AMD Cuts X2 Processor Prices

BDPrime writes "AMD is cutting prices for its X2 processors, according to an update on its microprocessor pricing list. The cuts refer to AMD's Athlon 64 FX and Athlon 64 X2 chips. Some of the price cuts are almost in half."

Comments

socket 939 seems to be screwed all-round

[Anonymous Coward]

Most of the remaining chips on AMD's price list use the Socket AM2 or Socket F form factor, rather than the older Socket 939 interfaces.

I just bought a 4200+ x2 for $159 from newegg. They sold out hours later. I don't think they even make 'em any more. Anything higher than a 4200 was plain sold out everywhere.

So if you've got a socket 939, I'd say you better upgrade with a quickness cuz those CPUs are going, going, gone.

Re:socket 939 seems to be screwed all-round

[alisson]

Indeed. My wife has a 939, but coincidentally, we're broke :D

Re:socket 939 seems to be screwed all-round

[Rudeboy777]

They haven't been produced in some time. I would love to get a dual-core 939, but they aren't in stock in any store in Canada AFAICT. OTOH the power consumption of any 939 X2 CPU is much higher than the newer AM2 chips.

Re:socket 939 seems to be screwed all-round

[billcopc]

The reason no one has any stock is because OEM Express / OEM Depot bought them all :) Give them a look if they're in your town, as they don't sell online.

Re:socket 939 seems to be screwed all-round

[Darby]

I would love to get a dual-core 939, but they aren't in stock in any store in Canada AFAICT.

Not sure if it's any help, but 939 Opterons work great for that as well... better than most of the X2s since all Opterons have 1MB cache per core and only some X2s do. Back a year and a half ago, the Opterons were even cheaper than the X2s as well. If you can find it for a good price, it's a great chip.

Re:socket 939 seems to be screwed all-round

[linguizic]

Man, this is a bad place to be making claims about your wife's socket.

Re:socket 939 seems to be screwed all-round

[MoxFulder]

I upgraded my work computer with an X2 3600+, ECS mobo with ATI radeon graphics, and 1gb DDR2 RAM for $170 a couple of weeks ago. Now I'd like to do the same at home, but I'd like to keep my current mobo and RAM.

But as far as I can find, none of the Socket 939 X2's support hardware virtualization, which is a biggie for me as I do a lot of cross-platform development.

Are there *any* socket 939 X2's that support hardware virt?

Re:socket 939 seems to be screwed all-round

[Mad Merlin]

Are there *any* socket 939 X2's that support hardware virt?

No, unfortunately not.

Re:socket 939 seems to be screwed all-round

[Anonymous Coward]

www.mysterybyte.com in Nova Scotia

Re:socket 939 seems to be screwed all-round

[MoxFulder]

Darn it! That's what I thought :-(

According to the wikipedia page on the X2, the E4 and E6 steppings were for Socket 939 and didn't support AMD-V. While the F2, F3, and G1 steppings support it, but are socket AM2 only.

It doesn't seem like there's any deep reason why AMD couldn't offer the latter steppings in a Socket 939 form... but I guess they've moved on and now concentrate on the Socket AM2 exclusively.

Re:socket 939 seems to be screwed all-round

[Anonymous Coward]

The reviews don't point out that Intel E4000 CPUs don't do hardware virtualization either.

Re:Buy one get one free!

[renegadesx]

I get it now!!!! The lawsuit between Intel and AMD is just sleazy advertising for AMD!! Why didn't I see it sooner!! Thank you Richard Dawkings!!!

To tell you the truth...

[ZDRuX]

I`m quite surprised by how long it took for this to come into affect after the Core Duo's were released. It seems like they gained immediate acceptance (Core2 Duo) and were getting great reviews from just about anybody, you'd think AMD would drop prices earlier on, I didn't think it would take this long.. I don't think this it was a question of IF... but WHEN.

Re:To tell you the truth...

[ZachPruckowski]

There was a summer price cut around the time of the C2D launch. These are timed to deal with the launch of the ultra-low-end C2Ds, and the price-cuts expected as the new Core 2 Quadros push C2D prices down. AMD's got nothing new until Q3's Barcelona, so they're fighting better chips with cheaper chips.

Re:To tell you the truth...

[Fulcrum of Evil]

Itanic was a completely separate architecture. This is x86 with some extra stuff, more or less.

Re:To tell you the truth...

[shawnce]

AMD's chips are proven, Intel's new offerrings may be technically interesting, but they are unproven

Huh? The Core 2 is based on technology used in the Pentium M which itself was based on technology from the original Pentium families (not P4). The Core 2 family in some ways has a more developed history then the Pentium 4 (NetBurst) family did. Also unlike the Itanium the Core 2 uses x86-32/64 ISA... just like current AMDs processors.

Anyway the Core 2 Duo (and Core Duo before it) have been out for a while now (around a year) and are used in a huge number of consumer, prosumer and workstation systems from many vendors. You think that would prove something...

Re:To tell you the truth...

[ZachPruckowski]

C2D is based off the P3/M architecture. It's not a dramatic, risky step like Itanium. The issue with Itanium was porting, but C2D follows pre-existing ISAs and ISA extensions like x86-64 and SSE1/2/3. There's nothing unproven about C2D at all.

Re:To tell you the truth...

[Technician]

I don't think this it was a question of IF... but WHEN.

Just to take a stupid guess... I think with the need for cash, AMD was hoping Microsoft releasing Vista (Biggest upgrade in 7 years) would create high demand for new PC's and they could sell product as demand exceeded supply. The demand for Vista didn't drive demand as expected.

Vista failing to launch put AMD in tight competition in a smaller market due the lack of demand for Vista. AMD didn't sell to Apple. Intel did. Mac's are selling where Vista is getting so-so response so Intel is selling the new chips into markets AMD is not in. If Intel didn't sell to Apple, and had to cut prices, AMD would be in an even worse position due to the low demand for new Vista machines.

http://finance.yahoo.com/q/is?s=AAPL&annual
Note that Apple has gone in revenue from 2004-2006 $8,279,000,000 to $19,315,000,000
Operating income has gone from $326,000,000 to $2,453,000,000. This is almost an order of magnitude growth in only 2 years. This isn't just from a few iPod sales. Vista's dead start and XP's malware flood is driving people away from Microsoft. The recent growth in Apple and Linux is not primarily new PC consumers. It's mostly ex-Microsoft consumers.

I'm wondering if AMD is selling chips at a loss instead of having to throw them out. I can't see them making money at that price, only cutting their loss.

Selling chips at half price is not profitable. I'm assuming most chips have only a 10-30% margin. Chopping the price in half is selling under the cost to manufacture.

Re:To tell you the truth...

[Anonymous Coward]

Don't be surprised, its was completely expected. Heres a link to the Intel price drop http://news.com.com/2100-1001-923046.html and theres a comment in it about the pending AMD price drop in response.

Re:To tell you the truth...

[Kjella]

I think that depends a lot on how you're defining it, most of the cost go into designing a processor, validating it and getting the production process working. Once you've done that, the marginal costs of producing more processors aren't that high. And in the short term, given the plants you already have, I doubt they cost much at all. So in the short term it might be better to get them sold at half price at a *marginal* profit, than not get them sold at all. Whether that's enough to cover the long-term investments and R&D is another matter...

Re:To tell you the truth...

[Technician]

Once you've done that, the marginal costs of producing more processors aren't that high.

Hmmm. Are you sure. I thought there was quite a bit of expense to produce high speed processors due to the costs associated in production. Not only is R&D costs high, but so is the technology for Lithography, Clean Rooms, Metrology (can't make and sell a product with 100% failure rates), packaging (Silicon in carrier, effective heat spreader, etc. Not the cardboard box), and of course Yeild. The more steps it takes to make a chip, the more steps that can kill some of the product. Adding more and more to a chip requires more layers to connect it all together. Making it smaller makes dust that would pit a line, now completely obliterate it. Optics required for lithography for features smaller than visable light wavelengths raise the cost due to the precision required for registration, focus, etc.

If you were making a sound chip for a greeting card is a cheap process. Making a modern high speed heat and power effecient CPU is not a cheap process. If it were, then there would be many more manufactures working to rake in the money. Money is made by having the capacity at high yeild and a high performance product to cover the R&D and manufacturing costs. If your product doesn't yeild, you have less sellable product for your capitol investments and materials & manpower costs. If your product doesn't perform against the competition, then making a profit isn't easy.

Making current generation microporcessors is a very risky business, unless you have a clear roadmap into the future and can stay on track to beat the competition. Getting off track is very expensive. Microsoft having a poor showing of Vista isn't helping pay the bills in the Microprocessor market for everyone at the moment. The Apple Intel deal has helped soften the poor Vista launch for Intel. AMD doesn't have that safety net.

Re:To tell you the truth...

[Anonymous Coward]

I take it further. It's not a question of if or when, but the question of stupidity. If the level of stupidity at AMD is not so high they would have reduced the price instantly. The Intel core 2 duo chips smoke the AMD's. I look at the price and performance comparison before and I saw AMD was almost competitive. I would go with Intel in a heart beat with the price almost as low as Intel for a same performance.

Re:To tell you the truth...

[drinkypoo]

Itanic was a completely separate architecture. This is x86 with some extra stuff, more or less.

Uh no. Even the K6 was a fully-RISC internal architecture, effectively with an x86 ISA emulator strapped on the front end of it. The K6 was a fantastic processor but not very good at being an 80386.

K8, similarly, is internally both much more and less complex than you can see with the x86 stuff. x86 processing uses a subset of the core's functionality, it's true, but that's not quite the same thing as having x86 with addons. They're emulating x86 again. And this time they're doing it on a 64 bit architecture (probably by discarding a lot of bits, and setting the carry flag from the 32nd instead of the 64th bit, and so on, although there must be plenty of other magic in there.)

This is actually true of all of these various processors these days; they're mostly or fully RISC and there's an emulator in there to handle x86.

Re:To tell you the truth...

[drinkypoo]

As the process shrinks, R&D seems to go up, but cost per part goes down, because flaws affect less chips on the wafer because the individual dies cut out are each smaller.

Re:To tell you the truth...

[Fulcrum of Evil]

You're missing the whole point: itanic is not RISC, it's VLIW, with a whole bunch of differences, not the least of which is the compiler level parallelism. Core duo is just x86 with some extra stuff - the previous processors are almost exactly the same.

Damn I just bought one!

[snuf23]

I just picked up a nice Athlon X2 system for my son. Total cost with shipping was $700. 1GB RAM, 160GB HD, DVD drive, 7600GT video card. For an extra bonus although I ordered a 3800+ X2, the system came with a 4200+ installed.

I have a Core2Duo system myself but currently the AMDs are a great value.

Re:Damn I just bought one!

[Seumas]

Check wherever you bought it from and see if they have one of those "if you find it cheaper in 30 days, we'll refund the difference (especially if it's found cheaper at our same store)" things. A lot of places have them even if they don't advertise it, so it's worth asking.

Re:Damn I just bought one!

[Anonymous Coward]

currently the AMDs are a great value.

A great value what?

Re:Damn I just bought one!

[Cerberus7]

I just sent off a message to Newegg. I hope they do this for me, as my 6000+ nearly halved in price, and I just ordered a week ago!

Re:Damn I just bought one!

[bcrowell]

I just bought one, too: AMD Athlon 64 X2 4200+. I can't say I'm too upset about missing a price drop. These things happen.

What's really cool about these chips, IMO, is that you can get the power-efficient ones that only draw 65 W max for the CPU. My new system only draws 133 W (monitor included) with both CPUs running full blast, and when they're idle and the monitor is powered down, it goes down to only 51 W! These chips have AMD's cool'n'quiet, which is fully supported in recent linux kernels.

This was my first time running a dual-core system. So far I really haven't seen any improvement in performance from the dual core. The sad truth is that most of the time when I end up waiting for my computer to do something, it's either (a) doing I/O, (b) doing something CPU-intensive that's not parallelizable, or (c) it's limited by the speed of the memory, not the speed of the CPU.

Re:Damn I just bought one!

[gmack]

I have a 4200+ as well and most of the time it's in it's lowest possible frequency setting. (you can check in /proc/cpuinfo) Playing videos won't tax the system but some games will and so will compiling for extended periods of time. I'd say for 95% of my computer's life it's throttled down.

Re:Damn I just bought one!

[AmaranthineNight]

My 5200+ just went down in price and I ordered it yesterday. If I could go back in time I'd wait till today and order a 6000+
Maybe I'll return it once it arrives (tomorrow or wednesday) and pick up a 6000+ since it'll still be in the box. That's a really substantial price cut, and I'd love to see that increase in performance for about $60...
I'm off to kick myself for not looking into price cuts >_>

Re:Damn I just bought one!

[GnuAge]

Not too shabby.
Let's see what it would cost for me to build a system like that and throw on Linux instead:
Retail X2 3800+ CPU & ECS GeForce6100SM-M motherboard (onboard GF6100): $90.
2 GB Kingston PC5300 DDR2 RAM (after $40 MIR): $100.
Maxtor 160 GB Serial ATA/300 hard drive: $40.
Geforce 7600 GT video card with 256 MB RAM: $110.
Samsung 18X DVD-RW: $30.
Apevia X-QPack-NW-AL MicroATX case with 420W power supply (after shipping & $10 MIR): $86.
Total: $556 plus tax, $607 total after $50 MIRs.

I guess the extra $93 is the Microsoft tax. Of course the system I describe comes with twice as much RAM and good quality onboard graphics for those wanting Xinerama Beryl dual-monitor goodness, plus a slick little case, probably much smaller and handier than what you ended up with. Since I'm not a gamer I'd save a little money and stick with the onboard graphics or maybe add a Geforce 6100 for $40 instead if I wanted dual monitors.

Waiting for the corresponding cut on Core2 Duo

[dilute]

Why bother with X2 when you know this will force a price cut on C2D?

I must presume, though, that this cut is just to clean up on the bottm end and make way for a new high-end line.

Re:Waiting for the corresponding cut on Core2 Duo

[rsmith-mac]

The price cut is in a little less than 2 weeks(April 22nd). Shamelessly ripped from AnandTech:

Intel Core 2 Extreme QX6800    2.93GHz x 2     4MB x 2     $1199
Intel Core 2 Extreme QX6700    2.66GHz x 2     4MB x 2     $999
Intel Core 2 Extreme X6800     2.93GHz         4MB         $999
Intel Core 2 Quad Q6600        2.40GHz x 2     4MB x 2     $530
Intel Core 2 Duo E6700         2.66GHz         4MB         $316
Intel Core 2 Duo E6600         2.40GHz         4MB         $224
Intel Core 2 Duo E6400         2.13GHz         2MB         $183
Intel Core 2 Duo E6300         1.86GHz         2MB         $163
Intel Core 2 Duo E4300         1.80GHz         2MB         $113

Re:Waiting for the corresponding cut on Core2 Duo

[straponego]

I think it depends on whether you're upgrading, and how fast you need to go (or how much you're willing to spend). If you're building from scratch, C2D is probably a better deal these days, especially at the high end. But at the low end, last I looked, it was much more competitive. And, lessee... okay, from Newegg six months ago, I got a socket 939 motherboard and AMD64 X2 4200 for $270. If I'd gone Intel, I'd have had to get DDR2 memory, which would have basically doubled the price for similar performance. As it was, I could re-use the memory I had.

So, depending on your situation, AMD is not in as grim a position as it looks... but they are hurting, and need to do something soon. Still, they were ahead (in absolute performance) for about three years, and have been behind for about one. It seems like they're in general ahead on innovation (cool new features like 64 bit, Hypertransport, multicore, asymmetric cores, tend to come out of AMD's labs), while Intel is well ahead on process.

Intel has to be careful, though. If they do manage to kill AMD, a few things will happen. First, they're susceptible to anti-trust suits. Second, IBM might pick up AMD and close the process gap. Third... well, where will Intel look for new ideas?

Re:Waiting for the corresponding cut on Core2 Duo

[Minwee]

That's no surprise. We have known for some time that Intel is planning to release a few new processors and slash CPU prices dramatically in about two weeks. AMD won't have anything new to show off until later this summer so all they can do is cut their own prices farther and sooner just to keep up.

Re:Waiting for the corresponding cut on Core2 Duo

[dreddnott]

Ouch! I just paid $308 for an E6600 from Newegg. Oh well. $224 would put it at the same price as the 3GHz Athlon 64 X2...

Re:Waiting for the corresponding cut on Core2 Duo

[Anonymous Coward]

I don't know what those figures are but I assume they are the raw CPU prices. In that case the E6700 will be more than $316. It will most definitely fall into the "best bang for the buck" that the E6600 is currently in but it will certainly be more expensive than what you paid for your E6600 for some time.

Re:Waiting for the corresponding cut on Core2 Duo

[Rudeboy777]

If I'd gone Intel, I'd have had to get DDR2 memory, which would have basically doubled the price for similar performance. As it was, I could re-use the memory I had.

There are motherboards that allow you to use DDR RAM with a Core 2 Duo CPU (example: http://www.asrock.com/mb/overview.asp?Model=775Dua l-VSTA&s=n ), I believe Anandtech or Toms Hardware had some benchmarks that showed the performance difference from using DDR RAM in one of these boards was negligable.

Re:Waiting for the corresponding cut on Core2 Duo

[Malc]

I wonder if that's going to affect mobile prices. I'm due to get a new Dell laptop this quarter - will be nice to see the T7600 cpu a little cheaper.

Re:Waiting for the corresponding cut on Core2 Duo

[Nozsd]

Not only that, but the entire 6000 series will be upgraded to 4MBs of cache.

Also shamelessly ripped from AnanadTech: Currently the E6300 and E6400 both have 2MB L2s, but both chips will be replaced by 4MB versions - the E6320 and E6420 respectively.

The best part is that they won't cost any different than the 2MB versions.

Re:Waiting for the corresponding cut on Core2 Duo

[straponego]

Interesting, thanks!

OT RAM prices

[lawpoop]

Mod me offtopic, but I need answers!

When the hell are memory prices going to come down!? A gig of RAM is still like $80 at the *minimum*. What is going on? Someone please explain this to me. This is like the single most expensive component of the computer. I'm looking to put together a new system for video editing, but either MAC or PC, memory prices are outrageous. If I'm forced to use Vista, I don't want to shell out $300 to upgrade to 4 gigs of ram to get a decent GUI experience.

I remember reading about price-fixing cases a few years back, have those gone anywhere? AFAICT, there is still some collusion going on.

Re:OT RAM prices

[CompMD]

If you're complaining about ONLY $80 for 1GB of RAM, you have no concept of what non-shitty memory for real workstations can cost.

Re:OT RAM prices

[lawpoop]

OK, I'll bite; how much?

Re:OT RAM prices

[TheRealMindChild]

Well, I can't give you a concrete answer, but I can put it into perspective a little better. See, here in the US, gas went up due to the war a LITTLE, but not to the extent that Hurricane Katrina caused. Gas prices skyrocketed because supply was short. Though as production increased, the prices of the gas didn't decrease in ratio. So now we are at productivity, gas is still much higher than it would have been without the disaster. See, they couldn't gouge directly, but they had reason to increase the prices, but no inclination to decrease them.

Re:OT RAM prices

[EvanED]

If I'm forced to use Vista, I don't want to shell out $300 to upgrade to 4 gigs of ram to get a decent GUI experience.

You don't need to... that article was full of crap. Vista has issues, including performance and memory use issues, but it's not nearly THAT bad.

As for your question, I have no clue. I bought a gig (one stick) a year and a half ago for about $110, and it's only recently that prices have come below that. It seems that prices shot up I guess not long after that and have been declining since. I would like to know as well.

Re:OT RAM prices

[tomhudson]

If you're complaining about ONLY $80 for 1GB of RAM, you have no concept of what non-shitty memory for real workstations can cost."

Hey, I paid more than that for 64k of ram (not 640k .. 64k). $100.00 on sale. A gig of ram for $80 - stop complaining, its cheaper than staples, thumbtacks, plant seeds, etc.

And you should have seen the cost of the abacus before that!!!

Re:OT RAM prices

[sumdumass]

If he is doing video editing like he suggested, it would be that bad. Of course this would be mostly because of the video files and editing on top of what Vista has issues with.

But on a side note, I don't understand why he is complaining about the cost of a special needs system when he is the one dictating the special needs. And $80 per gig isn't likely going to be the type of memory to excel in this area. I have dabbled with video editing in the past and found the rock bottom priced memory to have issues in some programs were they weren't ever noticed in others. After hours upon hours of trouble shooting and pulling my hair out, I have actually bought matched sticks of corsair memory to replace problematic generic el-cheapo memory and then used that cheap stuff in other systems without problems.

I'm wondering if he is going to be satisfied with any $80 per gig memory or any cheaper modules when they become available.

Re:OT RAM prices

[Lumpy]

$164.00 per 1 gig DDR2 ECC workstation level ram.

and that is for the cheap stuff.

Re:OT RAM prices

[rm999]

My guess:
RAM prices come down in steps. RAM gets cheaper because new fab technologies come out, but upgrading to these new factories costs a lot of money. This loss is recouped by selling the current technology at a higher price, while discounting the old technology to have half a chance in hell of selling enough to maintain a profit. This is the way CPU prices work (see current article). While prices come down slowly (I purchased a gig of RAM for 100 dollars last year, which is almost 50% more than current prices), it is not noticeable until one of these new technologies come out. I usually time my computer upgrades to coincide with these events.

I could be totally wrong. Sometimes catastrophic events can actually increase RAM prices, such as the earthquake in taiwan a few years back that drastically cut production of RAM.

Re:OT RAM prices

[MoOsEb0y]

I'll bite. 1 GB PC 5300 is $54.99 shipped.

Re:OT RAM prices

[Lumpy]

2gig is more than enough to edit video under XP.
After ordering hardware for a Vista Editor, and using it for 2 days we erased it and downgraded to XP. the machine is at LEAST 3X faster without Vista taking up all the resources. (64bit is useless in video editing right now, most editing apps are NOT optimized for it yet)

Only a complete fool would use Vista on a video editor, stay with XP until the last possible moment, you get more done faster.

Re:OT RAM prices

[Awod]

I use rambus it came with the comp 4 years ago.. if I wanted to upgrade to 512 it would cost roughly $600 [obviously no shot in hell of that happening..]

Re:OT RAM prices

[Anonymous Coward]

For a long time, you could run XP with only half a gig. It got to the point where even the cheap-o computers had half a gig, and a gig was only a few bucks more (relatively speaking).

But, as has happened every few years when a bleeding edge (in system requirements, not performance) version of Windows comes out, everyone goes out to upgrade and the new machines are requiring more resources. Now, PCs are being sold with the minimum required 1g, with 2g upgrades and so on. Essentially, where RAM used to be cheap because you didn't need much, now you need twice as much just to get going and more to really perform => higher cost of RAM.

So yeah, after a long period where prices went down because computers only needed 1G for about 5 years straight, we're seeing the other end. Once the market catches up and demand slows again, prices will drop, but I wouldn't expect it anytime soon.

Plus, I hear prices drop when Dell or HP orders huge amounts from a supplier.

Re:OT RAM prices

[Anonymous Coward]

I'd be interested in finding a good motherboard for a Core 2 Duo supporting > 4 GB of ECC... For a server. That said, I've got 4 GB of non-ECC "shitty memory" as you call it on my main development machine, which I'd pretty much call a workstation (Core 2 Duo 6600 / 4 GB or Ram... This thing smokes a great many "professional workstation" from not long ago). And I've got a 6 months uptime (Debian 'etch' non-final) on this baby. Which beggs a question: in this day and age of VCS (I'm using SVN usually), SSH, checksums everywhere, etc. how the fc*k would 4 GB of ECC help me vs 4 GB of non-ECC? I mean, with my "shitty memory" I get 6 months of uptime and have a hard time remembering an application crashing (OK, Firefox sometimes crashes... But this ain't due to the mem ;)

I only buy Kingston-lifetime-warranty RAM. But they are non-ECC. Are you just an elitist troll?

Re:OT RAM prices

[Merc248]

I just bought a set of G.SKILL 2x 1 GB PC-6400 modules for $130 from Newegg.

http://www.newegg.com/Product/Product.asp?Item=N82 E16820231098

Re:OT RAM prices

[Doddman]

When was this purchase made? Because if you did that recently, I would think that that finding a 64k ram card would be something that one would purchase to cherish (antique ram)

Re:OT RAM prices

[Anonymous Coward]

When the hell are memory prices going to come down!?

They have, by several orders of magnitude. 1GB memory modules at PC100 speeds (800MBps throughput) used to run upwards of $1000. Now, RAM that is five times faster sells for a tenth of the price, and people complain that it's expensive.

A gig of RAM is still like $80 at the *minimum*. What is going on? Someone please explain this to me. This is like the single most expensive component of the computer.

DRAM consists of a transistor and a capacitor for each bit. That means 1GB of RAM has a transistor count equal to a high-end modern CPU. It's simpler to produce, but not that much.

Also, if you're doing video editing, you don't need 4GB of RAM. My Final Cut Studio workstation runs fine with 2GB, in DV, DVCAM, and HDV, and I know some that run it on older model PowerBooks with under 1GB, with DV.

Re:OT RAM prices

[lawpoop]

All I'm asking is why RAM prices haven't fallen like they have for MHz in processors or gigs in hard drives.

Re:OT RAM prices

[ameoba]

Memory failure is memory failure, period. The application you're running doesn't change the fact that it's not working, it just accentuates it.

If you can't run memtest86 overnight on new RAM, you should return it. End of story.

Re:OT RAM prices

[Fulcrum of Evil]

If you want expensive, look at the prices on the 2G stick.

Re:OT RAM prices

[sumdumass]

Thats what I thought. But low and behold, this happened. And the sticks passed memtest86 and a couple of other memory testers.

I'm not sure if there wasn't something with the ability of the memory to keep up in the timing when a heavy demand was placed on it or if there was just a bad spot that only got accessed when using the video editing stuff. I mean working with a 2 gig raw video file compared to a couple hundred megs for a game or word processing is a little different in usage. But to tell you the truth, I couldn't tell you were the problem was and I'm guessing to why it was even present. I don't think I'm alone on this either. I have heard in the past of someone getting memory to work were someone else said it was bad.

However, I do suspect it involves how much was in use and how it was used. And I'm not willing to dismiss that something could be wrong with the memory that doesn't show up in lighter usages. With the switch from DOS to windows 95, MS had an article explaining how windows 95 used memory different and bad memory could work fine in DOS 3.1 systems and the problem would be noticeable in windows 95.

Re:OT RAM prices

[sumdumass]

I think they have. Although there have been fines imposed for fixing prices in the past that the memory companies needed to recuperate as well as supplies being limited because of DDR2 and 3 becoming more available.

There are several factors at work here but 5 years ago, 512 megs was going in the $100+ range. Now a full gig is under that in some cases. A key difference between the two might be the amount of competition that is limited by the CPU designs and motherboards that support them when dealing with memory. The memory is more of a reactionary process were they make something then wait for it to be used and the processors generally dictate what types of memory is to be used. You cannot have the same level of competition as you would with processors because of the dependence of usages. On the other hand, Taping into newer processes in production that would make them cheaper seems more reasonable for newer styles of memory when considering the retooling effort needed. Why spend million rebuilding the production system for a third generation technology when you can tool for DDR2 or DDR3 that will have a longer life span.

There are probably more factors involved. I think these are some that seem to pop out and could be likely to influence the price a bit.

Re:OT RAM prices

[tirefire]

And now, so have I.

*high-five*

Re:OT RAM prices

[Propaganda13]

I got new Registered ECC 256MB DDR for $25 about 5 years ago, so I can complain.

Re:OT RAM prices

[CompMD]

Whoa there AC, these ain't PCs I'm talking about. People commonly forget that there's far more to computing than PCs.

What is it that computers do? Compute! Mathematics. I just have to deal with computers that aren't exactly meant for Joe Windows User, Bob Apple User, or AC Linux User, so if that makes me an "elitist troll" so be it. Computational fluid dynamics, finite element modeling with millions upon millions of nodes in a mesh, mechanical system modeling, and more are all computational tasks suited for a different class of computer than most people are familiar with. Sun, SGI, IBM, heck, even some Dells require memory that is easily twice the $80/GB price point that makes you sweat. The components of these machines must be of the best quality, the most robust design, and extremely trustworthy. ECC is icing on the cake. When I submit a compute job that is going to take five days, I want to be damn sure that the results I'm getting are correct. There is no room for a bit error. There have been two occassions in my year and a half tenure at the aircraft design and engineering firm I work for where this has been a factor. A computer detects an uncorrectable error in memory and halts. I submit the job again, if the error occurs in about the same place, I run a diagnostic, verify that the address is roughly the same again, then pull that memory module and replace it. Its not cheap, but how safe would you feel flying in airplanes that have been modeled and analyzed with unverified computation?

In case you aren't aware, most workstations won't even function with bargain basement memory installed. The workstation is supposed to be reliable, and that stuff can't guarantee reliability all the time. I just mention this for workstations, but the same goes for servers and minis. So what if you have checksums everywhere? They don't necessarily tell you where a problem occurred, or even if its in hardware or software. If you don't believe that the better grade memory makes a difference, I have some SparcStations, Iris Indigos, and AS/400s in perfect, original, working condition (that get regular use) that you should check out.

Re:OT RAM prices

[TheLink]

Uh the word you are looking for is upgrade. You upgraded to XP from Vista.

If more people kept to XP, maybe we can make Vista to Microsoft what the Itanium was to Intel.

Now if the Wine etc people would come up with an XP+DirectX compatible ASAP, then when Microsoft tries to pull the plug on Windows XP, lots would switch to "XP" on Linux. Just like when Intel tried to kick people off x86 and on to the Itanium, AMD started raking it in ;).

Re:OT RAM prices

[oojah]

I'm completely with you on the need for better RAM, but you must still see the point the OP is making? RAM prices haven't changed for ages - I've been after a GB stick for one of my machines for over a year and it's been at ~£70= ~US$140 for all that time and I'm not *that* bothered at that price.

The real thing that bites for me, and I'm well aware that it's completely different technology but it's hard not to compare, is that flash ram is so cheap now. I'm looking at a 4GB usb drive for £20=~US$27. It is the cheap stuff, but that's nuts.

Cheers,

Roger

Re:OT RAM prices

[Anonymous Coward]

Sun? What the hell? That's it, I'm never flying again.

Thanks. Thanks a bunch.

Re:OT RAM prices

[TheGratefulNet]

memtest86 and ALSO prime95. one instance of p95 for each cpu core. let THAT cook overnight.

Re:OT RAM prices

[glsunder]

if you're complaining about ONLY $80 for 1GB of RAM, you have no concept of what non-shitty memory for real workstations can cost.

I read that as he's buying non-shitty ECC memory, as in name brand and not generic crap. I think you read it as non-ECC memory is shitty.

Having ECC is good for servers where since you don't want the server to crash due to bad memory, but you want to know asap about memory that's gone bad. You also want to minimize down time since there could be 100s or 1000s of people using that system where as with a desktop, it's generally 1 person. In the last 10 years or so, I've had 1 server have memory go bad. The server started beeping, told me what slot had the bad memory, but never crashed. I've got to say, it was much nicer than having a random crash or having to run memtest86 to find out if it was bad memory.

Re:Buy one get one free!

[romland]

I'll bite.

Reading "Honestly, Slashdot these days..." then following your sig and see: "Francisco Mota is a 16 year old Portuguese college student..."

When was it better?

(Off topic. Got mod points? You CAN pretend that you did not see this one.)

Re:Damn, another weekend at Frys.

[sumdumass]

You should be glad you can deal with Frys and don't have to goto Best Buy and put up with the "Geek squad".

Is it enough?

[eebra82]

I hate to ask this, but which of the top AMD CPU:s are truly competitive now that AMD has cut the prices in half?

At least before the price cuts, there was simply no way I would even consider an AMD CPU, after Intel got Core 2 Duo up and running.

So unless Barcelona changes AMD:s position, what CPU:s do you recommend that actually give us some serious $/performance sightings against Intel?

Re:Is it enough?

[BigBlockMopar]

At least before the price cuts, there was simply no way I would even consider an AMD CPU, after Intel got Core 2 Duo up and running.

Why even consider Intel? They're only relevant for the same reason as Micro$oft - right place, right time, and then gave IBM fellatio. (I especially love the claim that Intel invented the microprocessor. TI was at least two years ealier, but that's a story left to judicious reading of Wikipedia.)

Fat and stupid CISC design. Little Endian. Built a modern architecture on a house of cards (64k page limit). x86 interrupt handling could only be improved by a 14-year-old with a soldering iron and a box of relays. It's as gross and as much of a hatchet-job as Windows: Stupid, ugly, and so mass-produced that we're stuck with it.

AMD could double their prices, and there'd still be some people (like me) who would happily choose the organization which copied an asinine design for commercial purposes... instead of the organization which was responsible for the mess to begin with.

Never mind me, though. Intel might be perfectly right... hell, they might even employ a non-retard or two. It's just that I've been forced to program in assembly for their nightmares - it ain't exactly a TMS9900 or a Motorola 68000.

Re:Is it enough?

[tom8658]

The fact that we're stuck with x86 isn't Intel's fault. At the time, CISC was considered a good idea, and as for endianness.... I won't even go there.

Intel has tried to move away from x86: look at the Itanium and Itanium II. Intel gambled that they could find enough ILP with their compiler, and lost, but at least they moved off of x86, right?

The fact is, because x86 was so wildly successful, and because so much software was written for it, Intel had to ensure that future processors were compatible with the x86 instruction set. Doing otherwise would have been deliberately alienating a large part of their market share. It could be argued that x86 compatibility (or lack thereof, more specifically) is one of the major reasons why IA-64 was unsuccessful. Completely moving off of x86 would be devastating to the company, and irresponsible in the eyes of their shareholders/employees.

I can't believe that any engineer in his right mind would actually want to stick with IA-32 in the face of its glaring defects, they're very bright people, and if you need proof of that, just look at the Core. But if you want to sell consumer chips, you don't have any other choice.

Re:Is it enough?

[Anonymous Coward]

To generalize, the AMD processors will now give you slightly more value for money. But chances are that if you care about processor performance you also like to overclock your processor, which turns the situation over in favor of the Intel processors by a big margin. The AMD processors are already "overclocked" on default settings.

Re:Is it enough?

[fontkick]

I hate to ask this, but which of the top AMD CPU:s are truly competitive now that AMD has cut the prices in half?

None of them. The overclock potential on the C4D is so high, but the C4D's overlock isn't factored into the price.

I have a C4D rated at 2.4 GHz running at 2.9 GHz (an easy, downright effortless overlock if there ever was one) and there is almost no difference in temp - but that much overclock results in a very noticable speed gain. Many people are overclocking the 6600's into the mid-3GHz with decent cooling rigs.

It's not about getting the cheapest you can get away with (although that's normally the smart thing to do). If you are going to spend money on a new CPU or rig - you can spend just a little more and have the fastest desktop processor ever made. It's worth it in the case of the C4D.

Re:Is it enough?

[gmack]

That's not true.. The itanium failed because it's design was so bad that 32 bit x86 processors were outrunning it even when the same app was compiled natively for each architecture being compared.

Re:Is it enough?

[Weirsbaski]

The fact is, because x86 was so wildly successful, and because so much software was written for it, Intel had to ensure that future processors were compatible with the x86 instruction set. Doing otherwise would have been deliberately alienating a large part of their market share. It could be argued that x86 compatibility (or lack thereof, more specifically) is one of the major reasons why IA-64 was unsuccessful. Completely moving off of x86 would be devastating to the company, and irresponsible in the eyes of their shareholders/employees.

I can't believe that any engineer in his right mind would actually want to stick with IA-32 in the face of its glaring defects, they're very bright people, and if you need proof of that, just look at the Core. But if you want to sell consumer chips, you don't have any other choice.

Look at a modern x86 cpu, it actually doesn't have much x86-centric logic in them. They don't execute x86 instructions in x86-handling pipes, they internally convert each x86 instruction into one or more ops that are much more RISKy in nature (3-operand format, access to internal-only temp registers, allow register renaming, more unified ops than x86 provides, etc).

Maybe 10% of the core (and even less taking caches+optional NB into account) needs to know x86, making the cpu more of a "superscaler RISK processor with built-in x86-instruction-translation", than an "x86 cpu".

For a lot of code sequences, the x86 nature of the opcodes isn't the bottleneck, other internal hardware is, which keeps the cpu on a much more level playing field with high-performance parts than a pure x86 could be.

Re:Is it enough?

[RightSaidFred99]

This is false. Itanium was (and is) very powerful in specific domains (scientific computing, database servers, etc...). The design is fine, it's just not been nurtured as much because Intel finally realized it wouldn't be a mainstream processor anytime soon.

Re:Is it enough?

[Anonymous Coward]

I don't see a problem with Intel. The 8086 was simply supposed to be a stop-gap between the 8085 and the iAPX432. The 432 was supposed to be the way of the future, but it never caught on. Then the i860 was Intel's RISC CPU, but that didn't catch on either. Finally, they bet the farm on Itanium, but that hasn't been able to replace the x86 either. What's Intel supposed to do? It's really AMD that's extending the x86 architecture to the point where we don't need Itanium!

dom

Re:Is it enough?

[TheLink]

Nah it's a crap design. Doesn't do that well in SPEC CPU 2000 or SPEC CPU 2006 and most real world apps. See: http://www.spec.org/cpu2006/results/cpu2006.html and http://www.spec.org/cpu/results/cpu2000.html

The fastest Itanium 2 there isn't faster than the fastest x86 in floating point performance or integer performance when you compare quad cores with quad cores, etc.

And look at price, power consumption for the performance you get. Expensive (the same number of transistors in an Itanium 2 system (18MB or even 24MB cache!) will get you a fair number of Core 2 Duo/X2 cores), hot, and only fast in very specific cases. If I wanted an expensive solution that's high performance only in specialized cases I'd be using an ASIC or DSP.

Compare the Itanic with IBM's POWER range and you'll see the problem- if you are going to pick a platform that is expensive, isn't x86 compatible, you might as well switch to IBM's POWER range. You provide the large sums of money and at least IBM will provide the hardware, O/S, apps, the people to blame aka consultants ;), and people to yell at aka support. Whereas Intel has nothing but the CPU.

Scientific computing? go talk to the HPC people nowadays and see what they are buying. They aren't betting on the Itanic.

Database? I don't see how VLIW helps for DB aps. AFAIK Opterons currently (better memory bandwidth) are the way to go for that. Whatever it is, it sure isn't Itanium 2 unless your "Enterprise" DB fits in the 24MB caches of the Itanium.

The advantage of the x86 ISA is that it is CISC and thus fairly dense - think of it as compressed instructions - and nowadays when bandwidth is an issue (memory bandwidth, cache bandwidth etc), CISC isn't so bad. IBM's POWER is pretty dense as well.

x86 is still ugly though. The pig with a jetpack is flying faster than the eagles etc, but it's still a pig ;).

Re:Is it enough?

[Anonymous Coward]

Intel has tried to move away from x86: look at the Itanium and Itanium II.

Also look at iAPX 432, i860 and i960, a long time before itanic (which is a science project gone wrong).

These price cuts are pretty sweet

[d12v10]

If you don't need the latest and greatest processors for a new system, these cuts are awesome. Just two months ago the X2 3800+ was about $109.99, but can now be purchased for close to $86.00!

Re:These price cuts are pretty sweet

[bsharitt]

I just paid $90 for the 3600+ less than 2 weeks ago. No I wish I would have waited since NewEgg has the same chip for $73 now.

Planned obsolescence

[Anonymous Coward]

"Most of the remaining chips on AMD's price list use the Socket AM2 or Socket F form factor, rather than the older Socket 939 interfaces."

This is the thing I don't like. When I was shopping for a processor it was either socket 754 or 939. I chose the 754 even though it was going to be obsoleted by the 939, because I knew that as predictable as day and night AMD would do the same thing to the 939 leaving me hanging and having to change everything to keep up. And the 939 hasn't been that long on the market. Intel does the same thing but not as bad as AMD.

Re:Planned obsolescence

[mrbcs]

Actually socket 939 had the shortest lifespan of any AMD socket to date. I was told by my wholesaler to avoid it like the plague cause it was only going to be around for about a year and a half.

Quite amazing when you realize how long we sold socket A boards. They went from a duron 700 to well over Athlon 2400+ over about 3 years IIRC.

AM2 is supposed to be here for a while now. Just ordered a 4400 today. It's gonna cost me about $138 canadian. Got an Asus board, the chip, Nvidia 256 video card gig of ram and a 200 gig seagate for under $500 canadian. Might buy the 6000 if the price does drop that much in the next little while.

Re:Planned obsolescence

[plasmoidia]

Intel does the same thing but not as bad as AMD. Ummm, right.... Before the 64s came out, AMD had the Socket A, which was around for a very long time (Duron through Athlon XP). Before that was (Super) Socket 7. How many sockets did Intel go through in the same amount of time? Check out Wikipedia. Also, remember that with CPUs, even a few months can be a *long* time. Only recently has AMD been changing sockets with any sort of frequency. And each change has been for a single reason: memory. Since the memory controller is integrated on the CPU, the socket and memory type are closely related. Socket 754 uses single channel DDR SDRAM. Socket 939 uses dual-channel DDR SDRAM. Now the new Socket AM2 uses DDR2 SDRAM. Different memory, different controller, different socket. It is that simple. AMD has been much more stable than Intel in the area of sockets. Period.

Not that I like the changes either. I have a 939 motherboard and only a single core processor that I bought about a year and a half ago. I would like to be able to pick up a cheap 939 x2, but looks like maybe I waited too long. But that is the way it goes.

Re:Planned obsolescence

[Anonymous Coward]

http://www.newegg.com/Product/Product.aspx?Item=N8 2E16819103052

socket 939 X2 3800+ for $95.

Re:Planned obsolescence

[TheLink]

And AMD is better in the "if the CPU fits in the motherboard socket it's likely to be compatible". Even if it's not recognized in the BIOS and you just see the MHz etc, it doesn't matter - it still works.

Whereas for recent Intel CPUs even if their CPUs fit in the socket, they often only work with particular chipsets.

Intel has definitely been worse- they moved from socket to slot (remember slot1?), to socket etc.

Re:Planned obsolescence

[drsmithy]

Ummm, right.... Before the 64s came out, AMD had the Socket A, which was around for a very long time (Duron through Athlon XP). Before that was (Super) Socket 7. How many sockets did Intel go through in the same amount of time? Check out Wikipedia.

Sticking with "mainstream" CPU sockets/slots (ie: no mobile or server parts, since they're not really relevant) since the K6 and Pentium 2 (1997), we have:

AMD (6)
Super socket 7
Slot A
Socket A
Socket 754
Socket 939
Socket AM2

Intel (5)
Slot 1
Socket 370
Socket 423
Socket 478
LGA 775

Seems pretty even to me...

Re:Planned obsolescence

[plasmoidia]

socket 939 X2 3800+ for $95. Yeah, but I was hoping for something a little more powerful. Or at least cheaper. I currently have a 2.2 GHz Athlon 64 3500+. And for $140 I could get a 2.5 GHz, 65 nm, x2 for Socket AM2. For a processor, I consider the extra $45 almost negligible (plus the AM2 is a retail, whereas the 939 is an OEM).

The only problem is I have to, at minimum, also get a new motherboard (~$100-150 maybe) and new memory (I currently have 2 GB, so that would probably be ~$175-200 to match that) to use it. And then it is probably worth it to go ahead and get a new video card, case, PSU, and hard disk to make a whole new system.

...Unless there is someone out there who would like to buy a used socket 939 motherboard with a 3500+ CPU and 2 GB RAM... :-)

Re:Planned obsolescence

[plasmoidia]

Seems pretty even to me... Okay, so it does seem close. It looks like you forgot Socket M for Intel, making it dead even. And I think we might could debate about whether Slot A, under AMD, should count ;-). But regardless, look at the timeline here. For AMD, the last three sockets have come out in the last few years (with AM2 being almost brand new). I believe all five of Intel's sockets were out during the time that the first three AMD sockets were used. That was mostly my point. That and basically any time AMD has changed the socket, it has been necessitated by a change in the memory architecture.

Question

[joshier]

Just wondering..

I'm an electronic musician, I use Ableton live and it does indeed support multiple CPU threading for dual cores.

Now, I currently have a 3200+ (single core) Athlon and I'm wondering whether to upgrade to a 4600+ dual core... Will this improve performance much? or is it best to wait for a 5000+?

Thanks

To tell you the truth...You CAN handle the truth.

[Anonymous Coward]

"Who remembers Itanic? Itanium2 was less of a joke, but not great."

Itanium's failure wasn't so much chip as the fact that compiler technology wasn't up to the task. Anyway remember while the CPU is important. It's the chipset that's the heart of any system.

Price isn't everything; boycott AMD

[vandan]

AMD / ATI have always had quite a negative attitude towards open-source. Their closed-source Radeon drivers lag behind the open-source drivers severely in features and stability, but AMD refuses to allow open-source developers access to information required to complete support for newer Radeons.

Intel, on the other hand, actively support the development of open-source drivers for their graphics chipsets. Indeed, just today, I noticed this announcement:

Announcing a new video driver called xf86-video-vermilion which provides
basic 2D functionality using blits & fills for the Intel Vermilion Range
chipsets.

Intel funded Tungsten Graphics to do this work. There's also an FBDev
driver being submitted to the linux-kernel.

... which is a world of difference from what we expect from AMD. The last release from AMD managed to add ... wait for it ... a new control panel that looks like arse and adds precisely nothing of value. Previous releases have done things like remove support for current hardware, remove support for custom modelines, etc.

So to Linux users wanting to support the 'underdog' ... forget it. Support the company that supports open-source software instead. Imagine what would happen to the performance and stability of your AMD system if AMD suddenly decided to stop playing ball with gcc developers, and instead release their own compiler. And yeah, I realise that Intel released their own icc compiler, but at least they continue to support gcc developers as well. But if you take AMD's stance towards open-source developers to its logical conclusion, you've got gcc that doesn't support AMD chips. Is that what we want?

Re:Price isn't everything; boycott AMD

[Anonymous Coward]

And AMD propaganda @ http://www.youtube.com/watch?v=6Vsz0jgICoo ?

Re:Price isn't everything; boycott AMD

[Anonymous Coward]

Didn't you work on the 5th floor of SC12?

Re:Price isn't everything; boycott AMD

[mrchaotica]

So, where's that Free Software Intel wireless chipset driver, then? (Just sayin'...)

Also, although I agree with you about old ATI's shitty attitude, it seems a little premature to condemn AMD for it. They haven't been the same company for that long yet, you know, and it remains to be seen whether AMD's leadership might change things.

(Note: I'm not an AMD fanboy; in fact I'm posting this from a Core Duo laptop.)

Re:Price isn't everything; boycott AMD

[UncleFluffy]

You're comparing apples and oranges, and, not only that, but comparing two different types of apple with a single type of orange. Or something like that, whatever the correct metaphor should be.

AMD and ATI have only just merged, so it's a little early to judge the "open-source friendliness" of the new company. However, the history is varied. AMD provides very strong support for gcc and other projects that are important to it, funding a number of full-time developers. ATI hasn't provided as much support for its graphics chipsets - partial documentation rather than active development - but the open-source drivers for ATI still do more than those for Intel, just because Intel graphics chips are a whole lot simpler. (Which means that Intel aren't losing any trade secrets by exposing their internals - their graphics chips are using less "clever stuff" than ATI or nVidia). Every ATI chip that has capabilities in the same class as any Intel graphics chip you care to name has a complete open-source driver, and for every closed ATI GPU there's an equally closed Intel wireless chipset.

Both companies play nice where it suits them and take their ball and go home where that is percieved as the more profitable option. My advice to you is to do the same and pick whichever option gives you more performance per dollar.

Re:Price isn't everything; boycott AMD

[Martin Blank]

So, where's that Free Software Intel wireless chipset driver, then? (Just sayin'...)

That would be right here. Everything non-binary is licensed under either GPLv2 or dual-license BSD/GPLv2, according to the documentation. The binaries are released in that form because they are prohibited by FCC regulations from releasing anything that could be modified by the end user to violate regulatory limits. Exactly the same thing applies to the MADWifi drivers for Atheros, who makes available a binary HAL to the developers.

Re:Price isn't everything; boycott AMD

[Anonymous Coward]

WTF?

Re:Price isn't everything; boycott AMD

[renegadesx]

Exactly, AMD hasn't exactly been THAT bad to open source, ATi has always hated it, but so does nVidia. The graphics card market is a cuthroat market between those two to get the most out of their chips and I don't see them opening up their drivers anytime soon but anything can happen. AMD is frendlier to open source than ATi is and once they mould into one company who knows what will happen.

Re:Price isn't everything; boycott AMD

[Mad Merlin]

Soon the binary only regulatory daemon won't be necessary, see http://intellinuxwireless.org/.

Re:Price isn't everything; boycott AMD

[mellonhead]

So, where's that Free Software Intel wireless chipset driver, then? (Just sayin'...)

An Openbsd guy wrote one but then a Linux guy accused him of stealing the code...

Re:Price isn't everything; boycott AMD

[JohnFluxx]

AMD spent a lot of money on improving gcc (for all architectures) while Intel only made their proprietary closed icc compiler.

I think blaming AMD for ATI's old mistakes is a bit premature at this stage.

Re:Price isn't everything; boycott AMD

[cortana]

Nonsense, that is a common lie trotted out by companies that don't really want to release free software drivers.

See the Ralink cards for examples of drivers that only consist of free software.

Re:Price isn't everything; boycott AMD

[tuxic]

That's very enlightening and explains a lot. I had wished this for be more well-known to the public and talked about. So, Intel can't do anything about it but they get disappointments from users who aren't aware of the situation. Hmm.

Re:Price isn't everything; boycott AMD

[raddan]

The "we kept it in binary to appease the FCC" is bogus argument. Anyone can build a transmitter that interferes on any nearly frequency they so desire with simple analog electronics. If Intel was really so concerned-- why not prevent the device from broadcasting on those frequencies in hardware? Also, why not allow free redistribution of the firmware? The GPL/BSD drivers are useless without it. No, Intel's using that argument to distract you from the fact that the binaries compensate for garbage hardware, and they don't really give a shit about F/OSS.

Re:Price isn't everything; boycott AMD

[Martin Blank]

The ability to build a transmitter that can interfere with others does not mean that it's legal, and in fact is expressly illegal in the United States. 47 CFR 15.5b states:

Operation of an intentional, unintentional, or incidental radiator is subject to the conditions that no harmful interference is caused and that interference must be accepted that may be caused by the operation of an authorized radio station, by another intentional or unintentional radiator, by industrial, scientific and medical (ISM) equipment, or by an incidental radiator.

Emphasis added.

I can drive my car the wrong way on the road, but that doesn't make it legal.

Looking around, it seems that Intel released in February a new driver using the d80211 subsystem that does away with the regulatory binary. There's still a binary microcode update that has to be loaded, and it's freely (as in beer) redistributable. I haven't played with it yet, but the initial response seems to be relatively positive.

Re:Price isn't everything; boycott AMD

[Anonymous Coward]

No, actually, AMD will not open source ATI's drivers or specifications anytime in the future:

http://liquidat.wordpress.com/2007/03/13/amdati-li nux-interview-with-terry-makedon-the-head-of-softw are-development/

"AMD/ATI & Linux - interview with Terry Makedon, the head of software development
March 13th, 2007 -- liquidat

cube-with-matrix
German golem.de features an interview with Terry Makedown about the current situation and the future prospects of Linux drivers for the AMD (former ATI) graphic cards. He makes clear that the legal problems alone are enough to keep the drivers close.

Thanks to German Golem.de we have an excellent interview with Terry Makedon concentrating mainly on the Linux driver development. Since the interview is published in German I will try to give a rough overview of the main points.
Actually the interview was mainly dedicated to the Vista drivers, but since these are coming along nicely the topic shifts to Linux quite fast.

Makedon points out that the driver development efforts for Linux were increased in the last two years. But he also identifies a major problem with the resources: the man power put into the Linux drivers is proportional to the percentage of Linux usage in the world. In the interview he mentions 5 % as an example.
Also, the development focuses on Workstations and professional use. The focus is explicitly not on games on Linux because that is not justified by the current situation in the real world.
Btw., I think that could be a reason why there is still no support for AIGLX in the AMD drivers, but quite nice support for dynamic display switching - although I'm still waiting for a nice GUI.

Anyway, opening the drivers is no alternative for Makedon: first of all there is too much 3rd party technology in the drivers "like for example compression algorithms". The 3rd party companies certainly wouldn't allow AMD to release that source code.
Second there is too much IP in the drivers. He compares the situation with McDonalds which will not release the recipe to the sauce of the BigMac also: "It is just not economically justifiable to do that."
Also, an NDA like Intel uses is not possible because - according to Makedon - there is not "so much 3D intelligence" inside the Intel drivers.
However, he adds that the 2D part is already open source and that there are discussions to also release other parts like the install routines as open source as well. He cannot say never, but that AMD will not take this path for at least the next 6 till 12 months.

The interview closes with the note that it is not a political decision but that the right approach is still missing: "Of course we want to talk with the Linux community to open the code. If it helps to improve our products and doesn't bring us into legal trouble or brakes our business secrets [...]".

The golem.de article adds that the current development focuses on the "Catalyst Control Center" - it is supposed to be released for Linux until May 2007."

Re:Price isn't everything; boycott AMD

[slashdot_commentator]

An Openbsd guy wrote one but then a Linux guy accused him of stealing the code...

Yes, I wrote about it in a journal entry.

So, where's that Free Software Intel wireless chipset driver, then?

Its there if you run linux.

Re:Price isn't everything; boycott AMD

[raddan]

You missed my point. I'm not advocating that Intel should allow us to create interference; I'm saying that Intel's keeping drivers in binary form to prevent users from creating interference is a red herring. If a user wants to broadcast at a different frequency, they can have it easily. Intel is keeping parts of the driver closed for other reasons.

It does appear that Intel has changed their stance wrt to redistributing the microcode binaries. It is unclear, however, if this would meet the requirements needed by, say, OpenBSD. There is mention that the microcode requires a user-space regulatory daemon in some cases-- this probably wouldn't fly with that project.

Re:Price isn't everything; boycott AMD

[dbIII]

AMD / ATI have always had quite a negative attitude towards open-source.

How many weeks have AMD owned ATI for? It certainly is not a full year yet - so I would say the post above is offtopic an irrelevant since we are talking about AMD and central processing units.

Re:Price isn't everything; boycott AMD

[Martin Blank]

There seems to be a lively debate on this, and it's never been tested in the courts, but a company with an investment of hundreds of millions of dollars in the technology cannot afford to release the source code to the binary and allow any random person to easily modify the binary to actively interfere in others' transmissions, especially when their software radio is perfectly capable of working in restricted frequencies such as the 4.9GHz range, not to mention potential interference in the radar frequencies that 802.11h was designed to mitigate.

OpenBSD's development staff desires to remain pure in this matter. That is their decision, but they may end up having to make some sacrifices to do so, as might Debian and other projects that refuse to use anything not under clear and accepted open source licenses.

I know why

[OrangeTide]

It was because I just bought a new Athlon X2 cpu last week.

sockets, or, how AMD made sure I won't upgrade

[Anonymous Coward]

I got suckered into buying a socket 939 system. Now I'm at a dead end. I can't get one of those higher end 3GHz parts even if I wanted to, because they're AM2 only.

Good job AMD! You made sure I won't be buying one of your latest CPUs, when I otherwise would have done so.

Contrast that to the older Athlon XP socket that lasted me from from a 900 MHz part through 3 upgrades to my XP2700 - and let me hand down the older CPUs to backup systems running the same socket.

Grumble grumble.

Re:Damn, another weekend at Frys.

[Professor_UNIX]

I'll still get the RAM at NewEgg, I DO have SOME standards.

Why would you buy RAM at Newegg, but processors at a box store? They're almost certainly going to be more expensive at Frys than they are at Newegg.

all the benchmarks are 32-bit

[r00t]

AMD chips go 30% faster when 64-bit.
Intel chips go 5% slower when 64-bit.

I suppose this is an indication that Intel marketing pays attention to the very lame old 32-bit benchmarks that are getting used.

64-bit is here now, even if you run Windows. Linux people have had pure 64-bit systems for many years.

Re:all the benchmarks are 32-bit

[Sancho]

I'm one of those freaky Linux people who doesn't mind the occasional bit of proprietary, closed-source software on his systems. Flash is a big reason I haven't made the 64-bit jump yet. I don't want to mess with broken chroot environments (and probably other gotchas with various binary software).

That, and going to 64-bit just isn't that useful right now. With AMD, you may get some speed increases (I really can't say, as I haven't seen the benchmarks or performed any tests myself for Linux--for FreeBSD, I know that the performance just isn't there) but that's architectural. Intel generally beats AMD64(in 32-bit mode), and if I don't have any reason to run in 64-bit mode (and plenty of reasons not to) then what's the point? My next machine will be from the Core2Duo line.

AMD had their day, and they may have it again, but for right now, I'm not interested.

Re:all the benchmarks are 32-bit

[na641]

actually you don't need a chroot at all to run 32bit programs in a 64bit linux environment. Atleast with gentoo all you need are the correct 32 bit libraries installed and you can compile and run32bit programs (like firefox coupled with flash) natively.

As for running a 64bit environment, the biggest factor is memory usage. Its true that 64bit programs yield slightly faster performance boosts, they take a lot more ram (think cumulative of all programs running and you can imagine what i mean) than their 32bit counter parts.

Re:all the benchmarks are 32-bit

[Anonymous Coward]

AMD chips go 30% faster when 64-bit.
Intel chips go 5% slower when 64-bit. Without sources (e.g. links), hardware details (e.g. Athlon 64, x2, Prescott, Pentium D, Core 2 Duo), and software details (e.g. Linux, Windows, benchmark software), your claims seem like AMD fanboy bullshit.

Re:all the benchmarks are 32-bit

[MoxFulder]

actually you don't need a chroot at all to run 32bit programs in a 64bit linux environment. Atleast with gentoo all you need are the correct 32 bit libraries installed and you can compile and run32bit programs (like firefox coupled with flash) natively.

Ditto with Debian and Ubuntu and probably every other 64-bit distro :-) Just install the 32-bit libs and you're good to go. It's a little messy to set this up for flash, but frankly... we can blame that on Adobe/Macromedia and their proprietary not-64-bit-safe crap code. Millions of lines of open source code got cleaned up for 64-bit over the past few years: why can't Adobe get their ass in gear?

Frankly, I got a 64-bit CPU cause I wanted to *use* it. It gives me extra registers (important for x86 code!) and access to more memory. And the warm smug feeling of knowing that all my Linux apps run 64-bit native *years* before the same can be said of Windows.

As for running a 64bit environment, the biggest factor is memory usage. Its true that 64bit programs yield slightly faster performance boosts, they take a lot more ram (think cumulative of all programs running and you can imagine what i mean) than their 32bit counter parts.

This is certainly a noticeable effect. Though so many apps use so much memory, I imagine there's a lot of low-hanging fruit to improve memory usage without worrying too much about 64bit vs 32bit.

Re:all the benchmarks are 32-bit

[oojah]

Where's your proof?

You've already had someone respond with a link to benchmarks showing exactly the opposite of what you claim:

http://www.pcstats.com/articleview.cfm?articleid=2 097&page=12

Where are the benchmarks that show what you claim?

Cheers,

Roger

Re:all the benchmarks are 32-bit

[wild_berry]

You, sir, are a cad and a bounder. What do the ScienceMark and Primordia scores show on Page 11 of the article you linked (http://www.pcstats.com/articleview.cfm?articleid= 2097&page=11)? Did you not need to include this data so we can arrive at fair conclusions?

ScienceMark 2.0 has the Athlons run four times faster in 64-bit mode than 32; the Core2 Duo speeds up by a factor of three. Primordia has the AMD64 speed up by a factor near 8/7; Core2 by the smaller factor of 9/8. I'd say that it's swings and roundabouts, as ever with computer architectures.

Re:all the benchmarks are 32-bit

[Sproggit]

Actually I've had 64bit alpha with OpenVMS and linux for damn near a decade.

The Sproggg

Re:all the benchmarks are 32-bit

[swillden]

Ditto with Debian and Ubuntu and probably every other 64-bit distro :-) Just install the 32-bit libs and you're good to go. It's a little messy to set this up for flash, but frankly... we can blame that on Adobe/Macromedia and their proprietary not-64-bit-safe crap code.

The Debian team is working on a new multiarch system that will address this, making it simple to install mixed architecture software on machines that support it. Basically, the packaging system will understand all of the various ISAs and their relationships, and which ones will run on which processors, and the package dependencies.

This means that rather than having aptitude (or your apt front end of choice) show you six different versions of the linux-image package (-486, -586, -686, -k7, -k8, -amd64, etc.) or the mencoder package, there will only be one linux-image and one mencoder in the list. The various binary versions will still exist, and by default apt will pick the best of those available for your platform. If you want, however, you'll be able to override its choice and pick a different one. If the one you pick requires different versions of support packages, then dpkg will also know that and apt will handle all of the dependency management, making sure that the right versions of everything are installed.

So for example, if you have an Athlon 64, by default apt will install 64-bit versions of everything. If, however, you decide to install the flashplugin-nonfree package, which is 32-bit only, apt will recognize that it cannot be used with your 64-bit browsers and offer to replace them with 32-bit versions. Since the 32-bit versions require 32-bit libraries, it will also offer to install the required libs. Part of the multiarch specification is a scheme for making it easy to install multiple versions of a given library side by side, and for automatically configuring apps to find the correct library versions.

This might seem like an overly-general solution for addressing the temporary x86 32- to 64-bit transition, but the Debian developers doing it have recognized that as just one example of a much larger problem, including:

• A half dozen of Debian's target platforms are already in the same position with processors that support 32- and 64-bit versions of the same ISA. This isn't just an x86 issue.
• Some processors support running other ISAs via emulation, software or hardware. For example, you can run lots of ISAs via software emulation on i386 via qemu, and Itanium runs ia64 natively but provides hardware emulation for i386.
• It's possible to use compatibility libraries to run software from other OSes. You can run some Solaris/sparc applications on Linux/sparc, for example, and you can run DOS and Windows software on Linux via FreeDOS, WINE, etc.
• Some processors support mixed-endian binaries.
• Many processors, particularly in the x86 world, support lots of different ISA subsets -- i386, i486, i586, i686, k7, k8, MMX, SSE, etc.
• All of the above is complicated by the fact that some packages can run on multiple architectures, but perhaps not all architectures.

So, the plan is to develop a solution that addresses all of these issues in a general way, rather than continuing to use various architecture-specific hacks to get around the fact that dpkg currently believes a machine has only one architecture. It's a pretty big project, but people are pushing to get it included in Lenny (which, judging by past releases should be out mid-2009), and users of unstable and testing should see it much sooner.

Re:all the benchmarks are 32-bit

[oojah]

You're absolutely right. I was just carelessly pasting in a link from another comment. My only intention was to put up some argument to show people the gp wasn't completely in the right.

Re:all the benchmarks are 32-bit

[shawnce]

ScienceMark 2.0 has the Athlons run four times faster in 64-bit mode than 32; the Core2 Duo speeds up by a factor of three.

Exactly... which again disproves what the following bogus statement...

AMD chips go 30% faster when 64-bit.
Intel chips go 5% slower when 64-bit.

No data that I have seen anywhere supports a statement like that... Core 2 chips are faster when in 64-bit mode (assuming compiled to utilize additional registers, etc.). The only likely edge case would be a pointer heavy data stream that causes cache to be blown out because of the 2x pointer overhead but you would see such an issue earlier on the AMD chips because of smaller caches (granted the AMD chip can offset some of that because of the lower latency memory interface).

Re:all the benchmarks are 32-bit

[Anonymous Coward]

No Flash support is one of the joys of running 64-bit Linux. My browsing experience is *so* much less cluttered.

Re:I don't really care

[Anonymous Coward]

The Radeon Xpress onboard GPU should be up and running with the r300 DRI driver any day now...

consider AMD...

[r00t]

...unless you'd like reduced performance on 64-bit code.

Going from 32-bit to 64-bit, Intel performance drops 5%. AMD performance goes up 30%.

You can't stay 32-bit forever. Even the Windows gamer world will end up 64-bit. Linux has already moved, with 100% 64-bit being common for years now.

Intel also does badly when you have more than 4 GB of memory. The AMD chips have an on-chip IO-MMU that can be used to avoid bounce buffers. PCI DMA on an Intel box can only reach the low 4 GB of memory; the OS must copy the data around if you have more RAM.

Re:consider AMD...

[Capricous]

Can anyone confirm this?

Re:consider AMD...

[daft_one]

Netcraft confirms it!

Re:consider AMD...

[AnyoneEB]

Do you have a source or are you just trolling?

Re:consider AMD...

[Anonymous Coward]

Going from 32-bit to 64-bit, Intel performance drops 5%. AMD performance goes up 30%. Without sources (e.g. links), hardware details (e.g. Athlon 64, x2, Prescott, Pentium D, Core 2 Duo), and software details (e.g. Linux, Windows, benchmark software), your claims seem like AMD fanboy bullshit.

Re:consider AMD...

[saikou]

I decided to look for 64-bit and 32-bit comparison charts. So I googled and found one at PCStats.com. So what does zlib Mini-GZIP 1.2.3 32-bit vs. 64-bit benchmark says? It still says Intel Core 2 Duo E6600 (WinXP 64/64 Bit) beats AMD Athlon64 FX-62 (WinXP 64/64 Bit) by roughly 30%. So does 64-bit DivX encoding.
And even at Science Mark 2.0 at which Intel C2Duo is slower than FX-62, switching into 64 bit reduces time needed to run the test from 66.241sec to 21.36.

At least provide some sort of sources when claiming performance drop in 64 bit mode. According to the above benchmark I do want to buy C2Duo and run it in 64-bit mode to do all the gzipping.

By the time consumers will start to care about PCI DMA eating more than 4GB of memory, the new revision of Intel CPU will be out with on-die controller ;)

AMD is the best choice for budget customers right now

Re:consider AMD...

[kingbob2]

I was looking at building a new machine soon and was looking at the E6600. with the x2 6000 being around the same price and maybe cheaper I'm starting to consider that also, but with the overclocking abilities of the 6600 it might stilll put it ahead of the 6000+. anyone have any numbers to help me solve this with $/performance and average overclocking factored in?

Re:consider AMD...

[Anonymous Coward]

This is retarded. Intel has always been good for gzipping even with their Pentium 4 processors. You are free to think that a gzip 'benchmark' means that Intel cpus are much better than AMD's in a 64-bit environment. Do I get to claim that AMD cpus are much better than Intel cpus because they cream Intel cpus in SSL 'benchmarks'?

Re:I don't really care

[Anonymous Coward]

Yeah but does anyone use the Intel 3D stuff for anything? Performance-wise I don't think the very latest chipset is suitable for even last-generation stuff.

Besides, in my experience the Linux nVidia drivers have been excellent and I have been using all nVidia for my high-end stuff for the last 6-7 years. I did try ATI for a while in some side projects and was not impressed with their Windows or Linux support.

AMD X2 price cut

[IceMan33]

I'm glad that AMD has taken so long to finally cut the prices of their processors. It has been a long time since the introduction of the dual processors that intel has designed. With these price drops all I can say is that it is about time. AMD Needed to drop the prices earlier to help with profits which would lead to money to improve their processors. All I can say is that it's about time!

Re:AMD X2 price cut

[Anonymous Coward]

All you can say is it's about time! Several times!

Yawn

[jmitchel!jmitchel.co]

The Inquirer had these numbers a couple of weeks ago friday, newegg adjusted by the following Monday, and this past friday, the local Frys flogged me a retail boxed AM2 3800x2 EE and a cheap motherboard for $90 - which feels like it Must have cost SOMEBODY money.

One example doesn't make a rule

[jd]

In fact, AMD's HyperTransport specs are available for free to anyone. Intel's PCI/PCI-X/PCI-e specs are only available to paid subscribers. Not exactly open, is it? But, again, one example doesn't make a rule.

And what of chip companies that do publish specs? There are MANY chips from FreeScale (formerly Motorola's semiconductor division) that include fantastic levels of documentation. All the calls, all the functions, all the features. There are bugger all drivers in any Open Source *nix (xBSD, Linux, Plan9, you name it) for the S1 encryption chip. You want to talk about supporting those vendors who support Open Source? Then support them by adding that support.

Let us get down to basics, here. Part of the reason why companies like ATI can avoid supporting Open Source is because the Open Source community has, itself, failed to support the Open Source community. We have not been perfect, shining examples of our own standards and have no right to expect others to adhere to ideals we ourselves fall desperately short of.

Sure, the Open Source community lacks the kind of funding needed for this sort of stuff. So does AMD, whose profits were almost a billion short of expectation, whose net worth is now not much more than ATI prior to being bought, and whose future (due to Intel's near-monopolistic control over the industry and near-inexhaustible supply of funds) is severely in doubt. AMD has less than a tenth of the money of Intel and can't afford the current price-war for much longer. In the meantime, Intel can not only afford it but can afford to make next-gen components that have exactly the same flaws in concept as all their products have always have. Intel can afford it, Intel will essentially kill AMD, and Intel will only correct the flaws in the logic the next time it is threatened by a chip company.

(I may sound a little harsh on Intel there, but it's basically true of all corporations. Quality for the sake of quality is not a concept most managers comprehend, and "engineering excellence" is an oxymoron in any group outside of a few fringe development projects and maybe a couple of Formula 1 teams.)

If support for Open Source were a criteria, I'd say support nobody and move to another planet. As the old NASA joke goes, there is intelligent life on Earth but it's only visiting. There isn't any meaningful support for Open Source, outside of a handful of individuals.

What about IBM? All those 500+ patents they freed up! Yeah, and how many projects do you see based on them? None? Is that a surprise, when most were hardware patents? Outside of OpenCores, I really don't see many people being able to do much with pipeline optimization or CPU scheduling, and frankly most coders there working on CPUs have been doing just fine using their own methods of solving these problems, and anyone likely to want a high-end 64-bit Open Source chip would probably be looking at the Open Sourced UltraSparc. IBM have released lots of bits of project in the past, but never really maintained them and never really did anything with them. You been using IBM's GUI-based Apache management tool? Ever realized IBM had one?

The community should, by rights, support anything and anyone it can, AMD included, because a monoculture would be far far worse than the putrid stench we have at the moment. The existing mess can be fixed, with a lot of time and a lot of patience. Monocultures are stagnant cultures are cultures waiting to die. What we have right now is no great shakes, but I'll take it over a living death any day. The dead can't be cured - well, unless they're a kipper.

The Price/Performance Argument Hipocracy

[Deviant]

I remember during the Pentium IV days all the AMD fans were constantly talking about how AMD owned the price to performance crown and that Intel was overpriced, ran hot (energy inefficient), and was just all around not as good an architecture. They were right - and I bought an Athlon 64 instead of a P4.

Now those same people are trying to argue that the less expensive, cooler and more efficient Core2Duo are still not as good as their beloved AMD. They will point to 64 bit performance or performance over 4GB of ram - or a myraid of little things that are not relevant to the vast majority for at least the next couple years to support their bias.

The processor wars, just like the video wars, will go back and forth. Nobody stays on top forever. Intel, after many years trailing, had their leap ahead for a generation or two. The people who are the most rational go with the best architecture or company at the time. I bought an ATI 9600 instead of a Nvidia 5600, even though I had always owned Nvidia and loved the drivers, because it was the better value for the money at that time.

The bigger person, the more rational person, is the one who can be objective about these things. Which CPU company you "love" is a very strange thing to have an irrational passion about...

Re:The Price/Performance Argument Hipocracy

[gmack]

Last time I pieced together a system I found that I would end up paying more for a core2duo once I factored in the cost of the motherboard.

And that was before I found out about the 32bit vs 64 bit performance difference. It all worked out in my favor since I run my system in almost entirely 64 bit mode.

Re:The Price/Performance Argument Hipocracy

[Pope Raymond Lama]

Except that if you feed the giant because it is 10% faster, or 12% cooler now, the small one may starve out, and bye bye "processor wars enduring years". You get stuck into a monopoly that won't do any good.

That is why I went AMD on my new system, even though I was paying, right now, a little more for less. (Which, I learned later, as I am on 64 bit is not even less)

Re:The Price/Performance Argument Hipocracy

[JohnFluxx]

How about simply rooting for the underdog because it would really suck if Intel (or any other company for that matter) because a monopoly?

Re:The Price/Performance Argument Hipocracy

[javilon]

If you run Linux you don't buy ATI. Their drivers truly suck and any performance advantage the ATI card could have on paper will not be realized because of the drivers.

Re:The Price/Performance Argument Hipocracy

[Ed Avis]

64-bit performance is surely more important than 32-bit. Most Linux distributions have had an x86_64 version for years. Apart from that crappy Flash plugin to bombard you with animated advertisements when you browse the web, what 32-bit software would you want to run?

Re:The Price/Performance Argument Hipocracy

[lateralus_1024]

I'll admit that there is a part of me that wants AMD to win. For the little guy that doesn't advertise much and performs well. After years of painful Intel marketing and lackluster technology they have finally put out a good product and that's good for them. I'll likely continue to use AMD on the desktop and Intel on the mobile.

Re:The Price/Performance Argument Hipocracy

[Andrei D]

The bigger person, the more rational person, is the one who can be objective about these things. You must be new here...

Re:The Price/Performance Argument Hipocracy

[Deviant]

I am not sure about that. Comparing their 64 bit performances aside (where it is still equal if not ahead), if you compiled those linux binaries for 32 bit instead of 64 bit on your Core2Duo how much of a difference would you see versus 64 bit running on a AMD? Unless you were addressing large amounts of memory or doing something rather out of the ordinary I doubt that you really need 64 bit for anything but stroking your ego to know that you are more leet because you compiled it with that flag. If you are running XP or Vista than 64 bit is even worse - a bunch of driver and compatability issues with no real performance benefit.

Re:The Price/Performance Argument Hipocracy

[GooberToo]

The problem with running 64-bit Linux is you wind up with what almost equates to two Linux installations. This is because you wind up having both a 64-bit and 32-bit environment, otherwise you can't run 32-bit software. Worse, lots of software that you may want to play with has never been ported to 64-bit. This means you either wind up compiling these odds and ends as 32-bit or you wind up porting it to 64-bit. Again, this means a 32-bit and 64-bit development environment, plus all those dang libraries.

I got tired of constantly porting software to compile as 64-bit. Worse, there exists a number of audio issues. If for example, you want to use a 32-bit application, which plays audio, on your 64-bit system, which happens to use the old OSS interface, you'll be required to have the full 32-bit audio support tools in place and you'll still have problems muxing your audio. I no longer remember the exact details but I can tell you that Teamspeak was at the top of my curse list.

Worse, I also had a lot of application stability problems with my 64-bit applications which seemed to hint that many of the 64-bit applications were not quite ready for the 64-bit world. Granted, this was a year or two ago (wow...I guess it has been that long) and things have probably, greatly improved...but I'm certainly not in a hurry to go back to a 64-bit Linux desktop. Maybe next year...

Re:The Price/Performance Argument Hipocracy

[Ed Avis]

The problem with running 64-bit Linux is you wind up with what almost equates to two Linux installations. This is because you wind up having both a 64-bit and 32-bit environment, otherwise you can't run 32-bit software.

That used to be the case a couple of years ago, when OpenOffice.org was still 32-bit only. Indeed, you needed 32-bit and 64-bit copies of every library and it was a bit of a mess. But nowadays pretty much everything is 64-bit native. See the LWN story The end of the multiarch era?.

Even back in the relatively early days of x86_64 I never had any trouble running Fedora 64-bit.

Re:The Price/Performance Argument Hipocracy

[Ed Avis]

Yeah it probably is just stroking your ego to run a 64-bit system, then again, that's pretty much the only reason for upgrading to a faster CPU anyway.

On the Athlon 64, 64-bit code is about 10% faster than 32-bit, so I switched to an x86_64 native Linux distribution, and I don't want to switch back to i386.

Re:The Price/Performance Argument Hipocracy

[GooberToo]

Actually, I had a 64-bit OpenOffice installed. It was primarily third party games (NWN, America's Army, Teamspeak, etc) which required the 32-bit environment. At the time, 64-bit Blender was also experimental with known stability issues.

Nice

[ghostbar38]

So my next X2 notebook will be cheaper!! I wait and I get better prices :)

Re:Damn, another weekend at Frys.

[jozeph78]

I'll still get the RAM at NewEgg, I DO have SOME standards.

Why would you buy RAM at Newegg, but processors at a box store? They're almost certainly going to be more expensive at Frys than they are at Newegg.

Hold on, Fry's has some great deals sometimes. I love newegg and usually favor it because of no tax in IL, but Fry's is an unusually great brick and mortar to have nearby and has deals just a great or greater than the 'egg. Now if you said CC or BB I would agree completely, but Fry's kinda kicks ass.

VIA Chipset quality

[Anonymous Coward]

Though my comment is anecdotal of my experience, I have an Athlon 64 3000+ system for about 2 years now. Runs nice in Cool n' Quiet mode. Blazingly fast with XP and games. XP-64 runs phenomenally on the Athlon-64. Had an ASUS board with a VIA chipset in it.

I had no complaint about the system in general for about 6 months. Then it started crashing. Had the mobo exchanged. Swapped the power supply, etc. But it never quite worked well. The BIOS settings would go all wonky, wouldn't keep the right settings, etc. I could only blame the chipset. I replaced the board with one on a different chipset nVidia and its worked great since.

If I upgrade to another system, I'm making sure to avoid VIA again. With ATI/AMD on board, I'm hoping that they'll become a major volume producer of their own chipsets just as ATI had been before the merger. I'd like to see VIA get pushed out -- they always seem to have Q&A problems.

Unknown circumstances

[Almahtar]

There could be circumstances we know nothing of preventing AMD from opening their drivers - an example: what if they are making use of technology licensed from another company and that company made them sign a nondisclosure? What if, over the years ATI has been developing their stuff, there were thousands of components licensed from various entities and ATI had signed contracts never to disclose some or all of them? ATI's been around for a long time, and open source really wasn't an issue way back when.

Re:Unknown circumstances

[vandan]

You license technology, sure. But you don't license an interface to a technology. So AMD may claim that they can't open-source such-and-such software that was developed by someone else, but it's ridiculous to claim that they can't release documentation on how to talk to hardware which may or may not have been developed elsewhere. No-one is asking AMD to open-source their driver; we're asking them to provide documentation necessary for the development of open-source drivers.

Re:Damn, another weekend at Frys.

[LibertineR]

Frys has a bundle right now that NewEgg cant touch. I would NEVER buy Frys RAM though.

Quad core chips coming

[Anonymous Coward]

So its no suprise that dualcores are being shuffled down to accomodate the "next big thing".

xbitlabs to the rescue

[GoatVomit]

http://www.xbitlabs.com/articles/cpu/display/dualc ore-roundup.html sums things up rather nicely.

Tell me about it 379 to 317

[Anonymous Coward]

I just bought a 6000+ AM2 dual core, after reading about it on techbargains,
i thought it was one of these rare misprint deals,,motherboard and chip for 379, i was working with the idea of getting a intel core 2 duo6700 for around 316 for the chip, el cheapo board for 50, (el cheapo works with my video and memory)

So i thought sweet, NOW the combo deal NewEgg offers is 319 for the chip motherboard, or some combo of 4 other things, you may actually want, like hard drives...

If AMD plans to do this same type of Bulk Huge price drop, then i will plan to buy intel.

Please,

[tehdaemon]

Please buy another one next week!

What a crock of shit.

[Generic Player]

AMD has always been incredibly helpful and suportive of open source. They supply full documentation for their hardware, and even donate hardware to open source projects. Its just ati that has sucked, and even then it wasn't always that way. They used to provide docs before they started trying to seriously compete with nvidia. Give AMD some time to deal with the merger before deciding how the new company will behave.

Athlon 64 3600+

[RzUpAnmsCwrds]

The real best-kept secret in the CPU world today is the X2 3600+. It's selling on Newegg for $65 right now, and while a dual-core 1.9GHz Athlon 64 isn't going to make Intel tremble, $65 is pretty darn close to Celeron price territory. Apparently the 3600+ overclocks well, too. Really well.

Re:dundant: prohibited by FCC regulations

[Anonymous Coward]

The binaries are released in that form because they are prohibited by FCC regulations from releasing anything that could be modified by the end user to violate regulatory limits.

Isn't that kind of a moot point? By releasing anything the end user can hack it. I know most people aren't comfortable jumping into machine code, but some do. Besides, HAM operators are in control of what they choose to do with the spectrum. They behave because they don't want to lose their license, but they have the power.

Kringe(r)

--

Ok I've captcha'd the 'puppies' what d'ya want me to do with 'em boss?

Man Intel still has the advantage

[aaronoaxaca]

Those price cuts just aren't cheap enough, and with the new core2 revisions later this month, too little, too late..Running a core2 6300 I got at fry's for $150 with mobo, and can't complain at all..for discount pc's the 805D still rules at 85$ at overclocks to 3.3ghz with stock cooling..I can't wait for the new AMD architecture in late July/august..

Old News... Kinda

[WhoBeDaPlaya]

The X2 3600+ could be had for $65 shipped free with a copy of Rainbox 6 : Vegas since last week. Sell the game for $15 - $20 and get the CPU for ~$45.
Pair it with a Biostar TA690G (the best overclocking 690G motherboard aside from the Sapphire PI-AM2RS690MHD) as well as some SuperTalent DDR2-667 (which seems to overclock pretty well) and you have a pretty nice setup.

Biostar TA690G at ZipZoomFly (the best deal atm w/free shipping)
2GB SuperTalent DDR2-667 at either newEgg or eWiz
If you need a cheap AM2 heatsink and some thermal paste, hit up SVC for the Arctic Cooling Alpine 64 and some Arctic Silver Ceramique

Planned obsolescence-slots.

[Anonymous Coward]

And do you remember that AMD had a slot at one time?

Re:Planned obsolescence-slots.

[plasmoidia]

And do you remember that AMD had a slot at one time? No... we try not to remember Slot A... ;-)

Ahhh.

[Filbertish]

Not even three months ago I spent $400 on my socket 939 Athalon 64 FX60 processor. I guess I couldn't really expect to know this three months ago, but the core2duo's success should have clued me in.

Re:Damn, another weekend at Frys.

[lateralus_1024]

You, my friend, have never been in a Fry's Electronics. Not only will the processor be less expensive than newegg, but they'll throw in an ECS motherboard (if you can call it that) to sweeten the deal.

Not Nice.

[wild_berry]

No price drop on the Turions. Sorry.

More than enough.

[Ihlosi]

At least before the price cuts, there was simply no way I would even consider an AMD CPU, after Intel got Core 2 Duo up and running.

So what's your usage profile ? High-end gaming or running some extremely CPU-intensive tasks where time is money ? Yup, then the C2D is for you.

But what about the mass market ? People whose CPU will mainly twiddle its thumbs (and other digits) while running web/email/office ? People who don't care if they get 100 fps or 145 ? For them, getting an Athlon makes more sense. It's more efficient while idle and more than fast enough for the usual tasks. It's fast enough for any current game (when combined with an appropriate graphics card). Why spend an extra $40 just to have "Intel inside" ?

MOD PARENT DOWN

[Anonymous Coward]

He makes outrageous claims with no sources and gets +5 Insightful?

Fanboyism is the only way to explain that.

So what you're saying is

[countvlad]

if AMD continues to drop their prices you'll...not buy from them? Err...k. If AMD plans to do this same type of Bulk Huge price drop, then i will plan to buy intel. Did you mean Intel and not AMD as you stated?

Re:Damn, another weekend at Frys.

[Ephemeriis]

You should both be glad you've actually got some way to purchase processors at a brick & mortar retailer. I'm lucky if I can pick up the right type of RAM around here... Processors, motherboards, decent sound and video cards, and empty computer cases are all out of the question. I have to order absolutely everything on-line and pay for shipping and handling. Very inconvenient when something sizzles and I just want to get back up and running fast.

It is a Microsoft Conspiracy

[Anonymous Coward]

1. Vista is Released
2. No one upgrades to Vista because the DRM and bad performance
3. Microsoft approaches AMD and Intel with a deal to make upgrading cheaper
4. AMD makes an anouncement knowing Intel will follow
5. Everyone can upgrade their hardware cheaper now and many begin to do so
6. Vista suddenly runs better on the faster hardware more vista is purchased
7. Microsoft Profits

Re:ATTN: SWITCHEURS!

[rustalot42684]

If you want to waste your money on a Mac, GTFO.
If you think that Steve Jobs is the greatest man in the universe, GTFO.
If you want to be an emo loser who cuts themselves in the shape of the Apple Logo, GTFO.
If you want to be instantly identified as someone with no computer proficiency, GTFO.

Stupid Mac users are not welcome among people who have a real life. Remember, cut down the road, not across the street.

Spell that right...

[swillden]

Also unlike the Itanium the Core 2 uses x86-32/64 ISA... just like current AMDs processors.

You mean to say the Core 2 uses the AMD64 ISA... just like current AMD processors.

Re:ATTN: SWITCHEURS!

[ifknot]

on Saturday March 24, @09:51PM you wrote: ' ...and I hate Macs* ' follow your own advice: ' *Fanboys: whether or not Macs are better is not the issue here. ' apart from the fact that statements 1 & 4 are just dumb... don't be troll fodder

Re:ATTN: SWITCHEURS!

[Anonymous Coward]

You sir, are a retard. GTFO!

Re:Damn, another weekend at Frys.

[LibertineR]

Troll? Hell, when did those loser Fry's employees figure out how to moderate?

Re:ATTN: SWITCHEURS!

[ifknot]

yeah and THE POOLS CLOSED

pretty much off topic but maybe someone here knows

[Anonymous Coward]

The processor in my machine died. It's a socket A Athlon XP 3200+ barton.

Checking out ebay to buy a used one ... damn, they're going for $150. Any ideas why prices are so high? I may as well just build a new system once I start getting into the 100s of dollars.

Re:dundant: prohibited by FCC regulations

[Martin Blank]

Ham operators behave because they don't want to lose their license, and Intel and Atheros behave because they don't want to lose their licenses. (There's no license required to operate the devices, but devices have to be certified by the FCC as conforming to regulations.)

Pah!

[darrenkopp]

so my new 5200+ is being delivered today. guess i should send it back, order the 6000+, and STILL get a credit on my account. i wonder if this marks amd releasing a quad core? i'm probably wrong, but i sure hope i'm right!

Re:Tell me about it 379 to 317

[GnuAge]

Well then I probably shouldn't tell you about the X2 AM2 6000+ retail CPU complete with K8M890-M motherboard at Arbeit Macht Fry's for $230 then?

a bit different

[snuf23]

The computer I bought (for my son incidentally) has an SLI capable motherboard oh and the 7600GT also supports dual monitors. At 10 years old my son is more interested in a box that looks like a silver spaceship than anything else. Also FedEx shipping to Hawaii is expensive - but if you try to buy parts locally you get screwed.
The system also has a three year warranty. Of course most parts have decent warranties anyway - it's just a matter of having one place to call no matter what breaks.
I've built a lot of computers of the past 15 years and now I'd just rather spend an extra $100 and buy one out right. Especially since there are boutique system builders that let you pick specific parts to go into the system. I'm not much of an advocate of the big guys like Dell except in office situations. Generally half the parts (case, motherboard and power supply for example) are useless if you want to upgrade later.

Don't be so snobby.

[ToasterMonkey]

No, I'm sure he meant what he said. It's very common to call it that.
Used here for example...
Although AMD renamed it
So did Intel