DELL goes AMD
The Register reported here. Why? DELL has been hurt enough. Its results today show massive decline on server and desktop revenue. DELL needs growth to sustain its ridiculously high 17x price/book valuation ($3.37 billion stock holder equity, $55 billion market cap).
Previously, I projected that DELL has to go AMD before end of Q2 or go BK.
A few days ago, when I saw the news that AMD pushed Turion X2 launch from May 9 to one day before Dell's earnings release today, I conjectured that there might be a correlation.
The significance of the DELL-AMD deal needs no emphasizing. As I pointed out, Intel's guerilla benchmarketing on Conroe and Woodcrest can only dupe amateurs. DELL's going AMD proves that AMD64 rules now and in the future.
With the AMD announcement, DELL's stock went up 4% after hours despite its terrible results. Intel stock fell almost 5% after hours. I expect INTC to fall below $15 in the next few weeks, as its hope of growing 18% in 2H06 is now shattered. Perception is everything. AMD's launch of Socket AM2 on May 23, 2006 and launch of Socket 1207 Opteron in June will push AMD up and Intel down more. The June 1 AMD technology conference will put the final nail on Intel's coffin.
The DELL-AMD deal will put pressure on SUN and IBM to accelerate their Opteron plans.
posted by Sharikou, Ph. D. @ 5/18/2006 01:28:00 PM
35 Comments:
Fantastic news. I guess my next laptop my be a Dell after all. :)
If Dell goes all the way AMD and hires some *talented* industrial designers, then Dell can leverage its strengths in manufacturing capabilities and its supply chain to really make some noise in the market.
And if more companies swing their votes like Dell appears to be doing, then Intel will end up having no one to buy their chips besides Apple and Isreal.
Makes sense to me. After all, Dell is getting hammered in the server segment, and AMD's sole strength is in the server segment.
"... and AMD's sole strength is in the server segment."
Not only server, but also for desktop, Intel hasn't released a chip that can compete with AMD's in terms of performance/watt (or absolute performance if you OC).
Dell's announcement proves one thing: Core 2 arch doesn't scale (to qual cores) in near future, probably up to 1 year. Or alternatively it could be a gesture to Intel in demand of more Core 2 quota allocation - maybe the yield of Core 2 family processors isn't too good.
"(or absolute performance if you OC)."
Ever try to overclock a Dell? I've been using a Dell at my work(fairly recently purchased, hyperthreading p4 cpu, 9800 graphics) and there is no way to overclock it with the BIOS. so, it's more non-overclocked performance that matters to Dell. Except maybe for their gaming computers.
"(or absolute performance if you OC)."
I have a Pentium 4, 651, oc'ed to 4.5, and after running all the benchmarks its wipping the floor with an AMD FX 60, super pi, pcmark 05, and also 3dmark 05 scores. And just so you know its running idle temp. is 32°c, load is 55-60°c on the stock cooler. You people are gonna be quick to hide when
Conroe comes out. Not a fan boy, but benchmarks don't lie.
"I have a Pentium 4, 651, oc'ed to 4.5, and after running all the benchmarks its wipping the floor with an AMD FX 60, super pi, pcmark 05, and also 3dmark 05 scores. And just so you know its running idle temp. is 32°c, load is 55-60°c on the stock cooler. You people are gonna be quick to hide when Conroe comes out. Not a fan boy, but benchmarks don't lie."
Hello Intel fanboy! Lets put your oc'ed P4 up against a oc'ed FX-60 and run those same benchmarks again. If your arguement over a processor beating another processor beating another in toy benchmarks merely because it's overclocked while the other one is at stock, that's a pretty sad fallacy for your mere opinions and breathes you being a zealot plain and simple.
As for Conroe, IT IS NOT ON THE MARKET YET, therefore your comments again are mere opinions and nothing more. Repeat this with me now:
ENGINEERING SAMPLES DO NOT REFLECT THE PERFORMANCE OF WHAT AN ACTUAL RELEASED PRODUCT WILL DO.
This goes for Conroe and for AM2 as well and I have little faith in benchmark programs, especially the SuperPi toy benchmark.(it's worthless in real world indications of a processor's performance IMHO) Get some real world results and then get back to us.
Until then, why don't you wait like the rest of us for some REAL benchmarks of an actual product on the market to compare along with an actual AM2 top of the line FX processor that has been released on the market. Then LEGITIMATE conclusions can be drawn, until then keep your fanboyism to yourself.
FWIW, If Intel does take the performance crown this time I will be the first to say congrats to them and it's about time they got thier act together. However I will say it could be short lived with K8L coming in the 1H of 2007, it's shaping up to be quite a processor and I feel it will be more than a match to Conroe, Merom and especially Woodcrest on the server front due to FSB limitations.
Oh yea! Congrats to AMD for the business win and a big "thanks for finally removing your head from your sphincter region" to Dell. Hopefully this will be the beginning of a great working relationship between them both and give the processor market that %50/%50 market share between Intel and AMD many of us have been hoping for.
*************
I have a Pentium 4, 651, oc'ed to 4.5, and after running all the benchmarks its wipping the floor with an AMD FX 60, super pi, pcmark 05, and also 3dmark 05 scores. And just so you know its running idle temp. is 32°c, load is 55-60°c on the stock cooler. You people are gonna be quick to hide when
Conroe comes out. Not a fan boy, but benchmarks don't lie.
***********
Okay. Let me get this straight. You had to overclock a 3.4ghz Intel proccessor to 4.5ghz, a 32% boost in speed, in order to compete with a 2.6ghz processor. Then intel processor needed to have a 76% raw speed advantage over the AMD processor before it was able to surpass the AMD processor in benchmarks.
Then, you compare the Prescott based processor which is built on a huge 31+ pipeline stage, at a 32% speed increase over the default speed, to Conroe, which is built on an entirely different architecture with a much shorter pipeline and a much higher IPC rate. Not even mentioning that you could not have run any multi-threaded tests that would benifit from the second core in the AMD processor, all of your tests are single threaded.
I'm sorry, but I'm not really seeing a correlation.
I'm also going to call BS on your tempature levels. A 2.6ghz Pentium4 Northwood generates 31degrees Celcius at the base of a 92mm Zalman ALCU heatsink. A 3.0ghz Pentium 630 Prescott generates 33degrees Celcius at the base of a 120mm Zalman ALCU heatsink. Note these measurements are taken by an external thermometer while the system is sitting at the Grub loading screen. Of course, as one would expect, the internal temp of the processors is much hotter. The 3.0ghz tops 60degrees celcius when forcing time-demos of Quake4 under a Ubuntu 2.6 sourced kernel. The 2.6ghz processor easily hits 50 degrees celcius under similar conditions. Overclocking the 3.0ghz to 3.45ghz on a Radeon Xpress 200 results in an internal reported temp of 70degrees celcius. I'm sorry, but I'm not buying the tempature readings given for a 3.4ghz processor at 4.5ghz.
The final line you have there is that "benchmarks don't lie." Unfortunantly for your argument, they do.
For example, the benchmarks you chose were all single threaded benchmarks, while one of the processors you appearently tried is a dual core processor meant for multi-threaded tasks.
Another example is 3Dmark and Nvidia. Remember the mess Nvidia caused when they optimized their drivers for GeforceFX in order to obtain higher scores in 3Dmark?
Another example is found in compilers. If you use the GCC Compiler, the Borland Compiler, the Intel/Metrowerks Compiler, or whatever compiler you choose to use, you're source code will behave differently. Then there is the concern of compiler targets, targeted to i386, i586, i686, K7, x86-64, PPC, or sparc. Then there is the concern of multimedia enhancments and SIMD enhancments like MMX, 3dNow, and SSE targets.
Then you have the concerns of memory latency. Sandra for one (by SiSoft) is affected by memory latencies.
Then there is the benchmark size, something that has been covered in this particular blog. Processors with larger caches can fit smaller tests inside the cache alone and provide a percieved IPC rate far above anything actually encountered in real-life applications. Processors with smaller caches that can't fit the entire benchmark inside the cache will appear to be slower in certain benchmarks, even if the real-life IPC rate is much higher.
I'm sorry, but I'm not buying the hype around Conroe.
Yes, Intel did dodge a bullet on the FSB by dropping the raw gigahertz speed of Conroe and it's derivatives. Intel also dodged a bullet by getting X86-64 targets right, Conroe is capable of running Ubuntu sources compiled to X86-64 targets.
However, keep in mind that these processors have another 5months to get to market, and get to market at pricepoints that are affordable, and in quantities that are avaiable to be purchased.
In addition you are comparing a processor that is now 5months away with AMD processors that are now 4 or 5months old. Such comparisions assume that AMD is not going to do anything between now and when Conroe launches.
Now, in my own case, having visited Intel during E3 and attended the after hours E3 party with Intel, I think I'm fairly safe in saying that the systems demonstrated bring Conroe up to current AthlonFX performance in current applications. Wether or not this will hold true in 5months is doubtful.
So no, we won't run and hide. Intel still has a lot to prove in reguards to their processors.
Umm, there are B0 revision (read produciton) Conroes in the wild now. And they completely destroy AMD product that is 5 generations ahead according to Mr. Sharikou. And I wonder what will AMD do if Intel goes IMC way? Run for the hills?
In regards to:
"Hello Intel fanboy! Lets put your oc'ed P4 up against a oc'ed FX-60 and run those same benchmarks again."
1.My post was directed at what "Edward" said...
"Intel hasn't released a chip that can compete with AMD's in terms of performance/watt (or absolute performance if you OC)."
2. I am not trying to fuel a fire here about upcoming chips, I completetly agree with what your saying about waiting for real products.
3.I am all about better products, whether its AMD and INTEL, or ATI and NVIDIA.
4.Benchmarking is a big deal for me, the only time I really push my computer is during Battle Field 2.
In regards to: "Jason Frothingham"
Please read No.1.
1.I didnt compare my 651 to Conroe.
2."I'm also going to call BS on your tempature levels."
Call it and I'll say go fish... lol
I will post those later, check back tommorrow.
3.5 months, this stement makes it seem like only a couple.
http://www.bit-tech.net/news/2006/04/20/conroe_release_date/
And fianlly, sorry for rambling on, I am not a computer expert, I build my own but thats about the extent.
***
And fianlly, sorry for rambling on, I am not a computer expert, I build my own but thats about the extent.
***
um. then why are you speaking on this?
***
Umm, there are B0 revision (read produciton) Conroes in the wild now. And they completely destroy AMD product
***
I'm sorry, but would you mind proving this, because I'm afraid I can prove it is not true. One of the demonstrations Intel ran in their E3 booth pitted identical Mobile based platforms, one using Centrino the other using a Core2 processor with identical Graphics (nvidia), identical ram, and identical support chips. The Centrino unit scored 140fps under Quake4 Time Demo, while the Conroe chip scored 180fps.
Considering that the Centrino chipset is on the market and that notebook is available right now, it shouldn't be too hard to find out which one it is, or compare it to Mobile Turion and AthlonFX systems. Based on the measurements I saw, the Conroe chip in use is just about as powerful as the most power Turion chipset in use.
The 2.6ghz Conroe chips used in the middle of the floor to power a racing game (title withheld) netted a framerate that was identical to a 2.6ghz Dual Core AthlonFX.
I'm sorry, but your statement does not hold up, at all.
***
2."I'm also going to call BS on your tempature levels."
-
Call it and I'll say go fish... lol
-
I will post those later, check back tommorrow.
***
I'm sorry, but you already admitted you are not an expert at this. I am, and the procesors I have on hand are already topping the heat range you mention. Given that the processor would physically produce more heat, your measurements can hardly be accurate. The only way they could be accurate if the processor is dipping into C-states and powering down constantly to stay inside the thermal envelope. If that is the case, you are quickly going to toast that processor.
****
3.5 months, this stement makes it seem like only a couple.
http://www.bit-tech.net/news/2006/04/20/conroe_release_date/
****
hmm, you may be correct on that. Intel timelined 6months during IDF Spring ( March 7-9, 2006 ) http://www.intel.com/idf/us/spring2006/
Given that we just passed the 2month mark, 4 to 3.5 is a more accurate launch window.
****
1.My post was directed at what "Edward" said...
"Intel hasn't released a chip that can compete with AMD's in terms of performance/watt (or absolute performance if you OC)."
***
Again, I stand by my statement. You had to overclock a 3.4ghz processor by 32%+, get a raw speed advantage in excess of 76%+ in order to have a competetive chip, without overclocking the AMD processor.
I wonder what would happen if you achieved a 32%+ raw speed overclock on the AthlonFX in order to keep the ratios in line? Would it still be competitive?
Would the processors even run for an extended length of time (72+ hours) without needing to resort to phase-change based cooling?
****
2. I am not trying to fuel a fire here about upcoming chips, I completetly agree with what your saying about waiting for real products.
****
I'm sorry, but when I read the original post that I responded to, this is not the feeling that was imparted. Rather, looking at the original post, I would say it was explicitly designed to create a fire fight.
And reading your second post:
***
And I wonder what will AMD do if Intel goes IMC way? Run for the hills?
***
I really don't see any support behind the statement you were not intending to start a fight.
as usual you forgot something, dell will only use it in HIGH END SERVERS as in 4+ sockets. i'd been an fanboy not to say that amd has 4+ socket systems covered for at least another year. so there will not be ANY laptops, desktops, workstations, or one and two socket dell servers with an amd inside for a LONG time. get lost, and who would ever take stock advice from a fanboy especially from a site? NO ONE
Intel amateurish designers haven't gotten a clue. Intel's architecture looks so primitive compared to Direct Connect Architecture. Most of the benchmarks Conroe claimed to have won are basically running off the 4MB cache. Try something more real, the Conroe thing will slow down immediately. We have discussed this quite a bit on this site already.
In response to : "Jason Frothingham"
"Umm, there are B0 revision (read produciton) Conroes in the wild now. And they completely destroy AMD product that is 5 generations ahead according to Mr. Sharikou. And I wonder what will AMD do if Intel goes IMC way? Run for the hills?"
was not posted by me.
nor this one
"as usual you forgot something, dell will only use it in HIGH END SERVERS as in 4+ sockets. i'd been an fanboy not to say that amd has 4+ socket systems covered for at least another year. so there will not be ANY laptops, desktops, workstations, or one and two socket dell servers with an amd inside for a LONG time. get lost, and who would ever take stock advice from a fanboy especially from a site? NO ONE"
So I repeat the statement:
"2. I am not trying to fuel a fire here about upcoming chips, I completetly agree with what your saying about waiting for real products."
Here is a link to the temps I took today.
http://www.enumae.com/computer-temps.html
Its alot hotter today than it was when I saw 32°C.
If im wrong I will admit it but this is what it says.
Anonymous said...
I have a Pentium 4, 651, oc'ed to 4.5, and after running all the benchmarks its wipping the floor with an AMD FX 60, super pi, pcmark 05, and also 3dmark 05 scores. And just so you know its running idle temp. is 32°c, load is 55-60°c on the stock cooler. You people are gonna be quick to hide when
Conroe comes out. Not a fan boy, but benchmarks don't lie.
How about some screenshots? That sure sounds like bullshit to me.
"Most of the benchmarks Conroe claimed to have won are basically running off the 4MB cache."
So, Quake 4 runs entirely in the L2 cache? Damn, I knew Carmack was a hell of a coder, but I never knew he was that good! And don't F.E.A.R., Half-Life 2 and 3DMark2001 (the only realistic version of 3DMark, IMO) also show Conroe in the lead, as per the benchmarks done by Anandtech, Tech Report, Bit-Tech, Xtreme Systems et al?
good news, Dell will no dought bring competative marketing to the market to regain share, Other Amds will be on its menu as well
to the moron who said he posted some screenshots
those screenshots only proved your cpu runs at 32C, but didn't prove its run on stock cooler
2630RPM? on stock cooler to cool a 4.5Ghz Northwood? i'm pretty sure its bogus
what happened with CPU fan on the second screen shot? do you mind explaining that?
"IF" dual/multi core were to take off, what kind of companies would benefit? Power management by any chance?
the benchmarks done by Anandtech, Tech Report, Bit-Tech
These are basically paid pumpers by Intel. They did no benchmark. They pushed the buttons Intel pre-arranged and wrote down the numbers. They couldn't even look at the device manager. The whole thing is disgusting, showed a total lack of basic integrity from Anand and others.
Woodcrest will outperform all other CPUs on the market
http://www.theinquirer.net/?article=31836
"These are basically paid pumpers by Intel. They did no benchmark. They pushed the buttons Intel pre-arranged and wrote down the numbers."
A) Wow, nice little piece of slander there. Care to prove your words?
B) Bit-Tech didn't use the Intel-provided F.E.A.R. demo, they used one installed on a USB enclosure from their own testing labs. And it showed much the same results as what the other sites did on the Intel demo.
C) How does Intel providing the demo machines automatically invalidate their results? There's simply no way to increase a game's framerate by 20% without killing visual quality, and if there were then everybody would be doing it! Even if they used an nVIDIA style cheat by using artificial clipping planes and other illicit optimizations, the plan would have come undone when Bit-Tech used a different demo.
D) If between now and next year, AMD provide benchmarks of K8L beating Conroe by the same amount as Conroe seemingly beats K8, will you refuse to accept those as well?
The company plans to offer more details about its next generation architecture at the beginning of next month, and then will reveal still more as the date of the project’s release grows closer.
AMD said the new architecture would be available in the first half of 2007.
“We do have multiple generations under design now,” Moore said.
Chuck Moore was hired to head AMD’s next generation architecture development, beyond the generation that he offered a preview of during his keynote address on Tuesday.
Advanced Micro Devices is working to build its engineering muscle.
Five design sites – two in Austin, one in Sunnyvale, one in Boston and one in India, said Chuck Moore
Moore estimated that its headcount has doubled in the last few years, just in terms of silicon engineers. And the number of software engineers has also increased, he said.
source:reed-electronics.com
"Anonymous said...
to the moron who said he posted some screenshots
those screenshots only proved your cpu runs at 32C, but didn't prove its run on stock cooler
2630RPM? on stock cooler to cool a 4.5Ghz Northwood? i'm pretty sure its bogus"
1. The 651 is 3.4Ghz 65nm CedarMill proc. not a northwood, hence the 1 at the end, check online and then tell me im wrong.
"what happened with CPU fan on the second screen shot? do you mind explaining that? "
2. I am not sure, I will take some pics tomorrow, untill then i guess im just a moron.
ps: how old are you 12?
Wow, nice little piece of slander there. Care to prove your words?
Check out numerous articles I wrote during the IDF regarding Anand the paid pumper.
I really don't get it. Discussion what is faster Conroe or Athlon has no sense at all. Athlon IS the fastest now. Conroe WILL be the fastest soon. By 4%, maybe 10% , whatever...Who cares?
Buying Conroe means no HT and suspicious 64-bit. Athlon is a proven technology - it works.
Conroe is an experiment. Cool one. 65nm one. AMD must get there asap with K8L dual-core design and 65 chip. And then lower all prices...
"Anonymous said...
to the moron who said he posted some screenshots
those screenshots only proved your cpu runs at 32C, but didn't prove its run on stock cooler
2630RPM? on stock cooler to cool a 4.5Ghz Northwood? i'm pretty sure its bogus"
Updated images, stop whining.
http://enumae.com/computer-temps.html
"Paid pumper"? Isn't that a job in the pornography industry?
So the question is who to believe: some benchmarks that were perhaps rigged by Intel...err..."optimized for the Intel experience" or whatever, or some raving AMD fanboi with a blog who says that these new Conroe chips wouldn't outperform a PCjr. (they are after all 23 generations behind AMD) even though he's never seen, smelled, touched or benchmarked one. I think the only rational choice is clear--neither.
But here's some food for thought--don't you think it's possible that Dell is only using Opteron on 4 socket servers because they've actually gotten their hands on some of the early Conroe protos and have found them to be fully competitive with the Opteron in 1 and 2 socket systems? Certainly a possibility I'd say, given that Dell has been pretty adamant that this Opteron deal is only for 4 socket systems and that they are comfortable with the rest of Intel's roadmap. I think it's pretty well established Opteron will hold the 4+ socket crown for quite some time; even Intel has admitted the 4 socket space has been "challenging" for them, which certainly has to be a nominee for understatement of the year, microprocessor category.
to enumae:
lolz.. you still haven't answered my first question. where is your so called "stock" cooler? pictures?
i see absolutely no resemblence of your "651" statement and your 4.5GHz @ 32C. care to elaborate that?
ps. before you make any statement, be sure to back them up with substantial evidence, or you're no better than a 12yr old.
Answering Intel fanboy (because the rest is AMD of course). ...
But here's some food for thought--don't you think it's possible that Dell is only using Opteron on 4 socket servers because they've actually gotten their hands on some of the early Conroe protos
The problem is that Dell maybe is confident about that, but for last 3 years their customers demand Opterons and Athlons. Because they want AMD. So DELL as a customer advocacy should deliver them.
That's it. Customer choice and satisfaction. No benchmkarks and better knowledge.
I can't believe people are still questioning the superiority that the Core 2 will demonstrate when it hits the market shortly.
Woodcrest is coming out next month, Conroe a month later and Merom a little after that.
All benchmarks people have run on what is now B0 stepping (Core 2 is due to release on B0 or B1 stepping) have shown it to best AMD in every way.
It even overclocks like no tomorrow (50% on stock hsf?).
There's no longer any doubt that Core 2 is a winner. It's only a matter of when AMD will release K8L (a 65nm die-shrink of the current K8 core might not be enough - we'll see) and how Intel responds.
Take some time to browse the forums on www.xtremesystems.com and you'll see many examples of Core 2, some ES versions; some regular.
Intel have nothing to gain by falsely claiming success with their new architecture, they'd soon get found out. Everything since IDF has proven that Core is a good performer.
No, it's real and it's coming - soon.
I for one can't wait - next year will be even better with K8L coming too!
Let battle commence!
This is my original post...
"I have a Pentium 4, 651, oc'ed to 4.5, ...running idle temp. is 32°c, load is 55-60°c on the stock cooler..."
Here is link to the Pictures you all have wanted. To the person still asking for them, you should have been able to see them last night... try refreshing the page. :)
http://enumae.com/computer-temps.html
ps: I am trying to back it up.
"There's no longer any doubt that Core 2 is a winner."
That is precisely the way Intel wanted you to think. And remember Intel today is a marketing company at heart (unlike 10 years ago when it was a manufacturing company).
How many "independent tests" on Conroe have you seen? Aside from IDF's benchmarketing, I know only VictorWang (ps. CPU-Z doesn't even report his Vcore right). I'd really like to see it from others, on more real-world apps.
Then there are people touting Conroe's architecture superiority. Like the Ars Technica article, which is full of marketing language and only fools amateurs. Conroe is just a one issue-wider Yonah which is a DC P-M with large shared L2 and better memory prefetch. (Ok, probably with some other "minor" modifications.)
As for overclocking, it's nothing if you get one or two chips that overclock; not like Athlon FX where 80+% overclocks well.
Also be sure that Conroe's availability will be terribly limited. Intel haven't ramped up Yonah yet. What Intel's been doing is clearly a trick to drive up demand and price - in the end it doesn't matter how much those chips were officially priced if few is available to the public (there's a Economics term for that but I forgot it).
So Conroe might be a winner on several things, such as super pi or futuremarks. But there's still lots of doubt on it being a better choice for overall price/performance - we simply cannot tell before it is out and available.
About "K8L" vs Core V2 :
"The advantage it has over NGMA is the ability to do 2x128b LDs."
David Kanter
(dkanter@realworldtech.com)
"It will probably be able to start two different DP MUL's
or two different DP ADD's each cycle. according to AMD
patent Pat. No. 6,944,744.
Theoretically they could use the 3rd scheduler also which
now only handles loads/stores/transforms."
Hans de Vries
(hansdevries@chip-architect.com)
K8L to slaughter the new upcoming Conroe CPU's by well over 50% & well over by 100% in multitasking -clock for clock.
K8L Spec's
Quad-Core, w/ Shared L3 Cache for all 4 Cores
HyperTransport 3.0 - Providing 40GB/s Bandwidth per Link running at 5.2GHz+ (16x16)
Improved CrossBar and Optimizations for 65nm SOI and beyond
Support for DDR2, DDR3, and FB-DIMM's (End-User has choice of what to use)
Native support for DDR2-800 & DDR2-1066
4-6 Issue Wide Core
1/2 Cycle SSE processing (2 per Cycle) (The slide left me confused a little, correct me if I am wrong)
Additional SSE Instructions and registers
Power Consumptions of 50w or less
Co-Processing Support
AMD's Pure Virtualisation (Vista Ready)
etc.
Post a Comment
<< Home