Slashdot Mirror
AMD Announces 1GHz Athlon Imminent
Foxpaw writes, "Not to be outdone by Intel, according to ZDNet UK, AMD is planning on simultaneously releasing the 900MHz, 950MHz and 1GHz Athlons, maybe as early as later this month. " I do have to say that the corporate peeing match between these two is pretty amusing - but if Intel is still having production problems, then AMD can continue to leverage their high-end chips.
This article claims AMD is going to announce their 1GHz Athlon on Monday. Of course, one can only imagine what supply is going to be like. ;-)
sigs are a waste of space
"It's all about the Bejamins baby!"
This sums it up. Progress in this whole industry (hardware at least) is so very incremental. There is waaaaaayy too much $$$ to be made by putting out a little at a time. I think it is too funny (almost shamefully so on Intel's part) how just before the beginning of this year (end of 1999) we were hearing it would be third or fourth quarter of 2000 before we would see 1 GHz proc's running around. Funny how a little competition changes things and shifts paradigms. Moore's Law, Schmoores Law! Sometimes, I think that "law" (and I use that term very loosely) was only created for revenue purposes. It just keeps a steady, bankable flow in of cash coming in for Intel. Now, AMD has blown that "law" out of the water by taking the fight directly to Intel, then one little slip up in production schedules an BAMMM!! the carpet gets pulled out from beneath you.
Also, I remember not so long ago when IBM (maybe Motorola too?) comes out and says, "Hey we're going to use Copper interconnects, screw Aluminum." The word from Intel? "We're gonna stick with Aluminum. We still have a ways to go with it before we exhaust it's possibilities." What a load of crap!! They just wanted to hold out on using Copper so they can milk Aluminum (and milk the customer) for all it's worth. Well, that's MHO anyway. Anyone else agree? Disagree?
Official Disclaimer: Mind you, the above statements are not the original statements from the above mentioned companies, merely my recollections (to the best of my meager abilities.) Also, the opinions expressed above are not necessarily those of Slashdot, or me for that matter
kuro5hin.org
Co-founder and designer at Music Nearby: http://musicnearby.com
Intel and AMD keep releasing faster and faster chips but
1. They have trouble producing them in quantity.
2. They prices are simply ridiculous. Who the hell needs a 1GHz CPU if it costs more then an average computer?
So, in light of this, I am wondering: do they actually manage to sell any of these "super-fast" CPUs? And if they do, what kind of luser would buy them?
___
___
If you think big enough, you'll never have to do it.
Errr... my apologies.
I was moderating, and this idiot microsoft scroll mouse bumped this to offtopic while I was scrolling down. There is no verification before moderation, so the parent of this post got bumped to offtopic, which is not at all what I intended to do.
Other moderators, else, please moderate this back up.
Poster, please accept my humble apologies.
Scroll mice owners, carefull, if you happen to drag the mouse pointer over a drop down box while scrolling down (not hard to do), the value in the box will be silently changed.
Slashdot architects... could you give a "verify" screen before committing the moderation activities?
Microsoft... the scroll mouse is like a decent idea with a flakey implementation. Can't you just bind the thing to the main window scroll bars? That is really the main place it is usefull, as a drop down dialog selector it probably creates 10 errors for every proper action I use it for.
Everyone else... sorry for being offtopic, I am just trying to correct my mistake...
Bill "so much for my karma" Kilgallon
Mathematically impossible requirements are technically not against policy.
This kind of one upmanship is excellent on the whole for consumers like myself. However, I am going to be more than a little wary of new chipsets and chip designs in this kind of a cycle of fierce competition. Remember, testing is the easiest part of development to shorten, at least in a manager's eyes. These chips already have cooling systems from (or perhaps for) hell, and I'm going to wait until a certain ammount of "public beta testing" is over before I put them in a production system.
Thank you for not thinking.
I'm so sick of morons asking for dual Crusoe motherboards/setups. Are you people retarded? The Crusoe is designed for low power low profile devices, this means it is designed for things the size of laptops, not 2' beige towers! A dual Crusoe would mean a complete rework on the VLIW translation code and the ability to buy these things retail not to metnino MORE POWER. You're not going to buy these retail, they're going to be supplied by OEMs. Dual machines come from the need for beacoup processing power, the Crusoe is not designed for beaucoup processing power.
I'm a loner Dottie, a Rebel.
Since the marketing people at Dell, Compaq, and Gateway realize their consumer base is composed of idiots and buffoons they hype up these "ultra-fast" processors and tell people it will last them into the next geological epoch. However people with a little more technical savvy and reasoning skills (we also tie our shoes without help from our mommies, most of us anyways) realize that higher speed doesnt mean better performance. The anaology between a Ferarri and a Beetles is that both will get you there but one will get you there faster, that may be so but the Beetle will be able to go farther because it doesn't use as much gas and run for a longer time because it's less likely to break down because of less stress on the engine. Where consumers ought to start looking isn't faster processors, merely a larger number of them. Multiple processor setups will greatly increase the speed and efficiency of programs if they're written with multi-threading in mind. People who build servers and mainframes like Sun and IBM have realized this already. Now all it will take is a trickling down of the technology to the average consumer. The price/performance rate is much better on a dual setup than buying a 1Ghz processor. Be has really excellent SMP performance because all of it's apps are designed with SMP in mind, only a handful of Windows apps are and Linux apps usually need some code tweaking to work correctly. I'd love to see consumer SMP boxes, especially from people like Apple who are really creaming over media editing on personal systems (Be has already worked through this idea). While we're migrating to SMP set-ups, lets kill off ATA please.
I'm a loner Dottie, a Rebel.
Has anyone had any thoughts as how these chip frequencys might affect 900mhz phones? If at all? Maybe cause some problems if you have a phone near a fast box?
Microsoft aggravates my tourettes syndrome.
Is that these chips aren't as good as they ought to be. Many moons ago, before the Athlon, Intel was slow to put out new chips. They always had to attack the problem of getting Really Fast Cache(tm) to go with the Really Fast MHz(tm). The chips would only increment about 50-66 MHz in speed, but would actually show a decent performance increase.
/.'er can tell you that there's more to speed then just MHz, or we'd all just be using cache-less Celerons clocked at 2 GHz. AMD and Intel are just showing off who has the bigger [censored].
Today? It's a blatant race for pure MHz. Your average
Who said competition helped everyone? It certainly doesn't mean jack-all in a world where only one number means anything.
Have you ever used Office 2000? Did you ever see what was out in 1992?. Geez, in 1992 everyone was just upgrading from PFS:Works to some gui product.
Nowadays, with Office 2k, i got dynamic data exchange, ODBC connectivity, easy templates, WYSWIG (wasn't perfected in 1992 quite yet), support for more then 32k records/files. (when your saving history or documents for years on end for auditing reasons, the 1992 software won't cut it).
The list can go on and on. Alot has changed since then people. Mhz = power, and power = control.
He who controlls the spice, controlls the universe. And yes, there is lots that can be done, but it is simply remarkable what has been done.
Remember, some of us slashdot readers were born before 1980 and remember what was around and what has changed.
It kind of reminds me of amiga, seperating the subsystems and processing through multiple chips.
I think distributed os's and computing is the future! but hey, x86 and sparc architecure and specs are abundant.. when will we see someone run with it and create an open system?
"Nobody needs more than 640Kb of memory." - Bill Gates
On a more serious note, the chips will be in high demand for (inclusively, not exclusively) servers and gamers. The fact that more people are using the internet means that the demand for a faster server is going to always increase. As for gamers, computer games are probably one of the few genres of applications that continually push the limit of the state-of-the-art computers nowadays. You think having GeForce is enough? Now that we offload the geometry and lighting to the video card, the CPU can spend more time creating more lifelike and realistic AI.
And of course, chips introduced at a higher speed means that the lower-grade CPUs will have their costs slashed, which is a definite good thing for consumers.
And finally, do you expect Intel and AMD to close down their R&D because "nobody needs faster comptuers anymore"? there's always Microsoft to make them look slow ;)
Really? I'm surprised. I went with the FIC Mobo (only problem I hit was its size... it takes up the FULL tray which can be a problem if your case isn't deep enough) and haven't had a problem (FYI I'm running Win98).
It was the first one that AMD listed as being certified for their 800mhz chip (although by now so is everyone else). When I check AMD's site they no longer listed the K7M from ASUS _anywhere_ on the motherboard page. Could they have dropped it due to ASUS's lack of acknoledging the product existed?
This space for rent. All reasonable inquiries will be entertained at proprietors discretion.
Yeah, kind of funny how the "Coppermine" still uses Aluminum interconnect! AMD have already moved to copper in their Dresden mfg facility, so Intel's definitely going to have problems keeping up with AMD's speeds.
Apparently Intel is having *hideous* speed problems with Merced/Itanium. AMD's "Sledgehammer" may well crush Intel.
>* Dirt cheap price (i.e. a fast CPU for $20).
Well, a K6-2 500 runs ~$45-50 now, which is pretty darn cheap...
>* Negligible power consumption.
D'Oh!
>* Much smaller, less clunky systems (of which a low power, cheap CPU is an important part).
Well, not everyone always really wants this. That's great for nice products, like a Palm, or laptop, but I'd still rather have a 2' tower under my desk, with expansion room, and room to add horsepower as I see fit. It all depends what you want it to do... I wouldn't use a laptop for CAD or numbercrunching - I don't like the limitations of portable devices. I *don't* want my TV, microwave, etc enabled with networking... dreamcast networking is a neat step forward for that platform, and I think centrally controlled lighting (X-10, etc) is a great thing, but thesee things don't need real CPUs.... Many people think they *need* something, and end up never using it... I like to make the most use of what I have...
"It's tough to be bilingual when you get hit in the head."
They can take up a lot of extra wires and gates to try to overcome some of the higher speed problems... differential designs are great, but at the expense of increased chip design. At least you can be sure when your data is stable without clocking, though. Really neat stuff.
"It's tough to be bilingual when you get hit in the head."
I've been told the newer chipset versions have been more stable than the originals, too (stands to reason that things would improve). Asus has a great history of stability, and they have always impressed me greatly. I know someone who just got one of the new FIC Athlon boards, and he's been stress testing it for about 6 days now... so far so good (lots of apps/compiles/games all runs with scripts, most of which serve little purpose than to test all available paths and try to create tough situations for the board/chip). One thing to always check is that the chipset is properly cooled... it never used to be much of a problem, but ever since the BX started pushing to 100MHz, and now others are doing that and more, with more buffers and all sorts of good stuff, heat becomes a problem, especially when sitting beneath a sizzling processor. Just something to think about.
The CPU may be the brain of your system, but a top notch motherboard really makes the difference.
"It's tough to be bilingual when you get hit in the head."
Regarding Moore's Law (or hypothesism which is a more accurate description) - it can be maintained for a while usng x-ray lithography along with newer SOI and copper technologies, and in a few years, maybe IBM's electron-beam lith will be cheap or fast enough to use with larger scale mass production cards, but we'll have to see. Humans are amazingly inventive creatures who overcome seemingly impossible physical limitations via little tweaks and large leaps (with the tweaks sometimes giving the insight necessary to figure out the leaps...).
As always, the benefits fall to the consumer, who is pretty happy with a 200/333/450 MHz PC processor, but boy, can Excel go fast now, or just look at how smooth Quake IX Arena is now! Most people really don't need such capabilities for their PCs, but gamers do, CAD designers do, coders love fast compiles, accountants are made happier by fast refreshes, and everybody loves it when everything just 'feels' faster.
A lot of the technology research that is motivated by processor market and communications market allows for greater gains in other highly specific areas... look how far graphics cards have come in the last few years. Everyone gains from this, as long as they realize that even last months technology is good enough for most people (see the Gateway commercial where the guy is bringing home a T5 (can't help thinking about Cray) and sees the signs put up for a T6)...
Of course, if things continue to push the way they have, we might continue to see more power-sucking features that force out old computer faster (fade in/out pull down menus in Win2k?! I mean, really - it's a menu, not an ancient scroll from D&D). I haven't upgraded a processor in any of my boxes for over a year now (the horror), but a couple running at 450 leaves me pretty happy, and I'll wait a little while before I make a large leap to a really screaming CPU/vid upgrade... I also just spent a lot moving into a new house, so that saps
the tech budget a little, too 8^)
Life is good in the commodity PC market for buyers, though. Take advantage of it.
"It's tough to be bilingual when you get hit in the head."
No, she doesn't need a lot of things... she doesn't need a laptop, she doesn't even need a full computer (maybe a thin client). Specific devices for specific needs is the idea. There are a lot of applications for the low-power, dirt cheap processors, there's no denying that... it's just that a blanket statement like the following bothers me:
>Could we please get back on track and work on something that computer users really want, like
[snipped the list]
When one speaks for *all* computer users (without consulting us all first 8^) you take a chance... therefore, I offered a diffrence of opinion, which is why we get to comment on stories.
Another salient point is that pushing the performance limits leads to lower power consumption per MHz/MIP/whatever, and really *does* help out the dirt-cheap low-power stuff. As the newer lith processes improve, less voltage is required to drive the chip, so power can decrease. Yes, a 1GHz chip will create a lot of heat and use a lot of power, but the same process used on a 500MHz chip will use far less - power changes linear with frequency, and with the square of the voltage difference (a lower voltage can be used at lower frequenciesm since less current is required, due to longer setup times). If we can produce a 1.5GHz chip that sucks 50 watts at 2 volts, a 500MHz chip (same core) that runs at 1.2v would use 6 watts, and I'm sure you could make a much simpler core that wouldn't be as power-sucking in the first place, and get it down further.
Pushing the envelope from several direction can really help (just don't knock it off the table 8^)
"It's tough to be bilingual when you get hit in the head."
You mean like the next generation of Athlon motherboards that will feature DDR-SDRAM at 133(2) Mhz memory for a bandwidth of 2.1 GB/s. The EV6 bus that we all love running at 133(2) Mhz and 200(2) for server boards and AGP 4x/PRO? Oh, don't forget UltraWide SCSI 3 and ATA100 is also just around the corner. HDD manufacturers are also in the final stages of design and testing of 10,000 rpm IDE drives and before long 15,000 rpm SCSI drives will be commonplace. But you're right, faster processors are pretty useless. When the performance difference between an Athlon 650 and Athlon 700 is 7-8% realworld and 9.3% hypothetical, that 1-2 percent is just too much to bare. Optical Storage is vaporware, UltraWide SCSI 3 is not.
cheese logs keep my wang warm at night.
MSK
I can easily imagine the value of ever-increasing processor speeds. If 500MHZ was enough, why would people ever buy computers with 2, 4, or 8 processors? The big huge database server is never going to be too fast. Any web server doing a lot of dynamic content is never going to be too fast.
Even at home, I've written a program to do some simple stock market analysis. Since I get info on more than 12,000 stocks, and I have data going back 10 years, it takes hours to compute. It used to take 12 hours for it to work on a 486 I used way back when. Now, I have an AMD Athlon 600, and it takes 4 hours, and I've added a whole lot more computation in to the program since because I can.
But, I do agree about one aspect - games. I used to play Doom. I played Quake 1 a little, too. But not anymore. Why? Because who needs the headache of keeping up with Video cards, memory, and processors just to play games? I've always preferred non-realtime strategy games anyway (although there are no great ones out there), so I just decided no more games. It's not even close to worth it.
First, make it work, then make it right, then make it fast, then, make it bloated!
Ah, this is great. Chip manufacturers are increasing power consumption in exchange for no notcible benefit. Could we please get back on track and work on something that computer users really want, like:
* Dirt cheap price (i.e. a fast CPU for $20).
* Negligible power consumption.
* Much smaller, less clunky systems (of which a low power, cheap CPU is an important part).
* Zero boot time.
There are companies working on such things, but other than Transmeta you never hear about them. iTV has been working on a CPU that costs $2, with the goal of manufacturing network-aware devices for under $50. And there are others.
It seems to me everytime a CPU / Motherboard speed post is made, the 4+ mod'ed up respondents devolve their conversations into one of two camps, fighting the same old arguement:
1. "Why do people need faster computers? I just dropped in my shiny new _____(number) mhz processor which cost me $_____ (number), and I can't tell the difference when running _____ (noun). Sure I'll have lots of extra cpu cycles to donate to _____ (noun), but does that _____ (verb) me?
2. "Many people need faster computers. ______ (pural noun) need them to render gibs at a minimum of ___ (number) fps. _______ (plural noun) need them to handle ____ (number) hits a second, so they can survive _____ (number) simaltaneous DoS attacks! And don't forget about the ______ (plural noun), what would we do without quintupple anti-aliasing of Buzz Lightyear's helmet in Toy Story ___(number)!
Okay, so mild exagerations aside, both sides have valid points. The bottom line is that yes, their are many uses for ultra fast processors, and yes, John Zaibatsu won't notice the difference when typing up word documents, something more revolutionary than a higher clock speed and higher transistor count will have to come into play to get things rolling once again.
Why do we dally around with 50 mhz clock uppings? Because its a proven means of creating wealth for the high tech companies -- John Zaibatsu will go get that new processor regardless of whether he really needs it. Companies/research outside of the already established tech cash flow need to cause another revolution.
Bring out my microprocessors measured in atoms instead of nanometers. I want a distributed system for myself, with a processor count that can't be measured with one digit of binary. Revolutionary system design -- don't bother with getting rid of the arcane scroll lock key or ISA slot, build a new system of input! Moore's law isn't a law, only an observation. Breaking it isn't impossible.
Okay, got the futurist ramble out of the system.
(Enter flames about how I've been reading too much Science Fiction and why we need a scroll lock key:)
The scroll sucks A..um..butt when scrolling through snything with alot of drop downs. In essense yo still need to move the pointer away from the fields (towards the edge), defeating the purpose of the wheel. I have had that bite me on many online forms where I fill it out and submit to see suddnley that my age range is 75-90 instead of 20-25 and my income is 100-150k instead of the pathetica amount it is and usually I live in come country like Zaire or Zimbabwe instead of the lousy old USofA. Fun huh?
www.mp3.com/Undocumented
The only reliable motherboard I've heard for Athlon chips is the recently released Asus line of Athlon MBs.
A friend of mine even had core dumps in Linux from his Athlon 800 MHz, until he got the Asus MBoard.
My decision, and I feel it was a wise one, was t oget a P3667 MHz Cu and Slot 1 board. It has reliability, it's cheaper than the very top end, and I know it can do everything I need it to do, in a production environment and in a gaming environment.
I actually don't plan on upgrading for a while, though my computer supports up to 1.2 GHz.
I'm not anti-Athlon, I'm anti-risk (having had a bad experience with my K6-300 having floating point problems). I know my Intel chip and Asus board won't cause me any worries, and I'm willing to pay a little more for that reassurance.
It seems to me that unlike other big corporate competitions (example: supermarket price wars), the speed war between Intel and AMD runs the risk of being bad news for consumers.
:)] - reduced reliability, it's probably not worth it.
:)
As the two companies rush to get newer and faster models of their processors to market before the other, they run the risk of sloppiness.
It's all well and good if AMD can release 900MHz+ Athlons and they work properlly, just as much as it's fresh 'n' funky if Intel can achieve the same with the PIII - there'll always be someone out there who'll buy them, be it to get that extra little kick in Q3A, or to process those SQL statements that little bit faster. However, if the price we have to pay for the extra speed is more than a few extra dollars [or pounds
On the other hand, if the machines these processors are going to go in are gonna run Windoze, they'll probably be beaten to crashing by the OS anyway
--
Listening for the sound of the coming rain...
Since AMD came out with the 750 Athlon, all the faster chips ship with cache divisors no higher than 2/5. I suspect that the 1GHz processors will ship with 1/3 cache divisors until they integrate the L2 cache onto the die with the Thunderbird processor.
This is the one area where Intel seems to be outpacing AMD, the CuMine cache isn't crippled at higher clock speeds. From my own experience, my Athlon benchmarks higher at 750 with 1/2 cache than at 823 with 2/5 cache.
Although I use AMD at home, when I went to go find a Dual-Althon board I was surprised I could not find any. Do any of the AMD chips support SMP? If not AMD is going to have a hard time with the really high end business market.
I've been doing this computer thing for a shorter time, but I think I've still got a handle on this computing power thing. Right now, my computer is streaming music over the Internet. At the same time, I'm writing you this message. In the background I've got mathematica open because I'm a grad student and spent part of the night doing math. I've also got an open document that looks better than what most professional publishers could do a decade and a half ago. E-bay just told me I had won an auction. I IMed with some friends in CA (I'm in NY) as they were going to sleep. Yesterday I sent a friend a short movie, a joke movie, but a movie none-the-less. I live in the middle of nowhere, Ithaca NY, and have ordered off of E-bay. I don't use my CD player any more and have 200 CDs in a big binder gathering dust. I've played games with friends 3500 miles away. I've video conferenced with my brother. And I have a talking paperclip on my screen that yells at me when I don't it ctrl-s to save. (I kill him when he does that) Anyway, what I'm trying to say is that while incremental improvements in chip speeds don't have much of an impact. My computer system (probably worth about 2.5-3k) has changed the way I live my life. And the speed increases are part of it. I realize that a 1 GHz chip on a 133MHz bus has issues, but at the same time, I'm still wanting a faster computer. Maybe I'll be able to put a 3-D model of myself online. Maybe speech recognization will finally work. Maybe my design rule checklist for MEMS design will run in less than 3 hours. There's plenty of ways for a computer's speed to be used, and we're not even close yet. Back in the mid 80's (before my modem) I used my computer for games and word processing. Now it is central to my life.
On a side note, in a Tom's hardware article that mentioned how Intel has higher margins it has to meet, thus they sell their chips for more. In order for Intel to not die they need to not lose their dominance they will have to fix some problems and then take lower margins. This will most likely result in a drop in Intel stock and I don't know if Intel execs are willing to do this. Before Intel sees the light and changes their ways, AMD may have stolen the show. I'm not saying Intel will go out of business, for I think that would probably be impossible, but they will have to change the way they do business. The question is will they change soon enough to stay the market leader.
Molog
So Linus, what are we doing tonight?
So Linus, what are we going to do tonight?
The same thing we do every night Tux. Try to take over the world!
The highest frequency of any digital system is determined, at the end, by how fast the transistors can switch on and off. A really cool thing will happen when asynchronous processors will come out. Asynchronous devices don't have clocked speeds, so the only limitation is the speed of the logic gates through which the signal must go. The problem with asynchronous devices is that they're a bitch to design, because you have race conditions to worry about and you don't know if you can accept the input and such ( since there is no clock cycle to tell you that )
-----
Athlons have moved the goalposts for processors slightly, but what we need is reliable fast memory, Fast interfaces and more development in storage devices (like the next stage of HDD's as they've pretty much not changed for longer than x86 chipsets... apart from more plates, higher density and faster spinning, where's decent optical storage as replacement, etc.)
Most of this seems to be intercompany bravado and marketing without much in the way of gestalt system development.
I for one would look more favourable on the company that doesn't push clock speeds but goes for partnerships with other technologies for holistic solutions.
Working for the (other) man
What, me worry?
(I needed this special isa card since its the only supported digital audio (spdif) card for linux.)
I'll try the tyan kx133 board next - hope I have better luck since I sure love the k7 chip, itself.
--
--
"It is now safe to switch off your computer."
Intel's mindshare with the public is slowly but surely eroding and the x86 CPU war is turning into a fairer fight. Me, personally, I'm just happy I can get a 700 MHz CPU for under $300 now.
- A.P.
--
"One World, one Web, one Program" - Microsoft promotional ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
So you need a new motherboard.
I for one like the competition to Intel's domination of the high-end processor market. In a relatively short space of time (since the Athlon release) wehave seen "fast" processor's come out at a rapid rate and at a quickly reducing price.
And this leads me to a couple of questions...
1) Assuming AMD can maintain their product releases to leapfrog Intel, I am wondering how long Moore's Law will last.
2) Intel keep showcasing exceptionally fast processor's dipped in oil and outside of a working case - so will these start coming to market quicker than usual.
What, me worry?
AMD is going to build ``normal'' x86 chips with special 64 bit instructions, while Intel is trying to push IA64... which hasn't been released in a commercial version, and which many people think won't be popular until 2003 or so. AMD has a brand new Athlon core to work with, but Intel is doing all they can to s t r e c h the PPro/PII/PIII core to faster speeds by increasing the length of the pipeline everywhere they can. Will Intel make a re-designed new core (I'm a big fan of calling the chip the Sexium if they do...)? Do they have to to compete with AMD?
And then there's the issue of other people... will we finally see a board capable of running new PowerPC chips that doesn't come from Apple? Will the portability of Linux allow other chip makers to enter the playing field? Maybe the Digi... er, Compaq Alpha people will get someone to fab their ``old, slow'' 21164s for really cheap.
Anyway, where is technology heading? Certainly low-power-consumption chips are a booming industry, microcontrollers are everywhere, but for the desktop, what will the landscape look like in 4 years?
-- Erich
Slashdot reader since 1997
How many applications these days are really processor bound? Sure this'll speed up your SETI rate, but is that a good reason to buy a $500 processor? I just went from a 300 Mhz to a 466 Mhz chip. You'd think that would be a noticeable difference in the everyday user experience. Nope. Feels just about the same -- even on Win 2K. Now what would a 1 Ghz do for me?
I know there are some people who can appreciate this. But it's getting to the point where most people have no reason to upgrade.
Their take is basically that Intel can't handle how to deal with competition, and is kinda panicing.
The Register is also suggesting that AMD might start fabbing Alpha's.
It's been pretty much established at this point that $800 is roughly the maximum price that the market will bear for the top-of-the-line processor from a given manufacturer in the general computing market. This has the effect of driving down prices on the rest of the processor speeds in the family.
A 1GHz CPU, by itself, is nearly useless for most purposes given the massive I/O bottlenecks we have to deal with. The fastest system buses out there are the CuMine 133 MHz FSB, and the Athlon/Alpha 200 MHz bus (that's really a 100 MHz bus). Memory is a huge constraining factor at these speeds, too, and the ATA-66 drives can just barely keep pace with those needs. When the FSB runs at 400 MHz, with RAM speeds to match, and everyone uses 66 MHz, 64-bit PCI and 4x AGP, then the rest of the system can keep pace. What a 1 GHz processor does in today's platforms is basically run benchmarks faster. There's not going to be a huge impact. To really get into equipment that supports that kind of I/O, you're talking about mainframes and super high-end workstations, not PC's anymore. And the costs rise accordingly.
The impact will be in the downward price pressure across the board - 600 MHz processors will be under $200 soon. Being able to get a lot of bang for a lot less buck is more compelling, I think.
And for the Beowulf trolls out there - for the money it'll cost for a single 1 GHz processor-based PC you'll be able to run a pair of 600's in a cluster, or buy a SMP board with a couple of 650 MHz CuMine's. Now that's cool!
- -Josh Turiel
-- Josh Turiel
"2. Do not eat iPod Shuffle."
IBM is pretty sure 4Ghz is doable in the next two years by doing multiple independent clocks with async buffering between them. Slashdot ran something about that within the last month or so.
I don't know how fast CPUs can really get, but clock speed isn't the only factor. Work done per cycle is also important. If you look at today's (desktop/workstation) CPUs they all can execute more then one instruction per cycle. Two is a minimum. Three is the sustained max for Intel/AMD. Some RISC CPUs can do 5 instructions (UltraSPARC) if they are just the right set. This rarely happens, so most of hte time your two/three/five banger is executing zero (waiting for main memory), or one (too many read-after-write hazards) instruction. Transmeta is trying to fix this by making the (emulated) CPU really smart. The Alpha is trying to fix this by executing instrctions from more then one process at the same time (i.e. an ADD and MUL from your MP3 player, a LOAD and FMUL for your Quake server, and a STORE and a CMP for the re-compile of your Linux kernel). Compaq's simulations say this is really the shit. In two years I guess we will be able to see if it is. (the idea isn't new, the Terra supercomputer does something similar, and barrel processors have been doing something similar for decades).
As for 1Ghz being unthinkable four years ago, I assure you the design cycle for CPUs is long enough that the team that built the Athalon was almost certonally thinking of it. About eight years ago the Alpha archature manual was talking about it. Seven years I would have taken a bet that the Alpha would be the first "desktop" CPU to hit it. Apparently later this month I can be thankful nobody took me up on it. :-)
I have a 200Mhz machine at work, and a 266Mhz at work. Both do nicely for a lot of tasks. I bought a new machine to ray trace on (you could give me a million 1Ghz machines and I could still use more CPU to raytrace on...well maybe not me, but a real artest I guess). Neither do well for modern games, which seem to be a big driver for fast desktop CPUs. My C++ compiler could use a faster CPU too, which is why I don't do compiles on my 200Mhz box at work.
A fast desktop machine doesn't have the I/O of an oldish supercomputer. It does have a faster CPU. It's CPU is even competitave with modern supercomputers (sometimes it shares the CPU!). It's memory system and I/O is staggeringly far behind. Then again so is it's microscopic price tag.
So a fast desktop is good for anything that can get away with less memory, slower memory, and slower I/O, and just pain can't afford a supercomputer.
Since supercomputers are staggeringly expensave, I would guess lots of stuff "makes do" with desktop CPUs. Off the top of my head I would say: