Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Interesting process article at xbitlabs
Paul
http://www.xbitlabs.com/articles/editorial/display/tech-process.html
from ptanner on SI, re: Sun
Sounds like there's no reason to believe that Sun's use of Intel in the new systems means that they've dropped the plans for Opteron, but that it's in the future, while the Intel system was already in the pipeline.
Paul
I am not sure that yesterday's Sun announcement was really news wrt AMD. This older article noted they planned an Intel expansion to their entry level servers while at the same time...
http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=18961681
gollem, chipguy does have a point. It's one thing to post pertinent excerpts from an article with a link, it's another thing entirely to post the entire article, regardless whether you include the link or not. If you post the entire article, there's no motivation for the reader to follow the link, thereby allowing the source site to satisfy its advertisers. It's far better to post a "teaser" excerpt, and let the reader get the rest from the source. They have to make a living, too.
Paul
belgiangenius, what's with the s*xual/an*l fixations? You're starting to sound a lot like vaneesbe from RB. How about some constructive input?
Paul
dougSF30, That SK8N is an Asus board.
wbmw, interesting that they didn't try the DVD playback using "maximum battery" mode. If you'll note the results from this old Anandtech chipset DVD playback test, you'll see that DVD playback shouldn't be all that stressful on these CPUs. Yes I know these are desktop scores, but they are on Duron 800s, with 100MHz FSB, after all. If you toss out the dog VIA chipsets, the worst CPU utilization is the Aladdin TNT2 chipset at 51-61%. I seriously doubt the mobile chipsets Tom's tested were less efficient power-wise than these desktop chipsets. Too bad Tom's didn't bother to try max battery or even tell us what the CPU utilizations were.
Paul
http://images.anandtech.com/reviews/video/Integrated0100/DVD.gif
Tenchu, I'm not sure, but regardless, that(model#/MHz comparison) is the reality.
Paul
Interesting new microscope.
Paul
http://www.physlink.com/News/Index.cfm?ID=42
sgolds - Well, I got my response. Not good. Anyone have the e-mail addresses for de Ruiz and Meyer? I have posted the full e-mail text, as well as my response to Ruth Cotten's response.
Paul
Ruth,
This is unacceptable. Deny it all you want, the fact is that anyone who really matters, whether it's the end user, the OEMs or the hardware reviewers and press, compare the AMD chips by model number to the numerically equivalent(in frequency) Pentium 4. To deny this is to deny reality and will ensure the discrediting and demise of the model number system. I certainly hope this will be corrected with ensuing releases. The current degradation of the model number system does not at all bode well for AMD. Many are watching now, willing to give AMD the benefit of the doubt on this release because the model number has been accurate and conservative in the past, but if this degradation continues, all support for the model number system will be lost.
I would sincerely appreciate it if you would forward this e-mail to the appropriate corporate executives, since the website is not very friendly as far as contact information. They need to know how AMD is being perceived in the real world, not in the minds of AMD's marketing staff.
Paul Flynn
AMD Stockholder
ruth.cotter@amd.com wrote:
>Hi Paul,
>
>All of the benchmarks that AMD used to both arrive at and substantiate the decision to launch this processor at a 3200+ model number rating have gone through extensive technical examination as well as legal approval from our legal benchmarking expert.
>
>Additionally, AMD has always reiterated that our model numbers and increases in those numbers refer to the performance within the AMD Athlon XP family of processors and are strictly NOT comparable in any way to megahertz or to the products offered by the competition.
>
>Thank you,
>INvestor Relations
>
>-----Original Message-----
>From: jjjpflynn@netscape.net [mailto:jjjpflynn@netscape.net]
>Sent: Wednesday, May 14, 2003 9:55 AM
>To: Relations, Investor
>Subject: 3200+ Quantispeed rating
>
>
>Dear Sir or Madam,
> I have been a supporter of the decision to institute the Quantispeed rating system, knowing the reality of the situation with actual performance vs. actual Megahertz that exists with Intel's Pentium 4, and AMD's XP model's far greater performance per clock cycle. However, the release of the latest model, the 3200+(as well as, to a lesser degree, the 3000+), is unacceptable. They ought to have been clocked at least 100MHz higher, or rated lower.
> The only way the Quantispeed(QS) system can survive as a legitimate rating
>is if it is a very conservative rating, leaving no real room for criticism in comparison to similar Pentium 4 frequency. This is not the case with the last 2 releases, but particularly the 3200+. There are far too many "benchmarks", which are commonly accepted, in which the 3200+ is beaten by the 3.0C and the 3.06 GHz Pentium 4s.
> Obviously, there are some benchmarks which will always be the victor's domain for the Pentium 4, but what is important for the QS system is that on the great majority of common benchmarks, the QS rated AMD processor is at least as powerful as the frequency equivalent Intel processor. This must be true for not only the benchmarks you use for establishing the rating, but also on most of the ones used by review sites and considered by those in the PC press who have influence. Otherwise the QS sytem will fall into disrepute, and you will have wasted the respect for your implementation of the QS system that had been established. If this happens, with the huge disparity in frequency between the XP and the Pentium 4, you can't expect the average PC buyer to know enough about PC components to understand that the lower frequency XP performs like a much higher clocked Pentium 4. They will assume the Pentium 4 is faster. This would be disastrous for AMD.
>
>Paul Flynn
>AMD stockholder
wbmw, I imagine they're offering the spread their customers want. With sales depressed, I'd say they'll take whatever sales they can get, so if it's a matter of selling $40 CPUs to a customer or not selling to the customer at all when you have capacity unsold...
There isn't a lot of call for the high end right now, without some sea change event to drive it. Perhaps like Doom3 and Opteron/Athlon64?
Paul
wbmw- re: QHz. A little dated, but shows the process.
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/Benchmarking_Methodology2_v2...
sgolds - Thanks, will do. None so far. eom
wbmw, re: availability
3 pages of pricewatch, eh? That's about 30 or 36 vendors, right? How many of those are high volume vendors that are likely to have a couple thousand chips, and how many are low volume vendors that probably have a hundred or less? If each vendor has 500 chips on average, then AMD only shipped 15,000 processors into their distribution channel.
A lot of distributors, both large and small, do not report to pricewatch. Don't base your numbers on that assumption.
Paul
Anyone play "America's Army"? It'll be interesting to see what effect the 88 Opteron 44node SCI/AMD/Racksaver cluster will have on their gameplay.
Paul
http://www.prnewswire.com/cgi-bin/stories.pl?ACCT=104&STORY=/www/story/05-13-2003/0001945872&...
sgolds, I agree. I wrote AMD about this. Sorry about the funky indentation or lack thereof, the copy/paste from webmail didn't quite fly.
Paul
Dear Sir or Madam,
I have been a supporter of the decision to institute the Quantispeed rating system, knowing the reality of the situation with actual performance vs. actual Megahertz that exists with Intel's Pentium 4, and AMD's XP model's far greater performance per clock cycle. However, the release of the latest model, the 3200+(as well as, to a lesser degree, the 3000+), is unacceptable. They ought to have been clocked at least 100MHz higher, or rated lower.
The only way the Quantispeed(QS) system can survive as a legitimate rating
is if it is a very conservative rating, leaving no real room for criticism in comparison to similar Pentium 4 frequency. This is not the case with the last 2 releases, but particularly the 3200+. There are far too many "benchmarks", which are commonly accepted, in which the 3200+ is beaten by the 3.0C and the 3.06 GHz Pentium 4s.
Obviously, there are some benchmarks which will always be the victor's domain for the Pentium 4, but what is important for the QS system is that on the great majority of common benchmarks, the QS rated AMD processor is at least as powerful as the frequency equivalent Intel processor. This must be true for not only the benchmarks you use for establishing the rating, but also on most of the ones used by review sites and considered by those in the PC press who have influence. Otherwise the QS sytem will fall into disrepute, and you will have wasted the respect for your implementation of the QS system that had been established. If this happens, with the huge disparity in frequency between the XP and the Pentium 4, you can't expect the average PC buyer to know enough about PC components to understand that the lower frequency XP performs like a much higher clocked Pentium 4. They will assume the Pentium 4 is faster. This would be disastrous for AMD.
Paul Flynn
AMD stockholder
kpf, yb, what do you mean rumors? nVidia has already specifically announced nForce3 for Opteron. It's way past rumor stage, it was demo'd at AMDZone's LAN party a week or two ago, and I imagine it could be again at E3.
http://www.nvidia.com/view.asp?PAGE=nforce3
OT: fyodor, re: SI post on chipsets -
I wouldn't count ALi totally out yet, although they're dying as a chipset provider, they are still providing southbridges for use with ATI mobile northbridges. They are actually bigger in DVD chips now than in chipsets, if I recall correctly.
Paul
no, yb, you may have misinterpreted slightly. The 84.7 is for the whole family, but we don't know what the final clock speed is, do we?
" I am confirming that this is for a maximum number for the entire
family. (including future higher clocked versions)"
This is what we suspected, but it's good to hear anyhow.
Paul
Tenchu, I imagine the truth is somewhere between the two extremes you presented. I'm sure folks out in the field are somewhat concerned, but not by any means pulling their hair out.
Paul
EP & wbmw, y'all know niceguy isn't here, he couldn't restrain his overblown optimism enough to remain under the radar. If you're so interested, he's been posting regularly again on SI Momderated AMD.
Paul
wbmw, DEC/Compaq isn't PC class EV6 hardware and Intel bus isn't the same as AMD's EV6 bus. Just because Intel's bus is "easy" to implement on PC class hardware doesn't mean that EV6, which wasn't designed for 4layer boards, should be as easy. It's just taken them(OEMs) a little longer to get it right.
wbmw, no different than downbinning.
wbmw, I wouldn't count on that. Several of the existing nForce2 boards are showing up with later revision northbridges, which while not labelled as Ultra 400, are having no problem going way over 200. 220-235 is not uncommon. The XPs haven't been the problem anyhow, even older ones with the 133/266 FSB have been hitting those speeds. nVidia seems to have it nailed now, and on a .15u process now, so it'll likely gain some headroom as new steppings of the Ultra 400 come out.
Paul
good point re: A64 memory speeds at xbitlabs.
They say Athlon64 will support PC3200, which will allow clock to be an even multiple of HT links, so the frequency will always be 200MHz based, rather than some lower frequency, as in the case of PC2700. Wonder if Opteron will support PC3200 soon, and if that will give us a more accurate picture of its scaling factors by eliminating the variations in RAM speed.
Paul
http://www.xbitlabs.com/news/cpu/display/20030510180321.html
CJ, LOL! Wouldn't Area51 be a lot more cost effective? I mean it's just a little down the road from Sunnyvale, not halfway around the world! ;)
CJ, I remember that, but have we ever heard what actually came of it? I remember some speculation about CVD SOI or something like that, but never heard anything official afterward.
Paul
No paper launch this time. Monarch computers has 3200+ in stock.
http://www.monarchcomputer.com/Merchant2/merchant.mv?Screen=PROD&Store_Code=M&Product_Code=1...
Keith, I believe the "conversations" took place before the relationship was publicly clarified, and sgolds was theorizing that they were closer than many were willing to believe at the time.
Edit: I see now that sgolds already answered far more eloquently than I.
Keith, I think he was talking about the increasingly tight relationship between IBM and AMD.
Paul
kpf, even more than working on the production process, they need to work on their memory performance - as I recall, it is atrocious!
Paul
Congrats, Tenchu!
kpf, ATI already has one(IGP320?); the problem is that its performance stinks. It uses an ALi southbridge, if I recall correctly. We need nVidia to stir up some activity here. If they can come up with a reasonably power-efficient high performance chipset / video system, a lot more folks would replace their desktops with one.
Paul
from SI. new thin film SOI Isonics product Sigma-1. So, did IBM or AMD or both order it jointly?
http://www.siliconinvestor.com/stocktalk/msg.gsp?msgid=18929102
Edit: Oops, I guess neither, unless Fishkill has either 100mm or 150mm setup. Does AMD at the Sunnyvale(?)test fab?
wbmw, I suspect you're right re: surviving chipset players.
kpf, 4. Lets face it, currently AMD does not have a lot of volume in mobile space. One reason, likely the main reason, for this is the lousy chipsets available for Athlon. If nVidia is smart enough, they could snatch a lot of market share from Intel based systems. If they have enough production resources. They're making nForce3 Pro on 0.15u, so maybe they'll do it after they've established that chipset.
Paul
Has anyone heard anything about the Newisys 4300's performance yet?
Haddock, yes, I found the article quite odd as well. But even with all the the inaccuracies and bogus negative commentary, I think it still reflected well on Opteron.
Paul
Minor good news. BSD AMD64 support soon.
http://www.amdzone.com/#2
Congrats! Although that A64 system sure would have been nice, eh?
Paul
econo, glad to hear something other than the black cloud mentality here. Your scenario in the last paragraph has the ring of truth to it, to me at least. Unlike '99/2000 though, OEMs of all sorts know the possibilities and that survival of(even prospering despite) Intel's wrath is possible and worth it. And in its own way, Opteron has the same sort of "enthusiast" support (in different circles) that Athlon had early on. I think the odds for success are better than those for failure.
Paul