929]

"An HP Superdome with 64 Itaniums running Oracle 10g was the first ever system to do OVER 1 MILLION transactions per minute."

And how much does that cost? As I said, the Opteron isn't designed to compete with the Itanium at this level. When you're looking for a comparison, you really need to look at the sub-$25,000 market to see where Opteron's upper-end target resides. Where Opteron will really shine is in the blade market. The low-power chips due out shortly are going to bring a whole new level of performance to these types of servers in areas where space is nearly as important as speed.

Show me how a 1 - 4 CPU Opteron looks in a price/performance ratio with a 1 - 4CPU Itanium box; that's where AMD's upper-end target market is. The large cluster deals we're seeing here and there are fun and nice press, but the sub-$25,000 server (reasonably high volume while maintaining an excellent profit margin) is really what AMD's shooting for. Breaking into this market will pave the way for higher end stuff around 2008 or so. Until that time, it's going to be Opteron (K8) and Opteronx2 (multi-core, K9) that power AMD's vision of the do-it-all chip.

This is really the market Intel would be throwing away if Xeon got 64-bit extensions. Intel wouldn't see any sales of sub-$25,000 servers using Itaniums simply because customers would use Opterons or Xeons. This becomes a problem because Intel's going to face massive competition above $25,000 or so, what with mutli-core Power5s from IBM (8 cores?!), Sparcs, etc. Don't forget that this is still fairly new territory for Intel as well. Itanium has only been sold for about 3 years now, which makes it barely an infant to an enterprise looking into $100,000 systems. The problem Intel faces is that Itanium would be left sandwiched between Xeons/Opterons and Sparcs/Power5s. Right now, Itanium has the entire 64-bit market to itself once you're under the cost for a Power4 or Sparc box. Intel's margins on these babies are its financial counterpoint to the P4 market, which has been massively discounted to compete with AMD.

By the way, I haven't looked just yet, but I'm guessing a SunFire 15k would most certainly "touch that right now". Though, for $6 million or so, one would certainly hope so. Aside from that, I'd bet real money that a nice little cluster of Opterons could also match/beat that mark, at probably less cost. I'm not saying that Itanium is scrap silicon, only that 10 years of joint research with HP plus billions of dollars spent yielded a large, hot Opteron with poor 32-bit performance. :)

Re:Itanium wouldn't go away if Xeon had 64-bit add

[Loki_1929]

"Check out the Top Ten Performance TPC-C benchmarks for on OLTP and 5 of the top 10 systems are Itanium and ALL of the the top 3. An HP Superdome with 64 Itaniums running Oracle 10g was the first ever system to do OVER 1 MILLION transactions per minute.

NO ARCHITECTURE, Opteron, Xeon, Power, (Certainly not SPARC or MIPS) can touch that right now."

Actually, Opteron could do it with 50 CPUs, where it took 64 Itaniums. If we look over here, we see that there are no Itaniums in the top 10 for price/performance. In fact, Opteron leads the pack in this field, with only Xeons competing with it at all. If we look at the performance in a 4-way Opteron system (number 8 on the price/performance list), we see that Opteron scales better than linear, and that roughly 50 Opterons would overtake the top Itanium system, which uses 64 Itaniums. Some rough calculations assuming a basic $55,000/CPU figure (ok, extremely rough estimates) based on the 4-way Rackserver system yield an Opteron-based system that outperforms the top Itanium box at a cost of around $2.8mUSD; a far cry from the $8.4mUSD Itanium system. Don't even try to come back with, "well a 50-CPU Opteron system doesn't exist", because neither does the system you're talking about. If you'll check the availability, you'll notice that it lists a future date, 04/14/04.

If that's considered "kicking butt", then I'd love to see what you'd consider calling a 64-CPU Opteron box, which should show something along the lines of ~1.3 million transactions per minute.

I'd say that's an architecture that goes a bit beyond 'touch'ing the Itaniums. ;)

Re:Itanium wouldn't go away if Xeon had 64-bit add

[Glasswire]

The original question was: would there be a role for Itanium in a world where Intel offered extended addressing for Xeon? My point was not about Itanium's price/perf but it's pure performance. If Opterons were faster in absolute terms then they would be in the Top Ten perf list not just the Price/Perf list. Instead, unlike the million+ tpm the best Itanium system supported, the best Opteron result was a piddly 20k (Because Racksaver is only showing results from ONE cpu. -if it scaled, maybe showing MORE than one cpu would make sense...) Anyway, if your solution really needs the high thoughput, it doesn't matter how cheap it is if you can't do the job. Price/perf only applies when you meet job criteria.

Your point about the Itanium system not existing, (like your 50-Opteron-box) is interesting. The Superdome certainly exists, (if not, what did they run the test on?) it's just not shipping yet - as opposed to the totally hypothetical Opteron system that exists only in your head. No one (that I know of) has built a large scale SMP or NUMA (non-clustered) architecture with Opterons (Hypertransport makes small sub-8-way SMP and clustering much more attractive) so they are not playing in this space yet. I'm skeptical about whether this would be worth the effort for Opterons -unlike Itaniums which were designed to be very 32-way+ scalable (explict parallelism, very large caches etc.) As for the performance of your multi-way Opteron system, you may want to read some CompSci texts on how SMP scaling works (hint: simply mulitplying 1 proc perf by number of procs is NOT the way it works). And one would think that the big system vendors who are adopting Opteron, IBM & Sun are not going to vandalize their own lucrative large-scale Power & SPARC markets by building inexpensive scaled Opteron solutions. If they don't, who will? Racksaver? Again, people will try to solve some of these problems with clusters, but there really are applications for which a large single image partition is what's really needed.

As long as some users need higher perf, bigger high-end iron, Itanium will do better. And as long as there is an installed base of IBM Power, Sun SPARC and other underperforming, overpriced (YES, relative to Itanium) RISC systems out there, there are customers with applications that need that large SMP perf and Itanium has a role that Xeon and Opteron aren't filling - even with or without memory extensions.

Re:Itanium wouldn't go away if Xeon had 64-bit add

[Loki_1929]

" If Opterons were faster in absolute terms then they would be in the Top Ten perf list not just the Price/Perf list. "

Using this logic, if I take 200,000,0000 Pentium Pro CPUs and put them into a system/cluster, then any system/cluster that doesn't beat it is slower in 'absolute terms'. They're not at the top because there were no systems listed with nearly that many Opterons in it. That's not because they don't exist or are not possible, but because no one has submitted the results of a specific benchmark to this specific site.

"Instead, unlike the million+ tpm the best Itanium system supported, the best Opteron result was a piddly 20k "

The best Opteron result on this specific benchmark submitted to this specific website was a 4-way system (that was a link, by the way) from Racksaver which trashed every Xeon and PIII Xeon system within $100,000 of its cost, with the exceptioon of one, (also a link) which was a mere $80,000 more expensive. But hey, what's $80,000 when you're looking at buying a $225,000 server? Oh yeah, 35%. That "piddly" 20k score, by the way, beat out 27 Intel machines running Xeons and PIII Xeons. In terms of price, this $42,000 machine outperformed everything Intel had up 'til you hit a $99,000 dual 2.8GHz Xeon system.

" (Because Racksaver is only showing results from ONE cpu. -if it scaled, maybe showing MORE than one cpu would make sense...)"

As would reading my post, or looking on the site itself. The system is listed here, and I posted a link to it specifically in my last posting. It's a $223,000 quad Opteron 844 system from Racksaver.

"Anyway, if your solution really needs the high thoughput, it doesn't matter how cheap it is if you can't do the job. Price/perf only applies when you meet job criteria."

Which you can do if you put together a system with enough CPUs. As I said, you can do the job of 64 Itaniums with around 50 Opterons. That's a conservative estimate, mind you. Someone putting together a more optimized solution may find even better results. Don't forget as well that the Itanium box is running on HP-UX, whereas the Opteron was running under Windows 2003. A more fair assessment would show results under something like Linux. Once FreeBSD's SMP support is improved, it would provide an even more interesting look into performance differences, considering its incredible high load scalability.

" No one (that I know of) has built a large scale SMP or NUMA (non-clustered) architecture with Opterons (Hypertransport makes small sub-8-way SMP and clustering much more attractive) so they are not playing in this space yet. I'm skeptical about whether this would be worth the effort for Opterons -unlike Itaniums which were designed to be very 32-way+ scalable (explict parallelism, very large caches etc.)"

AIST in Japan, Los Alamos National Labs, Texas A&M University and The University of Utah are putting together large-scale Opteron systems as we speak. While these are indeed all clusters, we're just talking about performance, right? Many of the systems listed for the TCP-C benchmarks are indeed clusters. In terms of Opteron's cache, AMD already has plans to launch them with larger caches. We'll most likely see those debut around the same time as the multi-core Opterons. Oh yes, Opteron was designed for multiple cores right from the start, and so will get them most likely in 2005. Oddly enough, that's the same year Intel plans to introduce multi-core Itaniums. Odd how they changed that from 2007 to 2005 right after AMD started answering questions about Opteron's multi-core future.

"As for the performance of your multi-way Opteron system, you may want to read some CompSci texts on how SMP scaling works (hint: simply mulitplying 1

Re:Itanium wouldn't go away if Xeon had 64-bit add

[Glasswire]

Thanks for conceding my original point:

In terms of Opteron not hitting the high end of the market, that was simply never their intended market.

But, unlike Xeon and Opteron, Itanium WAS intended to hit this market and as long as there's other less attractive cpus being sold out there like Power and SPARC and as long as commodity processors like Xeon/Operon can't fill that role, Itanium has a market. Either you ARE arguing that Xeon with 64 bit addressing OR Opteron will make Itanium unneeded or you're not. Please decide.

But I wouldn't assume Intel doesn't think there's a role for both architectures.

Re:Itanium wouldn't go away if Xeon had 64-bit add

[Loki_1929]

"Either you ARE arguing that Xeon with 64 bit addressing OR Opteron will make Itanium unneeded or you're not. Please decide. "

I've already decided. To quote myself in my previous post:

"[If Xeon goes 64-bit] Itanium will have a small, niche market alongside IBM's and Sun's offerings. If Intel can't push Itaniums en masse, then it stands to lose an incredible amount of money and clout. Selling 100 systems a year on that chip just isn't going to sustain the R&D."

And later:

"The point is, is Intel going to be able to build Itanium sales to the point where it's profitable and can sustain marketing and R&D while paying back all the money that's been spent on it in the past 10 years? That's really the essential question upon which Itanium's future rests."

My point in all of this is that without the smaller end of the market, which would be absorbed by 64-bit Xeons and Opterons, Itanium's only remaining niche is at the upper end, where sales are too small to sustain R&D + marketing + paying back 10 years of prior R&D.

"But I wouldn't assume Intel doesn't think there's a role for both architectures."

I think Intel is hoping there's a sustainable market for Itanium post-64-bit Xeon. I think they're scrambling to crunch the numbers and figure out what Itanium's market figures look like in a 64-bit Xeon world. I also think that Intel will continue to sell Itaniums, even if they lose a small amount of money each year thanks to R&D plus marketing costs, simply so they can continue to have a presence in the upper end of the market. If the numbers come out looking like Intel will lose money hand over fist as soon as 64-bit Xeon debuts, then I think it turns into a debate about how much it's worth to Intel to have the clout of Itanium. Once their foot is in the door of a large corporation thanks to a high end server based on Itanium, then they may be able to push enough Xeons through that same door to make up for the money lost sustaining Itanium's life.

In any case, I think that Opteron really has to make Intel sweat. Opteron competes with Itanium in a number of ways, and really forces Intel's hand. A current Intel customer with a need for 64-bits who can't afford the 2, 3, 4-way Itanium server could very well opt for an Opteron box instead. To avoid this happening en masse, Intel releases 64-bit Xeons. Companies a bit further up the food chain who were looking at Itanium servers suddenly see Xeon and Opteron machines that perform better at lower prices. Now they're only looking at it as Xeon vs Opteron, with Itanium no longer in the picture. The second type of customer is really Itanium's bread and butter. They might only buy 5 or 10 Itanium-based servers, but each has 2 or 4 CPUs. Multiply that by the number of customers who exist within this segment and you get into what makes this area so vital to Itanium.

But as I said before, Intel has more than enough money to sell Itaniums indefinitely, regardless of losses. The question then, however, is for investors: "do we really want our company selling a 'losing' product that doesn't look like it's going to recover any time soon?".

Let's see..

[Ancil]

What's 2005, in AMD years?

Hype

[jav1231]

It's all fine and good to talk about EOL'ing the 32bit processor. Notice they're not talking about the end of 32bit code. Therefore, this is largely marketing. *sigh*
Marketing reps suck....

Re:Ah, teenagers. always wanting the trendy machin

[mcrbids]

Oh, and don't forget:

Ctl-K B - start highlight,

Ctl-K K - end highlight

Ctl-K D - Delete highlighted text.

Ugh! I hated it. I always used the shareware PC-Write by QuickSoft. The link here provides a neat little historical summary of the WP.

If anybody ported *that* to Linux, I'd be in hog heaven... Shift-Ctl-Backspace would finally do what it should!

32-bits on BSD

[Anonymous Coward]

...as I sit at my toy keyboard and whistle a happy tune, 32-bits - you died too soon! (must've been because I ran BSD).

Forget 64, Where's 4096?!

[TexasCowboy23]

I'm ready to throw 32-bit out. I'm nearing purchase of a dual Opteron motherboard along with two Opteron 242 or 246 (haven't decided yet). But why stop at 64? Good Lord ... Move on to 4,096-bit processors. Scratch that ... Make it a 3^64 trigit (trinary digit) processor. Or even a 4^64 quadit (quadrinary digit) processor.

64-bits my ass. If this is the future, where's all the *good* technology? :)

If it doesn't have a 256-bit wide adder or shifter

[Ayanami Rei]

...it's NOT a 256-bit CPU.

There's your litmus test. No more discussion.

Re:Bush is beloved in UK- Guardian

[Anonymous Coward]

I, for another, would rather not see Bush come. I think Willy did that enough with that intern.

Life will be good when I have..

[Anonymous Coward]

64 bit PC, 64 bit OS, IPv6, Duke Nuke 'Em SixtyFour Ever

Re: Umm, you mean EGA quality graphics?

[benzapp]

This machine could barely manage a few megahertz, yet it was scaling/rotating/blitting out hi-res graphics with millions of colors.

I think you meant 256x244@with 256 onscreen colors.

Maybe you are young and don't remember ;) The rotating was cool though. I always loved the Actraiser opening sequence. What a cool game.

*cough* paged *cough*

[Ayanami Rei]

You're still limited to 4GB-(kernel prot.) per process.

You're supposed to treat the upper 16-bits...

[Ayanami Rei]

as "reserved", and always set to zero.

This will probably lead to a extension in later revisions of the chip with new paging schemes that enable the upper address bits, but not break compatibility with older software.

Are you kidding me?

[Ayanami Rei]

I've got my Opteron at work (to cheap to buy one myself) and I am content.

(no really, it kicks a large quantity of ass. At least 3700 furlongs)

Re: Umm, you mean EGA quality graphics?

[AKAImBatman]

Okay 32,768 colors. So sue me. :-) Since most games were very cartoony in nature, it wasn't usually obvious that the screen was palette swapping. And IIRC, the more advanced scaling and rotation stuff was called "Mode 7" or something like that. The first game to show it off was the second game released for the SNES: the psuedo 3D "Pilot Wings".

Ah, memories. Of course, with hit titles like "Sim City", "TMNT 4", "Street Fighter 2", "Contra III", "F-Zero" and "Zelda 3: A Link to the Past", the SNES was a shoe in.

Re:64-bit processors

[Anonymous Coward]

In that case, you must still be using an 8-bit Amiga to post, because you failed it.

The benefits outweigh the costs.

[Ayanami Rei]

Comparing user-accessible 64-bit to an SUV is assinine. SUVs have large costs that are only justified if they are used to their full extent.

AMD64s allow programmers to simplify memory management, access large integer datatypes easily, and potentially improve throughput. It's 32-bit compatible without bending-over backwards. They allow users a much less restricted upgrade path.
All in chips that run cooler clock for clock.

On the down side, the data cache might be used less effectively. And some OSs need a little bit of extra logic to keep 32-bit code and 64-bit seperate (mostly for performance reasons).

That being said, you don't need AMD64, G5 or any of thatbut it's a really good idea

I have lost all hope in humanity

[Anonymous Coward]

because the parent post is currently moderated as "Score:1, Informative". ::emoticon::

Super Emacs

[superpulpsicle]

Wow with a 64 bit processor my PC can finally...

# ./emacs3D

# echo and talk to me when I am bored

# make coffee with gcc

Unfortunately all I want is a 100baseT broadband network in hmmm 5 years.

Don't do as I say; do as I do

[rixstep]

Everyone's predicted the end of 32-bit computing. Some real pioneers have already predicted the end of 1,024-bit computing - somewhere four millennia from now, that is.

But AMD has always impressed me as a straight-shooting company. And if they stop producing 32-bit processors, that's a pretty good indication where the industry will go.

Now Apple are already on their way there, so it's no fantastic news exactly. But it's welcome.

Re:Ah, teenagers. always wanting the trendy machin

[Sivaram_Velauthapill]

Shouldn't geeks be LEADING? I think you are the opposite of a geek...which would be... hmm.. a caveman ;) When all of us start using 64 bits, you'll still be using 32 bits ... oh the horror >:o

Sivaram Velauthapillai

I have been using 64-bits for 10 years

[hughk]

Even on the desk top. It was called Alpha and it still works very nicely. HP being idiots are trying to push the users onto unobitaniums.

Actually, when Digital looked like they were going belly up, a bunch of the engineers went to AMD. I therefore have high hopes for Opteron/Athalons.

Shortage of Athlon MP's?

[wolverine1999]

This summer I wanted to upgrade my athlon mp system to a dual Athlon MP... and guess what, there's a shortage of Athlon MP's..

Re:Shortage of Athlon MP's?

[Anonymous Coward]

but not because of the demand, but because of the lack of demand.

Re:Bush is beloved in UK- Guardian

[ThaReetLad]

I happy he's here. It makes it so much easier to smack him around the head.

Re:AMD predicts it will stop producing 32-bit proc

[hughk]

Mike, I used to work for DEC and API (Alpha Processor Inc) Thats tough. Alphas are still going strong for two of the world's largest electronic financial exchanges running VMS. They are being promised a bug free Itanium which is good, because with the killing of the Alpha, they are relying on quiet markets to last until VMS/Itanium is ready. They are conservative too and it took until the mid nineties before they moved the hosts to Alpha, so I guess they won't move immediately.

To move to Linux/cheaper hardware at the host side would be extremely difficult. The new processors look great but Linux still lacks in some important areas (distributed lock manager).

Re:``640K of memory should be enough for anybody.'

[Anonymous Coward]

Bill Gates claims he never said it. And a quick googling turns up no reliable sources for when he was supposed to have said it (some make vague remarks about "some" tech demo in the early PC days).

I think it's quite conceivable that he never actually said this.

Also, the 640k limitation was enforced by IBM, not MS. MS just built a design around that limitation that haunted them for years and years.

Re: Umm, you mean EGA quality graphics?

[Anonymous Coward]

Ah zelda. Gladly did I give many a day of my illspent youth to your cause. Oh, how I long for the long-passed age of saving maidens from evil mages and actually enjoying it plentifully.

Nowadays, I want to be the evil mage doing stuff to maidens. But then between the first time I saved zelda and today lies the thing that turns innocent good-natured boys into evil perverted sicko's, puberty.

Re:AMD predicts it will stop producing 32-bit proc

[mikefoley]

I think Intel has positioned the Itanic as an Alpha/Sparc replacement and never really meant for it to go into things like laptops. AMD on the other hand sees AMD64 as a complete replacement of its major product line.

It'll be interesting to see how VMS does on Itanic. My fear is that HP will keep VMS in the "cash cow" position however. Too bad. It's still the worlds best OS. (I was the system manager for the VMS Development group about 10+ years ago)

Why I need 64 bits

[suitti]

There are several reasons I need 64 bits.

I need files that are larger than 4 GB. I've needed this for 5 years - ever since 2.3 GB hard disks became affordable for home use. We could have this now, with 32 bit procesors. Under Linux, the only way I know how to do this is to stream to a raw disk (no filesystem). I'm tempted to do this for backups - one disk to another. However, I'd like more than one backup image on the backup drive.
cd / ; tar cf - . | gzip -9 > /dev/hdb1

I need more than 32 bits of integer arithmetic. bzip2 is limited to 512 MB files (2 ^ 29 bytes), because it needs to calculate addresses to bits, and is limited to 32 bits. It probably could have been written using gcc's long long's (which are 64 bits), but it would likely hamper file portability. bzip2 can not produce a single archive file that fills a single CD.

I need more than 32 bits of RAM addressing. An app I wrote appears to require about 8 GB RAM for a full production run.

Yet, most of my applications ran fine on 16 bit machines.

I would have picked up an Alpha if I could have afforded it. This shows that just being the fastest machine on Earth does not get you a spot on the desktop. Did DEC offer last year's model for cheap? The only low end Alpha's were on the used market, as far as I know.

I would have picked up an AMD64 rather than an Athlon, if I could have gotten one cheap enough. My old machine died, so I had to upgrade. It was a surprise, and I didn't have much extra cash. I picked up a low end 1800+, but with a good motherboard, expandable to 3 GB RAM, and 3200+ chip. I may add RAM, but, so far, I'm happy with the speed.

Re:AMD predicts it will stop producing 32-bit proc

[hughk]

I don't even know what HP really wants to do with VMS. Compaq didn't no where to put it and HP really have it on the backburner. A shame, because an Alpha with VMS is reasonably fast and exceptionally reliable. The banking industry seems to have gone now to IBM and big Suns.

The word is that Itanium boots VMS now, but I don't particularly like the chip. The Alpha was cleaner and it was a real pity that it couldn't be mass-produced on the scale to bring costs down to desktop level (assuming that desktop

I was the system manager for the VMS Development group about 10+ years ago

Ah, thats where I know you from. I remember the name. I was somewhat active in DECUS Europe in the early nineties but later was too busy doing small consulting gigs mostly around exchange software. We had two platforms at the user end, AIX and VMS. This has now been replaced by Win and Solaris, driven mostly by the market.

Subsequently, I was working on a major gig out in the former USSR and got talking with people that knew VMS well from the Robotron days. HP got a very good foothold in Russia because after the fall of the USSR, Digital refused to give reasonable terms to license VMS (up until those days, their VAX/VMS was pirated). HP stepped in with some very attractive deals and they hoovered up the market. Another of Digital's marketing cockups.

Re:AMD predicts it will stop producing 32-bit proc

[special_agent]

Good points.

BTW: DEC, RSX11, VMS, OpenVMS and Alpha used to be my bread and butter. OpenVms in many ways, is still the best OS around in my opinion. If back
in the old days, Ken Olsen had released VMS under some sort of GPL
license, can you imagine how different the world might look today?

Now I'm a Linux head. Hard to resist e=mc**2.

Best regards,

Jeff

Only on slashdot...

[Loki_1929]

Only on Slashdot can a comment which pokes fun at a multi-billion dollar company's decisions on the basis that they don't fit the individual poster's immediate or predicted needs be modded 'insightful' while another comment from a poster who uses a bit of sarcasm to point out the fact that future market does not depend entirely on that individual is modded as a 'troll'.

It's these little idiosyncrasies that put a smile on my face and keep me coming back for more.

Seems reasonable to me... and I want one now!

[slagish666]

I work in a lab which has 24 P4 machines and we are to upgrade in the next year. We run lots of statistical software such as Stata, which require a lot of memory in order to grind those numbers.

I'm looking at trialing an Athlon 64 FX-51 machine with about 4GB of memory. It seems to me that this is the only realistic upgrade route, since we do need the extra memory, and whatever Intel can supply in the next year will only handle 2 or 4 GB max of RAM (and in our Dell machines, there's only room for 2GB of RAMBUS memory, which we've been told will cost $3,000 (CAN) -- the price of a new high powered workstation!)

As long as we can keep the noise down from these machines, AMD will have our business. I just don't see any other way to go right now.

Re:How many 32-bitters does it take to run a fridg

[EmbeddedJanitor]

The Win98 thing doesn't run the fridge. It is basically just a fridge with a built in PC.

Re:How many 32-bitters does it take to run a fridg

[bhtooefr]

I know. That's why I mentioned the 4 or 8-bitter.

Insane, though. I could build a sweet OCer, put it in my $200 fridge, cut out a hole for a touchscreen, and put it in for $1000. This is $8000.

Re:How many 32-bitters does it take to run a fridg

[EmbeddedJanitor]

And even with a Pentium, the ice compartment should be able to keep it cool :-)

But why, I ask, put internet access in your fridge. It isn't as if you want your girlfriend to get caught up on ICQ while she should be fetching you a beer.

USA Never went to the moon

[Anonymous Coward]

Appolo mission never took place and the evidence of that is there is no wind on the moon and when you look at the picture the flag is waving.