If some of you have some spare money by the end of this month to buy a new computer and have been waiting for some comparison between AMD's 45nm Shanghai against Intel's new Nehalem, well wait no more: Link (Thanks to Axel).
The good news is that PhenomII has shown significant improvement over Phenom. But then this shouldn't be a surprise considering Phenom was a step back from Athlon64. Two years and a process node later, AMD finally defeats Intel's Q6600. If some call this as AMD being back in the game, so be it. All I wish is for some of our friends to return and have something to say. Their silence is deafening.
153 comments:
some of you might have already noticed. this is in response to our old friend's newest blog. It acts as a place holder to allow some of you to make a comment.
Geez - it's been so long since Sci posted anything new that I rarely get over there to check.
His main beef seems to be Anandtech once again, and he mentions rumors "flying around" about how AT has already tested Deneb against i7 and are now having a hard time spinning the results in Intel's favor.
Wonder why AMD doesn't send him a test system over at AMDZone?? :)
From the discussion on the previous post:
"I suppose Cray did not bother looking at Nehalem performance and we should defer to your logic that Cray simply based their decision on a short term delay and a bug which was fixed on the next stepping."
"This deal doesn't go down until 2011 or 2012. Are you really so naive that you think that Cray entered an agreement for 3-4 years out based on the TLB bug?"
These points reminded me that many times, enthusiats drastically overstate the impact that harware review sites have on the decisions made at many industries. I've worked in IT for more than a dozen years and dealt with numerous IT professionals over that time. I'm also the go-to guy whenever a co-worker wants advice on a new computer or on upgrades or when they want some tech support questions answered. Having dealt with, literally hundreds of people, I can count the number of people who ever mentioned hardware websites on, well... one finger.
Joe average doesn't know AMD from Intel from ATI from NVIDIA from IBM from a hole in the ground. They typically diferentiate by company, based on marketing. So they know about Apple, and they know about Dell, and from that point all they know is Best Buy and Circuit City. They never ask me about AMD CPUs versus Intel CPUs. If they want to know anything at all, it's whether or not they should get a Mac instead of a PC. That's it.
Joe Professional knows about hardware websites, but would not trust one as far as he could throw it when it comes to judging hardware and hardware platforms. Most of these guys are in cutthroat industries where making a bad choice can cost them their jobs, or cost their companies money and clients. I hear IT techs talking about how they are trying out a new system or new hardware or new software. None of them, to a man, has ever mentioned that they think something is good because they read it on Toms or Anands or Tech Report or whatever.
Cray doesn't rely on some hobbyist who works with some buddies and occasionally has to edit his charts and reviews because he forgot to rename a file, or because he didn't double-check his settings. They don't rely on Fudzilla to determine whether or not they should sign that contract with AMD. Dreamworks has hundreds of millions of dollars at stake on their technology, and they're not making decisions based on the latest review at HardOCP. I really wouldn't want to be the guy at Dreamworks who tells his CIO that they signed with Intel because "they gave us some free CPUs."
By the way, the Phenom II CPUs performed pretty well on their CAD/3D/rendering, to the extent that their lower prices make them a very nice option for 3D modeling and rendering. For the aspiring 3D animator, who wants to create demo work on the cheap, they're tough to resist.
Nothing new here!
If you watch Shanghai marketing, they keep talking about performance per something which really in itself is a statement that they do not have a performace claim at this point. So, benchmarking the new AMD Shanghai should be read for how much it has a perfomance deficit.
wow, Roborat64, you able to find put that he posted within few hours :)
Blogger Tonus said...
By the way, the Phenom II CPUs performed pretty well on their CAD/3D/rendering, to the extent that their lower prices make them a very nice option for 3D modeling and rendering. For the aspiring 3D animator, who wants to create demo work on the cheap, they're tough to resist.
are you referring to the benchmark here http://www.hwbox.gr/showthread.php?t=3253 ?
It seems to lose out to the Kentsfield though ...
I’ll have to agree with TONUS, PII did very well. Now, if this were Barcelona back in 2007, well, the present day market landscape would have been entirely different, to say the least. Hence, I go back to my earlier post. The ATI purchase was a distraction AMD could not afford financially and strategically at the time. While they had they had some advantage; this indeed would have been a threat/ serious alternative to Kentsfield.
This first shot evaluation is significant for the main stream, and who knows, with a DDR3 part they just might make a real impact. However, conspicuously absent are the XE top bins. Obviously, the benchmark landscape does change a bit.
I said it before, they’ll have to get past a QX9770 (the only one with 1600 FSB) to be truly competitive. They didn’t.
In closing, credit must be given where credit is due. The AMD engineers did a great job in turning this thing around, albeit -12 to 18 months to late. To bad they weren’t running the company.
SPARKS
"are you referring to the benchmark here http://www.hwbox.gr/showthread.php?t=3253 ?"
Okay, that confused me for a moment, I was wondering why there were two sets of benchmarks showing differing results. The one Rob linked on the front page is the tests run at stock speeds, the one you linked is with all of the CPUs running at 3.7GHz.
The Phenom II 3.0GHz in the 'stock' benchmarks was second only to the i7 940 (and by a very small margin, at that) and outperformed all of the Intel CPUs in the UGS benchmark, while finishing last in the other tests, but not by a large amount. I still think that for an animator on a budget, it shows much promise.
More junk from Sharikou's brother. Ohh, sorry, I meant Roborat.
...I really wouldn't want to be the guy at Dreamworks who tells his CIO that they signed with Intel because "they gave us some free CPUs."
LOL, everyone knows that Deamworks went with intel this time because they got cpus + software optimizations for free:
InformationWeek
Under the alliance, Intel will provide advanced processors with multiple cores and adapt the technology, which is not yet available in the mainstream market, to DreamWorks' authoring tools. In essence, DreamWorks has agreed to convert its computing infrastructure for 3-D film creation to an Intel-based system.
In conclusion, they got a better deal because intel has all the money in t, and adding insult to injury, they have to optimize all of DreamWorks rendering software to take advantage of Penryn+HyperThreading (sorry, I meant Nehalem) :)
And this is a second source just in case you guys believe the 1st. one was "AMD biased" (only you guys can come up with such nonsense, lol):
SeekingAlpha
As part of that deal, Intel is providing advanced processors and (eventually) multi-core chips specially adapted by Intel engineers for animation and 3D rendering. DreamWorks, in turn, is converting its entire computing infrastructure to Intel-based environments.
By the way, the Phenom II CPUs performed pretty well on their CAD/3D/rendering, to the extent that their lower prices make them a very nice option for 3D modeling and rendering. For the aspiring 3D animator, who wants to create demo work on the cheap, they're tough to resist.
For me the question is how a 4P system with an AMD CPU will compare to a 4P Westmere in a year or two using a linpac benchmark. I'll be looking for a machine to run CFD models and think something along those lines will probably be the way to go for a personal system. I may have to scrimp and settle for a 2P system, but one can always hope.
Love the way you added this as if it were part of the article
In conclusion, they got a better deal because intel has all the money in t, and adding insult to injury, they have to optimize all of DreamWorks rendering software to take advantage of Penryn+HyperThreading (sorry, I meant Nehalem) :)
which was your commentary, not the authors.
So in short, Intel made them a better offer based on the ability to offer services that AMD couldn't offer. Do you really think they chose to go with an inferior platform that would take longer to do the work they wanted done? I don't. And there was nothing in your links that indicated that the system they were getting would be in any way inferior to what AMD offered.
“In conclusion, they got a better deal because intel has all the money in t, and adding insult to injury, they have to optimize all of DreamWorks rendering software to take advantage of Penryn+HyperThreading (sorry, I meant Nehalem) :)”
WTF is the color of the sun on your planet, and what rock did you crawl out from under?
Ok, let’s put this into perspective. I have an F1 chassis. Two engine manufactures are competing for my endorsements. One manufacturer SAYS he’s working on a killer engine, the other say’s he’s got them up and running and the dyno figures are off the charts.
The second guy is giving me FREE engines to tune to my chassis, AND a bunch of engineers to assist in the modification to optimize all that horsepower. He’ll help tune the package and even help with future development.
The first guys engines a plagued with delays and manufacturing issues. In fact, the sample engines don’t come half as close in power and reliability, and with high RPM loads they blow up. I’ve waited nearly 9 months for my first guy to come with something viable.
Then the shit hits the fan. I’m taking big losses, my competition is winning races, I’m losing races, the owners are not happy, and they want a change----NOW!
I get the first samples. I can’t believe the power and scalability of the second guy engines. His engineers are bending over backwards to help, and in a year or two I will a tremendous edge on the competition if I jump on this deal early.
Listen, junior, wake up and smell the coffee. INTC made CRAY and deal they couldn’t refuse. It was either take the sweetheart deal or wait for AMD and all its bullshit and die.
If ya gotta better product and support, people will change an entire company. That’s exactly what CRAY did.
It’s called business. Do you get it yet?
SPARKS
To the anonymous Dreamworks analyst:
I take it you assume "provide" means "provide for free"? I don't read it that way; if someone said Intel provides >80% of the marketplace that would not mean free. If someone said AMAT provides tools for Intel; does that mean AMAT is giving Intel free tools? The problem is you are assuming 'provided' means free. And when you assume you make an ASS out of U and... well just you....
Do you have a link where it indicates Intel IS GIVING AWAY FREE CHIPS to Dreamworks, or just the links that say Intel is 'providing them with chips"? You ignore ALL of the other links as to the reasons why the CEO and a president at Dreamworks give for making the switch and we are to believe they are simply lying and you are telling the truth?
Time to put up or shutup - provide a link where it says Intel is providing chips FOR FREE. Clearly the SW support is/was a key but you are reading into the word "provides".
I also love how you have completely dropped the Cray stuff (embarrassed by having your ridiculous "logic" exposed?)...
"LOL, everyone knows that Deamworks went with intel this time because they got cpus + software optimizations for free:"
I think that in this case, "everyone" means "people who don't understand what they are reading." Particularly, people who ignore terms such as "strategic alliance" and don't recognize how statements like the following fit into that:
"Under the alliance, Intel will provide advanced processors with multiple cores and adapt the technology, which is not yet available in the mainstream market, to DreamWorks' authoring tools. In essence, DreamWorks has agreed to convert its computing infrastructure for 3-D film creation to an Intel-based system.
In addition to working with DreamWorks, Intel plans to develop and promote the development of 3-D content on its platform in other markets, such as home theater, personal computers, videogames, online virtual worlds, and mobile devices."
Intel and Dreamworks have entered into an agreement where Intel will develop features in their future CPUs that work hand in hand with new 3D technologies being developed with Dreamworks, not just for them. The processors being provided are more likely engineering samples and chips in development, which serve as a test platform for the companies to develop the technology in question.
If someone wants to paint this as a case of Intel beating out AMD by waving free shinies in Dreamworks' faces, then I can see where the statements are being misunderstood or misinterpreted. But it requires a bit of a departure from reality to assume that Dreamworks pushed everything aside for a few free CPUs. I can imagine the conversation now:
DW rep: well, AMD made this offer and their products are so good, we're leaning in that direction.
Intel rep: well, how about if we give you the CPUs and software for free? Will you still consider AMD?
DW rep: consider who?
Intel rep: AMD. I'm saying that if we offer-
DW rep: consider WHO?
Intel rep: AM... oh okay, hahaha, I get it. Just sign on the dotted line. Bobby, give the man his bag of free Nehalems.
DW rep: Nice doing business with you!
Dreamworks went with Intel because Intel has the wherewithal (money, people, resources) to help them achieve their goal- to create an optimized system for developing 3D content. Don't think for a second that they didn't talk with AMD, or that AMD would have hesitated to offer CPU technology and software tools (for all of their financial problems, the cost of some processors and software development would be a drop in the bucket compared to the PR boost and potential for future revenue). Dreamworks went with Intel because they made the better offer, and Intel agreed because they get something out of it as well, software and hardware technology optimized for 3D development that they can license to others.
But if it makes you feel better to think that they just walked into Dreamworks with a bag of CPUs and walked out with a contract, go right ahead. Just don't assume that "everyone" feels the same way.
To meet the increased demands of creating 3-D animated feature films, Intel will provide DreamWorks Animation with the latest high-performance processing technologies, including future chips with multiple processing cores.
by reading the press release (I have no way associate to this project), it does mean DreamWorks has access to Intel Engineering sample and platform (particularly the Nehalem) for their early development work. Free or not free, i do not know. However, this is no difference from any support that intel would have to give to OEM/ODM, Apple/Sun and a like. Engineering samples for development are likely to be free (not sure about board). But, they would have to purchase for the production units.
And onto the 'main topic' .. I am still 'waiting' for comment to be appear at Sci's blog, which coincidentally with the title "Still Waiting" :)
may be Mister anonymous Dreamworks analyst can help him there? haahha
As I pointed out in the previous thread, there's a poster on AMDZone who works in the Hollywood rendering industry who pretty much covered the Dreamworks deal last August, and he had the same conclusions that we do - Nehalem systems show a large performance boost over the old Barcie systems which is why he was personally evaluating Nehalem systems for his company and personal use. Of course this didn't sit too well with the rest of AMDZone and so he got labeled as an Intel fanboy.
Anonymous A Nonny Moose said...
As I pointed out in the previous thread, there's a poster on AMDZone who works in the Hollywood rendering industry who pretty much covered the Dreamworks deal last August, and he had the same conclusions that we do - Nehalem systems show a large performance boost over the old Barcie systems
Nope, he actually referred to Yorkfield system (his company bought Apply system) being much better that other system available at that time and now consider the NHM system which is better.
Scentia is still waiting, meanwhile my 30 month old dual core will probably still wipe the floor with his Phenom II in most applications.
I've had 30 months of use out of it too.
"Scentia is still waiting, meanwhile my 30 month old dual core will probably still wipe the floor with his Phenom II in most applications."
You said it.
I started out with Q6600 (GO). I bought it for “shits and giggles” in July 2007. I built my wife a media machine and overclocked it to 3 Gig on the second boot. I was blown away by the performance. Then, I immediately under clocked it; she was so happy with her new machine. That weekend, when she went shopping, I swapped out the Q6600 with a 955EE out of my machine. She never got wise to it!
After that taste, I bought the QX9770 as soon as it was released, and there was no looking back.
Talking a bunch of crap is one thing living and working with these monsters is quite another.
Dementia is just an arm chair guy who is a self proclaimed computer expert with no practical experience, and no one to talk to. He couldn’t shine ROBO’s shoes.
“The sound of silence is deafening”
Jan 8th may just shut him up for another 5 months.
SPARKS
Heh, Johan stumbled upon a wasp's nest and the result was quite predictable. :)
Heh, Johan stumbled upon a wasp's nest and the result was quite predictable. :)
At least he admitted receiving intel advertising dollars and at the same time over-inflating intel's "unreleased" nehalem server CPUs benchmark numbers by a good margin. Shame on them.
At least he admitted receiving intel advertising dollars and at the same time over-inflating intel's "unreleased" nehalem server CPUs benchmark numbers by a good margin. Shame on them.
And that's exactly the attitude on display over at AMDZone that has earned them a reputation as a bunch of fanboy idiots.
Of course Johan admitted that Anandtech receives Intel advertising dollars - why wouldn't he when there's Intel ads all over the main page? He'd look pretty stupid if he alleged otherwise when the evidence is as obvious as your nose.
As for the "overinflating" - I see no admission from Johan that he did any such thing, in fact he outright denies it. He made a lot of sense whereas Assmountie and AhBenSToopid did not. This is why nobody wants anything to do with AMDZone - they're a useless pimple on the arse end of rabid fanbois and not even AMD respects them enough to send them free samples of anything.
/rant
"At least he admitted receiving intel advertising dollars"
He admitted that the site receives advertising dollars from Intel, which is obvious from the ads on the site. A quick look also reveals that they receive advertising dollars from NVIDIA, Corsair, NewEgg, OCZ, and Asus, at the very least.
This makes them... pretty much like any other publication that deals with the computer industry. If they have a hidden deal with Intel to promote their products, then you may have a smoking gun. Otherwise, the idea that getting ad dollars automatically makes them a shill strikes me as silly.
"and at the same time over-inflating intel's "unreleased" nehalem server CPUs benchmark numbers by a good margin. Shame on them."
He explained that the difference was negligible and within the margin of error. Anyone who would try to make an issue over numbers that fall within the margin of error do not understand statistics and should not make accusations lest they mark themselves as ignorant.
And anyone who refers to numbers that fall within the margin of error as being overstated "by a good margin" has no place calling someone else biased. You really have no idea what you're talking about, from Cray to Dreamworks to Johan's testing, do you?
However, I must say it was amusing to see Johan criticized for the memory he used in the AMD tests, implying that he was giving AMD the shaft... and then he points out that the memory kit is the one provided by AMD as part of the review system.
LOL.
He explained that the difference was negligible and within the margin of error.
LOL. If you call 5-10% "negligible" then there's nothing else I have to say here.
Hope you guys can be a Happy Intel family.
LOL. :D
However, I must say it was amusing to see Johan criticized for the memory he used in the AMD tests, implying that he was giving AMD the shaft... and then he points out that the memory kit is the one provided by AMD as part of the review system.
If he was professional enough, notihng would have stopped him from using buffered DDR2-800MHz kits to see how much does latency affects Shanghai. I bet you if it was intel, he would have a whole bunch of different memory kits with different latencies. Remember that this is not the first time Anand has been caught cheating.
In conclusion, I do feel really sad for Johan: working in a biased environment where everybody watches you to post pro-intel articles and not having the "true freedom" of posting the thruth!!
"LOL. If you call 5-10% "negligible" then there's nothing else I have to say here."
Johan called a 2% difference negligible and a 2-5% difference within the margin of error. It was the 'Zoner' who claimed upwards of 18% difference because he made a couple of incorrect assumptions. Much as you did, imagine that!
"I bet you if it was intel, he would have a whole bunch of different memory kits with different latencies."
And if the ones not provided by Intel turned out to be better, you would be mocking the company for providing substandard memory in a system submitted for review. If Shanghai or Barcelona would indeed benefit more from something other than what they themselves submitted for review, then you feel as if it is the reviewers job to correct AMD's screw-up?
LOL, indeed!
Tonus, as long as I have been on this site I can honestly say I have never detected any bias on your part, be it AMD or INTC. You've been fair, objective, and always the gentleman. I’m sure others will bare me out on this.
And, this is coming from a guy (me) who wouldn’t piss on an AMD chip. I make Andy Grove look like a friggin’ AMD cheerleader.
With all due respect, please ignore ‘Dreambo’. He’s trolling, flaming, and really not worth your intellect which, by my estimation, is profound and objective.
SPARKS
On a completely different topic, for those that want a peak into the world of yield improvement, I'd recommend that you take a look at this article.
You have to ignore the fact it is a self serving advertisement written as an article, but there are some interesting details in there.
Or maybe it's just my geek engineer nature showing through. :)
In conclusion, I do feel really sad for Johan: working in a biased environment where everybody watches you to post pro-intel articles and not having the "true freedom" of posting the thruth!!
I don't. No one is holding a gun to his head. If he weren't happy he could write for someone else.
And for the record, I hate conspiracy theories. If all that effort were put into something valuable, we'd all be better off.
Sparks - you'll love this - from our fave fanboy site AMDZone:
Foundry Company leader Hector Ruiz is on a list of 10 hispanic candidates to replace Bill Richardson who is dropping out after looks a corrupt contracts in New Mexico while he was governor. Really Ruiz's name is on a list made up by the fine folks at LULAC.
I guess he could do on a national scale what he has done for Motorola and now AMD :)
“Sparks - you'll love this –“
Well, there ya go. If many of you recall, I once passed a comment, which on the surface may have been interpreted as bigoted. I made a reference to Wrector’s Latin charm, poor boy, immigrant roots, success story, playing to a wide, largely understanding, sympathetic European audience.
In the Interest of discloser I am of Italian decent. (You know the type, Guinea, WOP, Day Go, Mafioso, etc.)
This short list of “qualified” replacements to Richardson not only supports my original premise, but it also supports another. Political qualifications are not determined by private sector successes alone (if you can find any), at the very least.
They are capitalizing on Wrectors Latin charm, poor boy, immigrant roots, success story, playing to a wide, largely understanding, sympathetic ‘AREA SPECIFIC AMERICAN’ audience”.
Hey, what ever works for him. Besides, performance wise, he’ll be in great company in Washington D.C.
Further, just imagine the help he could get for AMD on a Federal level!
(There ya go, GURU. “Welcome to my world”)
SPARKS
Yep, I'm thinking why wasn't JLo on the list? :) She sings, she dances and is a hell of a lot better looking than Hector. As for understanding commerce, she probably does a better job of it than Hector has shown...
Moose, at least he’s more qualified than, say, Timothy Robbins or Susan Sarandon. %D
Ahh,----- I think?
SPARKS
Lets see, Moose. JLO, Paula Abdul, hmmm, I'm starting to get a fix on the type of girl you fancy, the Latin Lovers, aye?
SPARKS
Well Obama just nominated a career politician, who has absolutely no experience in intelligence, to head the central intelligence agency. Change we can believe in?
You have a comedian stealing a Senatorial election in Minnesota, where 25 precincts had more votes then people who signed up to vote as many votes were double counted for some odd reason (brings me back to the good ole days in Chicago - vote early and vote often!)
And then you have a certain someone (Caroline Kennedy) who'll be appointed to replace Clinton in the NY Senate... she seems obviously qualified after touring the state for a couple of weeks (I hope she had GPS, because I doubt she's visited any part of the state outside of NYC)... oh I forgot she also headed a couple of charities... she hadn't voted for something like 16 years - yet suddenly she's interested in politics?
So when I hear Ruiz is on a list of candidates in New Mexico, call me jaded, but it really doesn't surprise me.
"a career politician, who has absolutely no experience in intelligence"
I can't figure out if this a contradiction in terms, the exception, or the rule. Ya got my head spinning on that statement!
SPARKS
"Caroline Kennedy"
What more harm could she do that hasn’t already been done? At least she's not a carpet bagger. Who knows, Hillary’s 'gift' to the Kennedy family may have its rewards. She's young, idealistic, untainted (for a career politician), may grow with job and rise to the occasion.
I’d like to see it. I could think of worse.
SPARKS
It's done. As if they thought it wouldn't happen.
http://www.tomshardware.com/forum/256785-28-clear-spinoff
SPARKS
Sparks - you'll actually love this... I just made a few grand off AMD stock (buying in the low 2's) - but don't worry some of it will go toward my next Intel build!
I would not buy/own this stock going into earnings, but if it drops near $2 again, it may be worth another flier. Barring a short term BK (which I don't see) the mere noise on this stock (a $0.20 move is ~10%) makes it an interesting trading vehicle if you have the cajones... (of course a -0.20 move is a quick 10% loss....)
And it is still not clear to me the logic of NY state giving 1.2Bil to a company (AMD foundry) that is chartered offshore (to avoid US taxes) and is majority owned outside of the US. (of course the issue here is that I'm assuming the NY pols are using logic). Would NY state give money to TSMC to build a fab in NY? Renasas? Toshiba? Samsung? It's not clear to me how the current situation is any different (except that there is a minority stake in the company by a US company?).
"Sparks - you'll actually love this... I just made a few grand off AMD stock (buying in the low 2's) -"
I remember. I knew/said it was a ballzy move at the time you made the purchase. I gotta give ya ctedit, ya got gillhonnies.
I saw the price last night at 2.78, after hours it climbed to 2.90. I knew someone, somewhere had to pull that trigger, it was you all along. -25% in month! KUDOS.
SPARKS
"He explained that the difference was negligible and within the margin of error. Anyone who would try to make an issue over numbers that fall within the margin of error do not understand statistics and should not make accusations lest they mark themselves as ignorant."
Ignorant is probably weak in this case... stupid is a better term.
Sparks: Lets see, Moose. JLO, Paula Abdul, hmmm, I'm starting to get a fix on the type of girl you fancy, the Latin Lovers, aye?
Heh, OK you outed me! However my wife is Vietnamese. Guess I go for that "exotic" look :).
I see that Intel has posted updated guidance and revenue down to 8.2B for Q4. Interestingly enough, AMD's stock which as been climbing since last week, is down by about the same percentage as Intel's. Now I wonder what will happen to Intel's stock when AMD comes out with the Q4 report in about 2 weeks.
Now I wonder what will happen to Intel's stock when AMD comes out with the Q4 report in about 2 weeks.
AMD had their 4th quarter guidance after Intel, and if I recall correctly it was much worse than Intel at the time (Intel at the time had projected down ~10%, and I think AMD said something like 20-25%). With the guidance today, Intel is revenue will be down ~20%, so I think it is inline with AMD's early warning.
The market is just punishing AMD along with Intel (and forgot AMD had already warned by more than Intel originally?). Unless the market tanks between now and earnings both will probably trade flat or even drift upward a bit upward into earnings. (I'm kind of hoping for a nice drop again toward 2 so I can put on another trade, but I'm not sure thats's happening unless the market drops significantly)
The whole market's nuts.
Yesterday, INTC was up 63 cents. After today’s guidance warning, they're down 93.
Timing that AMD move from the low 2's to yesterdays Gov. Approval @ 2.73, was the about the best move anyone could make in this hunkered in, nose to the ground, bear market.
AMD shed 12 cents today alone, and is down 4 more after hours.
The bottom line, money ain't moving, especially with the semi’s.
Those crazy bastards on Wall Street are positively manic. My buddy tells me it’s like trading nitroglycerin.
SPARKS
Just found one review on Deneb:
http://www.extremetech.com/article2/0,2845,2338338,00.asp
Just quote part of the conclusion below, the rest, read it yourself .. and more will be coming tonight :)
AMD has finally achieved performance parity with competition in its price range. Or, you could simply say that AMD had to price Phenom II appropriately. We didn't even throw in benchmarks from the Core 2 Quad 9650, which is at clock speed parity with Phenom II; it would have been no contest. And the two higher-end Core i7 would have simply run away with all the benchmarks.
ops, actually a lot reviews are already there .. reading now :)
Below are the what I compile from various side below on the OC on air achieved, I have to put semi colon to make it readable and if you wanna port that to excel :)
Cooler Master; 3500/3600; 1.425
Scythe Ninja; 3500; 1.425
Unknown; 3500; 1.488
Arctic Freezer 64; 3500; Unknown
Retail; 3900; 1.52
Unknown; 3755; 1.55
OCZ Vendetta 2 heatpipe; 3800; 1.5
first 2 results are from the same site
http://techreport.com/articles.x/16147/13
http://techreport.com/articles.x/16147/13
http://www.bit-tech.net/hardware/2009/01/08/amd-phenom-ii-x4-940-and-920-review/14
http://www.extremetech.com/article2/0,2845,2338342,00.asp
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3492&p=10
http://www.overclockersclub.com/reviews/phenomii940/4.htm
http://www.guru3d.com/article/amd-phenom-ii-x4-920-and-940-review-test/6
Anand seems to have a very good chip in his hand :)
Basically the OC is quite good, but some may disappointed after having high expectation due to the earlier hype
I laughed at this comment in the HardOCP review:
"I think that the enthusiast is wondering exactly what I am wondering. “How does this Phenom II do clock-per-clock compared to Intel’s Core 2 and Core i7?” Now while some of you will whine loudly, “That’s not fair! That flagship Core i7 is $1000!” Well you people just need to shut the hell up and go have a drink. You are already unreasonable, so a little alcohol won’t hurt you in the least."
To say the least, they were not impressed with Phenom II...
Tonus, I love you.
I just finished reading Anand’s Pheromone II review. MAN, he put a positive spin on this thing. Dementia’s and Suck-in-a-kook’s Fanboi base must be getting to him with phrases like, “The AMD we used to know and love is back”.
Noticeably absent are the large INTC ad’s (for this particular review, anyway), with only a small blue box in the upper left hand corner of the page.
Then there’s this gem, it sounds like Anand had either a lawyer or a diplomat phrase this one.
“Similarly, the Phenom II X4 920 is generally better than or equal to Intel's Core 2 Quad Q9300, and priced less (it's more of a competitor to the slower Q8300, but I didn't have one available for testing). The same stipulations detailed above exist here as well; there are some areas where Intel is going to be faster but for the most part our tests showed the Phenom II to be a better option.”
And………
“Competition is back. Let's get to it.”
What I’d like to know is back from where to where?
[H}Enthusiast had no such agenda. They gave this unvarnished review with the bark on. This is a real review with no caveats, QX9770, i965, top bin vs. top bin.
I am sick to death with performance per dollar and price per watt horseshit!! Hey if ya can’t afford ‘em, TFB!
http://enthusiast.hardocp.com/
SPARKS
Phenom II isn't a bad buy today for a mainstream machine, but it's price competitiveness is completely at Intel's mercy. With Core i7 920 at <$300, it's only logical that the Core 2 Quads are due for major price drops. Rumor has it that at least part of those drops are coming later this month.
See, in general the $300 Core i7 920 (2.66 GHz) easily outperforms the $550 Q9650 (3.0 GHz). Unless the i7 ramp is being held back for some reason, it makes sense for the Q9650 to come down way below $300, perhaps $250. With Intel's entire line of quads falling below that. Well, I'm not sure that Intel have enough inventory of i7s to support that pricing structure just yet.
And what's funny on the PhII big day, AMD stock down 4.5%, Intel's stock up by a penny :)
So what happened to all the PhII oc's to 4GHz easily on air BS?? From Pointer's post, it seems more like 3.6-3.7 with good cooling, and I wonder what the power draw is?
I read Tom's review earlier and it was about the same as Anand's - even though it seems to finally outperform the Q6600, it lags behind the better Yorkies and can't hold a candle to the i7, including the gaming tests that some of the pre-NDA-lift sites were showing a big improvement with the PhII. Anyway, Tom's gave it their rare "preferred" or recommended rating which should make a bunch of AMDZonerz finally happy :)
At HardOCP they were able to OC one to 3.8GHz on water, but had to increase the voltage to 1.65v, which they felt was not advisable for long-term use.
I'm interested in what the fallout is going to be. Phenom II runs better than Phenom I, and the price is attractive compared to Core i7 on a platform basis (ie, mobo + RAM included). But I see people talking about the die size and that the pricing is probably going to kill them, especially if Intel drops the pricing on the Q9550/Q9650. Will Intel pull the trigger on that, considering that they just lowered their Q4 expectations a second time? Or does that not enter into the equation?
What are the feelings of the process guys here, regarding the die size of Deneb? The guys at Aces forum are cringing over it, including the ones who don't lean overtly in Intel's favor.
What are the feelings of the process guys here, regarding the die size of Deneb?
I didn't read the Ace language, but I don't see what the big deal is. On 45nm I think it is ~20% smaller than the 65nm Phenom I... sure that's not the 30-40% you'd expect from a straight shrink, but it does have an extra 4MB cache instead. Though with AMD still using SiON for the gate the continued large size may continue to have some binning problems (esp at the edge) - it's hard to say as there is no public info on what kind of process variation there is on their 45nm process. I wouldn't tend to think it is that big a deal.
Another item from the AT review:
The take away point is that compared to Penryn, Phenom II is slower clock-for-clock. The gap grows with Nehalem; Phenom II only gets close in older game engines, while the rest of the time Nehalem is 30-60% faster at the same clock speed.
So all those fanbois claiming i7 wasn't good for games clearly needed to get some modern game engines :).
Axel:
“Core 2 Quads are due for major price drops. Rumor has it that at least part of those drops are coming later this month.”
Tonus:
“especially if Intel drops the pricing on the Q9550/Q9650. Will Intel pull the trigger on that, considering that they just lowered their Q4 expectations a second time? Or does that not enter into the equation?”
Just a couple of things worth considering…
If INTC wants to boost sales they’ll drop prices.
If INTC wants to clear inventories they’ll drop prices
If INTC wants to ramp i7 they’ll drop prices.
A number of chips have already been scheduled for EOL.
Then there’s a Digitimes report that said the cuts will come on Jan 18th. (The link now is dead without a subscription)
In fact, I read somewhere, the INQ I think, where INTC is being accused of stuffing the channel.
The cuts were already set in stone last year
Let’s see:
No Atom equivalent: Strike One
Pheromone II barely competitive with a Q9450: Strike Two
When are Xeon 55xx’s being released? Strike……
http://www.electronista.com/articles/08/11/13/xeon.i7.in.early.2009/
SPARKS
Hey All,
Well it's Phenom II day... for all it's worth. I found the HardOCP review to be interesting because they actually OC'd the Phenom II 940 to 3.2Ghz and did a clock-for-clock comparison to the Core 2's and i7's at the same clock speed. I think the Phenom eked out a narrow win over the Core 2 in 1 or two benches, but even the Core 2 blows it away in a whole bunch of others.
Of course, doing clock-for-clock comparisons is now "unfair" and "fanboy" because AMD no longer wins those comparisons, but I'm sure some heretics around here would be interested :-)
And now for my spiel:
1. According to the review websites, Phenom II has about 758 million transistors... roughly 27 million MORE transistors than the i7, while at the same time the i7 is whacking the Phenom II at performance... so much for the fanboys who say Intel can only win by just throwing more resources at the problem.
2. Remember that piece of crap the P4? Remember how Intel didn't really improve the architecture but just tried to move to smaller lithography and slap on extra cache and ramp up the clockspeed? Well I guarantee AMD remembers because that is exactly what they did with Phenom II with the same results... it's faster, but it's not blowing anything away.
3. Right now Intel doesn't have to do a single stitch of innovation to put AMD back on the ropes. All it has to do is go to the accounting department and drop the Q9550 down to just under the price of the Phenom II 940. A couple of CPAs at Intel can roll back everything AMD has been working on for years.
4. AMD's marketing department is full of dillweeds for making the Phenom model numbers match up to the i7 model numbers... they are showing that they do not believe in their own product and need to intentionally confuse it with Intel's.... grow a pair and think up your own name assholes.
Spark said...
Let’s see:
No Atom equivalent: Strike One
Pheromone II barely competitive with a Q9450: Strike Two
When are Xeon 55xx’s being released? Strike……
It's obvious what AMD doesn't have, and that is a top bin processor. So they are trying to give more bang for the buck. Which puts them at Intel's mercy.
But Intel still has some weaknesses.
First, perhaps surprisingly, is Atom. It rules the roost right now, but the ARM processors are beginning to enter the netbook space at CES. Intel needs to get the Menlow platform out the door, but it is still 6-9 months away.
Then there is graphics. As much as it pains me to agree with Scientia, he has a point when he says that Intel's integrated graphics are weak. And Intel knows it. That is why they are in the process of accelerating the cadence of their graphics development. This could give AMD an opportunity in the short term to exploit that weakness in the niche between netbooks and high end notebooks with their Yukon platform.
Intel is also having issues with their transition to a "platform" company. The Centrino platform was late by 4-6 weeks. While this didn't cost them any significant market share it did mean that their OEM customers missed the back to school schedule. Intel still has some things to learn on how to bring the whole platform together on schedule. Putting together a strong platform seems to be something AMD does fairly well.
And finally, there is Nvidia's CUDA and AMD's stream processing initiatives. These are on the market right now with the first rev of Larabee still a year away.
While Intel has answers in the works to all of these threats, the window of opportunity is open for the competition. So there is a lot more going on right now than just watching Shanghai catch up to C2D.
If Intel doesn't execute and possibly even find a way to accelerate some of these initiatives, they could find themselves losing mindshare.
INTEL got a few holes in its armour, but come on after seeing how they recovered from their Pentium IV and Itanium disaster do people really think that their sorry ass integrated graphics is a hole they can't close.
If you look back Intel's current technology, product portfolio and bank account couldn't be more insurmoutable and AMDs more weak.
Its easy for those in the peanut gallery and especially the fabnois in green like Scientia to spin things but get real. AMD is done as a long term thread, only poor decision and miss execution by INTEl can change the game.
Let the tick tock continue
“Intel is also having issues with their transition to a "platform" company…………..Intel still has some things to learn on how to bring the whole platform together on schedule. Putting together a strong platform seems to be something AMD does fairly well.”
When? When they dropped the ball with the i740 years back? Actually, INTC is and has always been the largest supplier of graphics solutions. Further, I always thought they WERE a platform company.
My wife’s machine, a Gigabyte GA-G33M-DS2R (Q6600) sports a GMA3100 graphics solution. It was cheap ($110), clocks like a bastard (3 gig), “for shits and giggles”, with a 300W powersupply! She’s never complained.
My daughters machine, Asus P4P800 MX (P4 3.2 478 w/HT) also has an IGP. That thing is as old as the hills and it runs just great.
Both these solutions have worked fine for years for their applications. Aren’t these one board “platform” solutions? Hasn’t INTC been doing platforms for years?
ITK, I don’t wish to sound trite or facetious, but what am I missing here? Perhaps, an IGP that can swing Crysis @ 30 FPS that no eleven year old, or a work at home mom will ever need, and never use?
It cost AMD 5.4B, 65 % of the company, and near bankruptcy to do the same. I believe Centrino, or any branded name for that mater, is some kind of marketing gimmick designed to attract uneducated buyers into something “special”.
Of course AMD can put “together a strong platform”. They had better. It cost them enough.
Sorry, INTC invented the Integrated Platform Solution years ago, and is still the leader.
SPARKS
Further, I always thought they WERE a platform company.
Not according to Paul Otellini. I've seen numerous interviews with him where he talks about converting Intel into a platform company.
You can say he is blowing smoke if you want to, but he is responsible for the centrino platform which was an integral part necessary for the growth of wireless hot-spots. Sure, it is a chicken and egg thing, but centrino and selling Wi-Fi have made Intel a lot of money. They are hoping to do the same thing again with WiMax.
So I see platforms as a way of increasing your share of the silicon inside any given device. And that means more cash in your pocket and less in someone else's. Which is the name of the game after all.
As to Intel having issues with their platforms, I did give an example. They were late on the Centrino2 launch and that did cost their customers money. Not a good thing.
I believe they have also slipped on their originally projected release date for mid-range Nehalems. And we know that can't be a CPU issue, because the CPU is out there for the high end machines. So if there is a delay, it must be in some other part of the platform.
And regarding graphics, again, I'll let the market place speak for me. Apple dumped Intel graphics and went with an Nvidia chipset in the Macbook. If Intel had a superior solution, there is no way Nvidia gets that sale.
That is why Intel is doubling the pace of their graphics development. They saw it coming, but can't close the gap fast enough to prevent this from happening.
Again, it is about taking the money on the table and putting it in your pocket so the other guy comes up empty. Intel's graphics aren't good enough to ensure that happens every time.
What are the feelings of the process guys here, regarding the die size of Deneb?
Off the cuff, I would speculate that Shanghai has gotten AMD off of the process cliff that they were on with Barcelona. (I base this assumption on the ability to overclock the chip.)
So if they are in the same ballpark they were in before for defect density (which they should be) and their parametrics are better (which the OCs would seem to indicate), the resulting yield improvement probably covers the price drop.
Forrest... Trees... It's amazing how many folks ("enthusiasts") have such a weak grasp of the overall market.
IGP in the past (and I'd argue in the present) is all about "good enough". How many people buy retail notebooks, looking to do heavy gaming with just an IGP? How many CORPORATE notebooks are purchased based on gaming capability on IGP's? If you work at a company which provides a work laptop, do you think they pay attention to whether you can run Crysis (and how many FPS)? For the small minority that may have a business need for good graphics, do you think they use a simple IGP solution in those cases? If/when netbooks get a little more capability I can see the corporate market shifting rapidly to netbook solutions to keep costs down.
How many CORPORATE desktops are purchased based on graphics ability? And of those buying based on this, how many do you think rely on an IGP solution as opposed to a discrete card?
Let's face it, the only reason AMD is spending such time talking about IGP's is that is the only area they can talk about these days.
Intel's roadmap is clear... a SOC solution for atom which will put any low end, downclocked single/dual core out of reach in terms of power consumption and probably cost/price (and thus force AMD to commit resources to develop a specific solution if they truly want to compete) and just keep pushing Si process tech and CPU capability while keeping a good enough platform.
Intel is, and has been, running a marathon, driving R&D, manufacturing tech and driving toward a finish line based on profitability and being accountable to stockholders. AMD seems to sprint out every once in awhile, flap their arms about "yelling look at me I'm in the lead" (K8, ATI acquistion, now IGP performance) only to realize by sprinting out they have caused themselves to cramp up ((taking on the Dell biz, mountains of debt from ATI acquisition) which causes them to fall back quickly.
I believe they have also slipped on their originally projected release date for mid-range Nehalems. And we know that can't be a CPU issue, because the CPU is out there for the high end machines. So if there is a delay, it must be in some other part of the platform.
This could be the case - with a different socket perhaps there are issues. The other possibility is the decision was market, and not technology, driven. Exactly what are the mainstream Nehalems going to be competing with (I'm assuming these will be sub $250 chips)? In addition to competing with AMD, they will in all likelihood competing directly with Core2 solutions which are smaller (and thus cheaper for Intel to produce) and use less power. Until Intel retires Core2, why would they really want mainstream Nehalems from a business perspective?
Anyone hear much about the Nehalem roadmap for the notebook space? For markets with a single socket solution with no "performance is king" needs, Core 2 is probably a better economic solution for Intel. I would not be surprised if there were no real mainstream Nehalems until 32nm or unless it is a integrate GPU/CPU product which serves a specific need.
ITK, good points all.
As far as midrange Nehalem’s are concerned, this delay (I’m guessing) was due a number of factors.
Midrange C2D’s are very strong (as we saw just today).
INTC is clearing inventories (in cooperation with partners, memory makers, motherboard makers, etc.)
There’s no competition. They can ramp i7 at their leisure.
They have a 12~18 month technological lead.
Souring economy.
You may have a point with Centrino, the boys at INTC didn’t take it too seriously the first time around, referring to it as Latrino. That’s changed, however. I’ll concede the Centrino2 launch.
Nvidia, you say? Laptops? Nope.
You, GURU, and Orthogonal once explained to me that chips will peel apart layer by layer if not processed correctly. Ya can’t layer marble and chauk. This is precisely what happen to NVDA when they rushed development of their laptop graphics chips. They’re falling apart and melting at the seams, costing them tens of millions. Things are not so rosy on that side of the street.
Finally, there is no doubt INTC is doubling their efforts on the graphics front. However I’m not underestimating their potential here. Not with 10B in cash and a 35B a year revenue stream. It’s enough to scare the bejesus out of NVDA’s CEO. He must have a reason for all that whining and pissing. After all, it did blow up in his face, and I think we all might be in for a surprise. (That anonymous poster seems confident enough.)
I won’t concede the IGP point, especially for the business sector, bread and butter, machines. INTC’s CPU’s and chipsets will carry their graphics, not the other way around. You guys are always hounding me about price performance ratios. Ok, I’m a power whore, but my wife and daughters are plain Jane, Barbie, cut and paste gals.
They’re happy and so am I. I’ve got a feeling a lot of IT pro’s are going to be happy, too, especially when this market turns around. The CPU is still King.
"Forrest... Trees... It's amazing how many folks ("enthusiasts") have such a weak grasp of the overall market."
Ouch!
SPARKS
I agree with ITK and Guru. Yes, Intel's integrated graphics are awful. And yes, few people are buying a notebook, or even a desktop system, with even mid-range graphics capability in mind.
The company I work for purchases at least a couple dozen computers a year, sometimes quite a few more. Nearly all of them include integrated graphics. I have yet to have a person ask why their graphics are so bad, or so slow. The Intel IGP does the job for them because they're not concerned with graphics performance.
If you're in an industry where graphics performance is key, you are avoiding Intel integrated graphics like the plague. You're also avoiding any other kind of integrated graphics. Your desktops will contain an ATI or NVIDIA card, your notebooks will too.
PS- Spore requires a fair amount of processing muscle, both CPU and video. My feeling is that a $699 notebook might display Spore beautifully, but I hope you aren't expecting to be able to play it on anything bigger than a 10" screen unless you turn off most of the eye candy. Anyone who buys a low end notebook to play Spore is going to feel pretty cheated as they watch that pretty slide show on their screen.
Allow me to help 'defending' Intel IGP a bit:)
Let's over simplified the consumer market segments (read computing needs) here:
1) Desktop
a) Enthusiast/Mid-High mainstream - likely to have discrete GFX
b) Mid-low - might have GFX if they do some modern games
c) Entry - likely IGP only.
d) Business - depends on graphics need, normal office would be sufficient on IGP
I see no much issue with Desktop.
Notebook
1) High - will have GFX
2) Mainstream - might or might not
3) Entry - No No
4) long battery time, etc
I do see issue with certain mainstream, long battery time and other needs, which require sufficient 3D graphic power while not as power hog with discrete GFX.
So, by not having a better IGP as compare to ATI/Nvidia is not a really a doom for intel, but it do create holes in which certain computing need can't be fulfilled by intel platform and hence less competitive in that market.
side note is that Intel north bridge has quite some other logic supporting AMT/Vpro, hence less Si are/power budget for the other functionality.
I do see issue with certain mainstream, long battery time and other needs, which require sufficient 3D graphic power while not as power hog with discrete GFX
I don't think anyone (and not me) is arguing this isn't a need/segment - but exactly how big a segment do you think this is? 5%?
People (esp enthusiasts) focus on gaming, but I don't see that as a decisive factor in the notebook market and in the desktop market the ability to use even cheap discrete graphics to serve this demand is pretty straightforward. And for folks serious about gaming and an extra FPS and eye candy, they will invest in discrete graphics.
While it is nice to have this in the form of an IGP solution is the the tail (IGP) going to wag the dog (computer sales)? Maybe for a small segment, but anyone who thinks this will have a MAJOR impact (like some other blogs) needs to share the stuff they are drinking/smoking.
First, let me clarify my position on Intel's IGP. I understand the concept of "good enough" graphics. I also understand that most laptops are currently business machines. And my employer not only doesn't care if my laptop can play games, he probably prefers that it doesn't. I also happen to think that these are valid arguments. All I'm saying here is what Pointer did a much better job of stating. Intel's IGP leaves some holes for the competition to exploit.
I've also seen it claimed here that the average, non-gaming user will be happy enough with Intel's IGP, and he probably will. But if he were to be shown Scientia's comparison of stills from Spore, which system do you think the average user will want? He will go for the one that appears to be better, regardless of his usage model and actual needs if the price is comparable. And integrated graphics aren't expensive, so the pricing concern is a wash. These are the kind of holes I'm talking about. So my contention is that Intel needs to be in the ballpark with the competitors IGP solutions, not clearly lagging behind.
Further, based on my reading, I would say that Intel sees graphics becoming an increasingly important part of the computing experience (their visual computing initiative comes to mind). If that is indeed the case, then roughly matching your competitors solution becomes increasingly important. If Intel didn't see this as a potential vulnerability, do you think they would have gone to the expense of doubling their graphics cadence for the next few years? I'm certain that it is not a cheap decision, so Intel must also see an issue there.
I think that Intel will continue to execute and will close this gap (be it real or perceived) but for the time being, there is a chink in the armor.
Incidentally, the economic arguments regarding a delay in the mid-range Nehalem launch are good ones. But doesn't that open another can of worms? The AMD faithful are often heard proclaiming that we need AMD to keep Intel from slowing the pace of technological growth. If Intel is indeed sitting on the mid-range Nehalem launch because they don't need the product to be competitive, doesn't that support the AMD argument?
If Intel is indeed sitting on the mid-range Nehalem launch because they don't need the product to be competitive, doesn't that support the AMD argument?
Nope, AMD is not the reason for the delay. That's all I can say :)
The Israel FAB rears its ugly head again, I'll wager.
They should have built the S.O.B. here away from the goddamned radical lunatics firing rockets from GAZA.
The Abdul's are really helping AMD now. Heh, Arab Micro Devices, found a new way to compete with INTC, while they get a billion+ from the NYS taxpayers.
You can't make this shit up!
I think I'll play some C.O.D. 4 after this post.
SPARKS
Oh yeah, you're not giving up nothing. INTC stock price fluctuated, in direct proportion to the attacks.
SPARKS
ITK: "I've also seen it claimed here that the average, non-gaming user will be happy enough with Intel's IGP, and he probably will. But if he were to be shown Scientia's comparison of stills from Spore, which system do you think the average user will want? He will go for the one that appears to be better, regardless of his usage model and actual needs if the price is comparable."
While I agree with the point you are making, this example may not be the best one, for the reason I explained before. SPORE can bring even a decent desktop gaming system to its knees due to its graphics demands (even on my 3GHz Core2Duo with ATI 4870, it recommended I go with moderate settings!). The person who buys a notebook based on those screenshots and tries to play SPORE is going to feel as if he's been scammed.
This is aside from the point you are making, that if a person can get better graphics for the same cost, then they will go for the better graphics. And yes, I think Intel feels that this is an area where they need to invest and develop a solution that doesn't suck, hence the spending on Larabee development.
As for sandbagging, I tend to feel the same way, that without a competitor to push you, you don't need to scramble to rush out technology, and can pace your product releases more conservatively. Of course, I think we're at that stage now, with AMD and its supporters apparently very pleased that their best desktop CPU is competing with Intel's mid-range offerings.
But if he were to be shown Scientia's comparison of stills from Spore, which system do you think the average user will want? He will go for the one that appears to be better, regardless of his usage model and actual needs if the price is comparable. And integrated graphics aren't expensive, so the pricing concern is a wash.
With respect there is a giant hole in this logic. It is not a matter of a user simply choosing between IGP's, they have to choose between the complete bundles (CPU, GPU, wirless to a lesser extent). So it is not just a matter of cost of the 2 different IGP solutions or the whole notebook for that matter. You have battery life, OEM support, and maybe CPU performance (?) to name a few things.
You make it sound as though if you get equivalent pricepoints, then it is just a question of good graphics or so-so graphics; but you fail to mention that choice of graphics also ties into battery life, CPU performance, etc...
Of course a user would choose the better graphics, if all other things were equal.... but all other things are not equal and that is the fundamental problem with using IGP as some sort of key determiner of market acceptance.
The key question is, is the difference between IGP's enough to force customers (or more importantly businesses) to change a decision. I submit if it is the amount of eye candy in Spore or extra FPS on some 3D games, that is way down the list in terms of making a deicsion (ESPECIALLY in the lucrative corporate market)
With respect there is a giant hole in this logic. It is not a matter of a user simply choosing between IGP's, they have to choose between the complete bundles (CPU, GPU, wirless to a lesser extent).
I will concede that there is more to this than just integrated graphics. I have focused the discussion on the end user, and I was wrong. The real discussion should be from the OEM's POV since we are talking laptops.
OEM's are who Intel needs to get the design wins with. And if Nvidia or ATI can put together an attractive chipset, then they will get the design win. They both have made chipsets for Intel CPU's and I expect they will continue to do so. And part of the equation to get that design win is the IGP. So while I agree with you that there is more to the equation, IGP is still a factor.
All this goes back to the platform emphasis. While Intel has put together some nice platforms, they still have some issues bringing it all together in a timely fashion.
While I agree with the point you are making, this example may not be the best one, for the reason I explained before. SPORE can bring even a decent desktop gaming system to its knees....
Which could go a long way to explain why Scientia used stills rather than FPS or some other metric. And you'll note that I did specify using stills. ;-)
Guys,
Here’s AMD Dragon ‘Platform’ the Dell XPS 625. (Why the hell they’re calling it a platform, I’ll never know. It’s still a marketing gimmick as far as I concerned.)
Prices start at $999 (no monitor) finishing at a cool $1749.
This, dear fellows, is a dead end machine, EOL’d while in production, as DDR3 which is scheduled to arrive within 6 months will supersede this AMT2+ derivative. (With negligable performance gains,~10%)
http://www.dell.com/content/products/productdetails.aspx/desktop-xps-625?c=us&cs=19&l=en&s=dhs
Here’s what I would do.
Core i7 920 300 -------------$300
Motherboard
GIGABYTE GA-EX58-UD4P ------$259
DDR3 Memory----------------------------$120
http://www.newegg.com/Product/Product.aspx?Item=N82E16820145223
BFG Tech BFGEGTX2951792E GeForce GTX 295----$500
http://www.newegg.com/Product/Product.aspx?Item=N82E16814143167
$1279
VERY few machines would come close, including mine and everyone else’s on this site.
1300 bucks, hello.
“Forrest... Trees... It's amazing how many folks ("enthusiasts") have such a weak grasp of the overall market.” Ah-hem.
Caveat emptor
SPARKS
Whoops! Enthusiats can't add!
That's $1179.
%)
SPARKS
I think that the XPS lineup shows one area where AMD's inability to catch up with Intel at the high-end hurts them. Dell basically added a lower tier to their enthusiast/gaming line in order to fit the Phenom II in.
Even assuming that those systems are better than similarly-priced (or even higher priced) Core2Duo/Quad systems, AMD is selling their CPUs for less, and they're only selling them in the lower-pricing tier to begin with!
The rest of the Dell XPS lines are built around Intel Core2Duo, Core2Quad, and Core i7 CPUs. Gaming systems are what people will likely pay more for, since they are systems that you buy for the muscle more than for the cost efficiency.
If the only niches that AMD can fill for Dell are at the lower end, and Dell is finding itself creating pricing tiers for AMD, it may not be long before Dell decides that it's not interested in selling systems with AMD processors.
"If the only niches that AMD can fill for Dell are at the lower end, and Dell is finding itself creating pricing tiers for AMD, it may not be long before Dell decides that it's not interested in selling systems with AMD processors."
That's an interesting point, and I agree. Considering DELL's positional loss in the market, in conjunction with their recent financial losses, points to two things. The pricing tier creation you mentioned, plus my belief that the “platform solution” hasn’t lived up to the hype we’ve been exposed to during the past few years. A lot of smoke, and too many mirrors. It certainly wasn’t worth 5.4B.
Additionally, Core i7 IS a future platform (read: LGA 1366), Socket 940 is getting rather long in the tooth. A new socket now would be devastating for AMD and its’ enthusiasts given the performance spread between the two companies CPU’s. In this scenario going with INTC is a given. I believe a socket change now is and will be AMD’s biggest hurdle in the near term. Basically, no gain, all pain.
Of course I am not referring to INTC’s “driving” force behind “SOC solution for ATOM”. This is the real deal and a completely separate animal. I don’t have the expertise in the industry that you guys do (e.g. ITK’s ATOM prediction), but I do know a loser when I see one.
“Dragon” ain’t cutting it, not with DDR2 or DDR3 on the high end, or on the low end.
SPARKS
Link from Aces forums. Could reports of Itanium's death be exaggerated? People have been calling it a failure for, how long... more than 10 years now?
Perhaps for Intel, $200 million in revenue per quarter constitutes a failure. :)
ITK- Now this is nice! This is the perfect accompaniment to an amateur chief in the kitchen (among other things).
The computer/keyboard would be perfect for whipping up those recipes on line, or from the home network. Let’s hope they make it wireless.
Sold.
SPARKS
Whoops, forgot the link!
http://www.fudzilla.com/index.php?option=com_content&task=view&id=11394&Itemid=1
SPARKS
Hey Sparks,
For the "low end" i7 system you spec'd out above, these parts might help lower the price even further:
Gigabyte UD3R X58 mobo: $199.99
And with the $60 bucks you save on the mobo you can even upgrade to 6Gigs of RAM:
from G.Skill
Chukula, I didn’t think there was an X58 Mobo below 200 bucks, well there it is.
I went with the Corsair memory because it’s rated for 1600 MHz. Personally; I’ll take the faster memory. The boost I got w/QX9770 @1800 FSB running the memory synchronous has gotten me terrific bandwidth (9.65 GB/s), almost but not quite, to AMD’s 10+K standard. Running low timings (7-7-7-21) (2.12 V) helped reduce latency down to an unheard of 57 ns on good ole FSB.
Check out those DOG DELL XPS 625 AMD systems for a $1000 to $1759. You can’t even buy these MUTTS with a 4870 or 4870X2, and PII 940 is gonna cost ya $200 more !!!
Anyway, what’s more interesting is how most everyone is saying the “high price of an X58 platform”. Horseshit. $120 memory, $300 i7 920, and now a $200 mobo! 620 bucks for a Core i7 system (Uh-- platform), GMFB!
Prices will continue to drop and these monsters will sell like mad!
And, let’s not forget SLI!
SPARKS
I didn’t think there was an X58 Mobo below 200 bucks, well there it is.
How long have I been saying that the mobo price is being driven by the components not the board? With x58 being made on 12" wafers, the price was due to come down. And that is for a top of the line chip. AMD's mobo advantage may well disappear when x58 is no longer the top of the line chipset.
How long have I been saying that the mobo price is being driven by the components not the board?
I'll defer to you on the MOBO items, but also there is a component of prices being high due to demand/small supply as well. As folks start recouping development costs and getting better economies of scale on these boards (along with the chipset price reductions that ITK talked about) prices will be down below $150 probably by mid-year.
And on a side note, Intel needs to start working on the power on the chipsets... while this is probably in part to them using a lagging technology node to fab these, they need to work on the design portion in terms of power consumption. They also should lower the idle speed on the Core i7 to further drop the idle power consumption.
And on a side note, Intel needs to start working on the power on the chipsets... while this is probably in part to them using a lagging technology node to fab these, they need to work on the design portion in terms of power consumption.
Part of that will be done by the repartitioning that is part of the Nehalem architecture going forward. The new chipset is called Ibex peak. The X58 chipset is really an enthusiast chipset and I'd bet power wasn't a big concern. An early look at Ibex peak from Oct'08 on VR-Zone.com said:
Ibex Peak will be able to support mainstream LGA1160 quad-core Lynnfield and dual-core Havendale with integrated GPU in Q3 2009. It is a single chip solution known as Platform Controller Hub (PCH) that has all the I/O features of the ICH.... It has the ability to support dual graphics like SLI and Crossfire in x8/x8 configuration. However, the P55 chipset interconnect is still on the DMI instead of the new QPI found on the X58 chipset. For the I/O aspect, it supports 14 USB 2.0 ports, 8 PCIe 2.0 x1 ports, 6 SATA 3.0Gb/s ports, Gigabit Ethernet and IMST 9.0 with FIS port multiplier.
Combine the single chip controller with the on die memory controller and you should start seeing improvements in power characteristics. Though I suspect you may be right and Intel will have to do more going forward.
For this year they are pretty well tied into this path though, as Ibex Peak is scheduled for a Q3'09 release.
Here ya go, a dual core, a 4 thread, fully functional microcomputer.
With NVDA’s strength in graphics, this will be a winner. The little monster has 2G of DDR3! It gives a new meaning to ‘hand held”.
http://anandtech.com/video/showdoc.aspx?i=3499
SPARKS
I see that since Jan. 9, the day after the Deneb NDA lifted and most reviews on it were out, AMD's stock price shed about a quarter of its value - down to $2.15 today from its high of $2.75. Granted the market as a whole has been trending lower this week but AMD is, as usual, the point of the falling spear :)...
http://www.eetimes.com/news/semi/showArticle.jhtml;jsessionid=YKDOQ3UR20F3IQSNDLOSKHSCJUNN2JVN?articleID=212900153
Pessimism grows for EUV litho
IBM Corp. agrees. ''EUV will not be ready for the 22-nm node,'' said Gary Patton, vice president for IBM's Semiconductor Research and Development Center, during a presentation at the Industry Strategy Symposium (ISS) here.
Funny wasn't IBM touting an EUV breakthrough not too long ago? (talking about a "researchy" 2 layer structure they produced). I'm stunned... another research announcement from IBM which has no actual representation of production worthiness.
IBM say’s this, and IBM say’s that, bla, bla, while the pesky cockroaches at AMD “flap their arms about "yelling look at me I'm in the lead". Then they spout, ‘we have this roadmap, the other roadmap, and other roadmap, coming out next week.
MUTT AND JEFF
Then you have this guy, MEGA SUPER GURU! When this guy talks, real quiet like, everyone should just STFU and listen! This is where is going and that’s that.
http://www.semiconductor.net/article/CA6621436.html?nid=3655
Imagine, Core i7 XE @ 32, with a few ‘refinements’!?!?!
Man, they know how to get my money.
Word to IBM: Move over rover, let Mr. Bohr (and the boys) take over!
HOO YA!
Oh yeah, EUV. This:
http://www.semiconductor.net/article/CA6553758.html
and this……. GURU is batting 1000, he’s NEVER wrong.
SPARKS
And here’s the difference:
This little excerpt shows the fundamental difference between INTC and IBM. Mr. Bohr (and company) deliver product, while IBM delivers papers!
http://www.semiconductor.net/blog/
270000427/post/1780037778.html
Imagine 1V transistors. I’ve got a cadre of diodes that won’t work at 1V.
SPARKS
Hey Sparks
IBM does more then publish papers. They did develop the technology and architecture and product the gaming chip. I believe over the last two years combined they have sold as much units as ITNEL sells in a month. SO don't slam them they publish lots of papers and make some stuff.
There is the minor detail that their customers have lost billions on the hardware while INTEL makes that much ever couple months.
Isn't there another company that relies on them for silicon. Its funny how they have lost money too.
Is there a correlation perhaps?
Hey Anonymous
Which gaming chip?
What product over the last two years?
What correlation?
By the way, IBM is primarily a services/solution provider who over the years has shed most, if not all, of its manufacturing.
(Their R+D arm, according to many posters who on this site WORKED for IBM, say is world class. I wouldn’t doubt these guys for a second.)
One of the last things they ‘divested’ themselves of was the infamous “Deskstar” series of hard drives. These were commonly known to enthusiasts as the ‘DeathStar’ series. Their legendary (‘click of death’) failure rate was nothing less than tragic.
Hitachi bought that failed mess lock, stock, and barrel, for 2B in 2002. (They laid off 1500 employees at the time) However, now, they are some of best and fastest drives around.
http://www.internetnews.com/infra/article.php/1183321
IBM used to manufacture laptops, they sold the “Thinkpad” Division to the Chinese Lenovo Group in 2004-5 for 1.75B.
http://www.laptopical.com/ibm-sells-to-lenovo.html
Believe me, I read your post quite a few times, what’s your point? POS retail computers perhaps?
SPARKS
SO don't slam them they publish lots of papers and make some stuff.
I'll slam them on their papers and announcements... they tend to announce things way too early before the integration is even proven (forget manufacturing worthiness). This is fine if you work in a research facility or in academia, but they intentionally word these announcements to sound like they are just a short step away from actual production.
They made a big deal about SiLK (spin on low K) a long time ago based on a simple 2 layer structure - the material/process was a complete flop as it could not hold up to a full backend metallization (with more than 2 layers). They announced highK/MG as "an option" for 45nm , when it was clear they did not have any integrated full loop data (they basically had done a short loop proof of concept). They announced a 2 layer EUV structure with the implication that it maybe ready for 22nm - of course, many folks can make a 2 layer structure... if you don't care about throughput (because the source isn't powerful enough) or any mention of whether they could do a realistic structure with reasonable yield or manufacturability tolerances.
Their latest 22nm SRAM cell announcement was also questionable... it was based on functional cells as opposed to a functional SRAM chip (with million/billions of cells)... the features were quoted as less then 25nm for the gate length (which is a bit ODD for a 22nm tech - why would they not say <22nm?). It is unclear how the litho would hold up on a large array of transistor (I have seen transistors printed down lower than 10-15nm, but that is pretty much meaningless in isolation).
In my view, IBM preys on the lack of knowledge of the press and allows the to take the announcement and misinterpret it for them... and technically IBM's hands are clean as they did not do any theoretically misleading, they just word things in a way that allows the press to do it for them.
“G”, in my above post, I mentioned “Imagine, Core i7 XE @ 32, with a few ‘refinements’!?!?!”.
I ran into this tidbit:
http://www.geek.com/articles/chips/32nm-tri-gate-
transistor-technology-from-intel-20060614/
Could there be any substance to this article?
The article contradicts INTC’s ‘Tick Tock’ strategy where going to 32nM mandates a process/architectural shrink of current Nehalem, as apposed to an architectural change required by implementing Tri Gate structures in a new design.
Pat # 7,422,946-
"Independently accessed double-gate and tri-gate transistors in the same process flow."
“A method for forming first and second devices from first and second silicon bodies is described. A sacrificial layer allows gate regions to be defined with underlying insulating members”
I assumed that Tri Gate Transistors would be an architectural change requiring a new “Tick” sometime in (Mid?) 2010 on a proven 32nM process. I’m not certain (because I’m an amateur), but wouldn’t Tri Gate necessitate different design rules?
SPARKS
Could there be any substance to this article?
Not "G" but I'll answer this one.
If you can believe Intel, this article is total hogwash.
Kelin Khun, who is in charge of 22nm transistor development (I posted the link a while back) considers tri-gate to be a high risk process for 22nm. Let alone 32nm.
If they aren't sure it will fly at 22nm, do you really think it will be on 32nm in 9-10 months?
It ain't going to happen. You can put that in the bank.
ITK Thanks. I thought so. I found the link.
http://www.semiconductor.net
/article/CA6622435.html
SPARKS
Gotta agree with ITK... I had originally thought 22nm (as 32nm will be more or a less of shrink of 45nm with some high K tweaks and some other stuff); but I have revised that prediction to also consider 22nm iffy.
I also do not think that trigate would necessitate an architectural change, I would consider it more like the high K change where you can use existing designs (think Core2 65nm to Penryn Core2 45nm transition); but you would probably gain more benefit from a new layout/architecture that is designed taking it into account. I would think you wouldn't be able to do a simple optical ("dumb") shrink on an existing product; but these days I don't think you have simple shrinks on any product regardless of what's in the technology node transition.
“I also do not think that trigate would necessitate an architectural change….”
Ah, you knew where I was going with that question. I looked at the microscopic images of the FIN FET devices in contrast to the planar devices, and I couldn’t understand how they could ‘plug-in’ the former without changing the overall topographical landscape of the entire “schematic” (for the lack of a better term), hundreds of million of times over on a single die.
Basically, without designing the entire front end of the vehicle, can you put a new fuel injected Chevy rat motor in an old Camaro?
I realize my loaded question was flawed with that ridiculous date, as ITK pointed out. Your answer, however, implies the engine change is within the realm possibility.
Further, if INTC does work the bugs (risk assessment?), they could drop the high performance motors in at anytime.
Instant horsepower, lower ET’s.
SPARKS
Further, if INTC does work the bugs (risk assessment?), they could drop the high performance motors in at anytime.
Not sure drop in would be the right term - certainly some layout work would be needed; but I don't see trigate requiring a new architecture. Even the Penryn "shrink" was more than just a shrink. Also trigate is a substantial enough change, I don't see how it could be anything but a start the new technology node with it type of change.
This is also why I think AMD switching to 45nm high K mid-node (or late node) is unlikely - while certainly doable, it is not a "drop-in" to existing mask/layouts. The necessary process qualification and layout optimization work would probably not be justified with small/moderate volumes, so unless 32nm is late, I don't see this making economic sense for AMD.
A couple of sites are reporting that Intel will cut prices on the C2Q line in a few days, bringing the Q9550 to just under the Phenom2 940 and the Q9400 well below it. I am interested in whether or not Intel will also lower Core i7 prices, or if it will hold the line on any processor that is Q9600 or better.
If they hold the line above the Q9550, it makes you wonder how anyone could see this as a case of AMD putting pressure on Intel instead of the other way around. And it seems that Intel is not going to play any games- the two CPUs that are the most direct competition for the Phenom2 (the Q9300 and Q9400) are pushed down well below the 920/940, while the Q9550 hits a price point just lower than the 940.
This is, again, what happens when your best CPU is only competitive with the other companies' midrange offering. Intel can afford to price a large part of its CPU lineup according to market demand and other factors (inventory, etc), whereas AMD will have to price their entire lineup according to Intel's pricing on their mid/low end. Q1 may be particularly brutal for AMD.
Intel almost reported a Q4 2008 loss today, due mainly to a huge writedown on the WiMax venture. The other surprise was the big expected drop in gross margins for Q1 2009, down to the low 40s.
Otherwise, not much that was unexpected given their warning last week. Not good news but they're still far better off than most of the other companies set to report their results over the next week or two.
Tonus
Intel can afford to price a large part of its CPU lineup according to market demand and other factors (inventory, etc), whereas AMD will have to price their entire lineup according to Intel's pricing on their mid/low end. Q1 may be particularly brutal for AMD.
Well, it's basically the same situation as last year continued indefinitely until possibly Bulldozer: Not enough cash flow to pay for the fab upgrades but likely more than enough to fund design. The difference this year is that Arab money will start flowing in to make AMD's finances look much better, at least in the short term. If ATIC doesn't amend the terms of the deal again, that is.
If they hold the line above the Q9550, it makes you wonder how anyone could see this as a case of AMD putting pressure on Intel instead of the other way around.
As the C2Q's are MCM they still have to be extremely cost effective (and not to mention smaller) than the Phenom II's. The cuts are probably more of an economic (overall sales slowing); then an AMD response - but the timing of the cuts is clearly designed to hit AMD. The problem I see is that the low end Core i7 is not much of a jump up in performance, so cuts to the C2Q's may start cutting into the 920 i7 sales (and maybe 940's)?
I think Intel wants i7 to be the premium line (much like Ed at overclockers theory) and will leave it alone. What all the people who scream price/performance don't realize is that in many markets (most?) there is a segment of the market that is price inelastic (meaning they will pay for the incremental performance almost regardless of the cost). I wonder what will happen when the C2Q MCM's get long in the tooth and what Intel does to attempt to segment mid-range/mainstream from the high end.
"A couple of sites are reporting that Intel will cut prices on the C2Q line in a few days."
This price cut was scheduled back in Sept, as Digitimes (dead link w/o script) reported. THG picked it up, however.
It makes you wonder what comes first the chicken or the egg.
Do AMD’s releases coincide with INTC’s roadmap, or does INTC orchestrate them that way?
http://www.tomshardware.com/news/Intel-Roadmap-Prices,6379.html
SPARKS
Do AMD’s releases coincide with INTC’s roadmap, or does INTC orchestrate them that way?
If you think AMD has the execution capability to match up a product release to a price cut date, well... you've been asleep at the wheel the lats few years. AMD's roadmap are shall we say "fluid" and happen when they happen, not when they are planned to happen!
Hmm
AMD big write down for ATI miss adventure and loses billions
INTEL big write down for clearwire and still squeaks to a profit.
If INTEL's GM is going to be in the 40's can you imagine what kind of quarter is in store for Arab Micro Devices.
Anyone check what has happened at the other foundrys TSMC, UMC, Charger, SMIC etc. etc. They are empty. Now you got Arab Micro Devices investing billions to own AMD fabs, want a cut of the money, demand evaporating, AMD parts barely competitive.
AMD is going to be BK.
As to IBM.. you arm chair boys got no clue.
Sparks for your information both Cell and the xbox CPU are both derivatives of the IBM PowerPC. There is a new book about the next generation gaming chip on sale and its a hilarous read about how SOny unkowingly funded microsoft and how the engineers at IBM were hiding things from each other. This is what happens when competitors get togather in a consortium. It dont' work out.
Its comforting to know I'm not the only one who remembers IBM SILK, IBM strain and other grand papers from them experts. SOI is no different.
AMD going BK. Where is Sharikou and Scientia these days.
On Monday, the Inq reported that Intel feels that it won't have to lay off any employees (having laid off 20,000 in 2006) even though they took a beating in Q4. Today Inq is reporting that AMD may lay off another 9% of their workforce, and will cut pay for the rest. Nasty times ahead. The company I work for just laid off seven people, and it sucks.
"Sparks for your information both Cell and the xbox CPU are both derivatives of the IBM PowerPC. There is a new book about the next generation gaming chip on sale and its a hilarous read about how SOny unkowingly funded microsoft and how the engineers at IBM were hiding things from each other. This is what happens when competitors get togather in a consortium. It dont' work out."
Any links there big fella?
SPARKS
Here you go Sparks,
http://online.wsj.com/article/SB123069467545545011.html
Ok, you computer genus/financial wizards. Does anybody care to take a shot on how much AMD will write down the ATI purchase, yet again???
Joe at Overclockers says there’s only 735M left.
“AMD has already taken two writeoffs amounting to $1,606 and $880 million relative to the ATI acquisition. AMD initially booked ATI goodwill at $3.22 billion, so this leaves about $734 million left.”
I say they take $350M, half now, allowing them another $350M write down in 1Q ’09.
The winner gets a Wrector Ruinz photo link, and an autographed copy or his new book, “How to kill two companies and lose 10B in three years” (I figure 6B in cumulative losses plus 5+ B in debt)
What say you ‘Moose’?
SPARKS
Sparks: Ok, you computer genus/financial wizards. Does anybody care to take a shot on how much AMD will write down the ATI purchase, yet again???
OK, I'll bite - I'd guess $5.4B :). Going by Hector's reasoning (and I use the term loosely), this writedown to -$4.5B total means AMD will get a bailout loan of +$4.5B from our friendly federal gov't, since the Dems hate to see a cash hole like that and will attempt to fill it up with our tax dollars :).
Seriously, since ATI actually made a small profit last quarter, I'd guess the writedown won't be that much. However I saw elsewhere that Intel actually increased its marketshare in Q4, which if true means it was a disastrous Q4 for AMD. So I'd guess a loss of at least half a billion, maybe more.
And I recall last August somebody from Intel saying that AMD's 45nm yields were poor, at the time anyway. Seeing that Shanghais and Denebs seem to be readily available, then either they improved the yields greatly or else they have sold so few of them that the few they can make is more than the market demand.
If the economy doesn't recover in the next couple of quarters, I'd have to agree that AMD is gonna be Chapt. 7-11 before the end of the year...
Looks like we were all wrong - Reuters is reporting [AMD will] take a new $622 million charge for its acquisition of graphics chip maker ATI, bringing total writeoffs for the deal to $3.17 billion.
Doc,
It seems imitation is indeed the sincerest form of flattery.
I give you Dementia’s latest post. Although I try not to either contaminate or waste the precious few working brain cells I have left by reading his biased bazaar pretzel logic, it’s interesting to know he’s searching for a new system, too. Just like you.
But his requirements are 3X the power of his present dog. (That should be easy)
He conveniently fails to mention anything about the “wait” and “the deafening silence” mantra of his last post. In fact, he’s again leading the charge on the performance per dollar comparisons. He mentions absolutely NOTHING about power consumption.
Obviously, he’s gone out of his way to be “the objective analyst” in reviewing today’s purchase options. Further, he totally ignores future upgrade options as AM2+ is done in 6 months. I suppose claiming to be a performance authority while upgrading every five years speaks volumes of his hands on experience.
You, in contrast, put a more reasonable $1000 cap on your personal requirements; he on the other hand puts his cap at a ridiculous $500. I suppose that price point is where AMD lives, where he could make a ‘fair and objective’ recommendation.
He mentions a (GO) stepped Q6600 inferring to a VERY cheep overclocked solution. Dirt cheep DDR2, dirt cheep INTC 3x 4x mobo’s, coupled with a Q6600 is the way I went with my wife’s machine. OVER ONE AND HALF YEARS AGO!
His mention of a Q6600 as a purchase option signifies, by default, a comparison to INTC’s oldest quad to AMD’s latest and greatest newest Quad. He unwittingly concedes performance issue. (That why lawyers tell you not to talk.)
The silence is deafening.
SPARKS
BTW: I was wrong. Core i7’s performance didn’t shut him up for another six months.
Moose-
Peh! So much for ATI’s graphics for doing “very well”, I think AMD would write them off even if they had a total market share. I mean really, ATI is the only thing that’s making them any money.
Whew, $622M! So, they overpaid how much, 3.17B?
Hey, if their writing off nearly ¾ of B, plus laying off an additional 9%, they really must be scrambling to keep this cinderblock afloat.
SPARKS
Here is and interesting take on the MID vs smart phone argument.
I think it’s fair to say that if you use a smartphone like a MID, it isn’t going to last long. My N82 lasts for about three hours; less if I’m using 3G. The iPhone, about 4 hours under heavy use and the Android-based G1 seems to be even worse. The fact is that radios and processing take energy and although your smartphone will sit doing nothing for days on end, once you start using it connected over Wifi or 3G, the clock starts ticking.
So what I take away from this is that the real advantage of the smart phone (from a power perspective) is pretty much limited to the "always on" when idle feature. If my interpretation is accurate, then the Moorestown platform should do just fine in smart phones with its 10x reduction in idle power usage.
So much for ARM's vaunted power advantage.
sparks: "His mention of a Q6600 as a purchase option signifies, by default, a comparison to INTC’s oldest quad to AMD’s latest and greatest newest Quad. He unwittingly concedes performance issue. (That why lawyers tell you not to talk.)"
It is, once again, an example of what I've described before, which is the willingness to view things through a 'wishful thinking' filter, or at least any filter that avoids analyzing any aspect that shines a negative light on AMD.
Much of what is discussed here, which is also what Ed at overclockers used to discuss, is how the bottom line is affected by the decisions and technology that AMD and Intel make. This is not an area that many enthusiasts even attempt to understand, so their view of what is positive and negative is a bit skewed.
For example, silentpcreview.com reviewed the Phenom II and felt that AMD had done what it had to do, which was to close the performance gap and hold their pricing to be competitive in the niche that they covered, performance-wise. Now, that works out great for the consumer, who can consider a drop-in replacement for his AM2+ motherboard, or who is looking to upgrade to quad-core on the cheap (and somehow manages not to notice the Intel bargains in that range, heh). But it's a view that ignores how AMD is affected by the fact that they have ceded the high-end to Intel.
The problem is that some reviewers make assessments of the market based on their approach that is limited to performance-based factors. To them, it's great that AMD has closed the gap ever slightly, because it means they're "headed in the right direction." As if they have unlimited cash and unlimited time, and it's simply a matter of eventually developing a chip that gets them back in the game. Reality is a lot harsher than that.
sparks: "His mention of a Q6600 as a purchase option signifies....
I went there and took a look, and i stop reading through somewhere around this:
scientia blog: Intel integrated graphics still lag behind both nVidia and AMD, so an nVidia motherboard is required for Intel unless I want to add a discrete graphics card from the start
basically he is telling (or making up) his point saying that he is going to buy a mid-high quad core system with IGP first ... :)
Sparks: Hey, if their writing off nearly ¾ of B, plus laying off an additional 9%, they really must be scrambling to keep this cinderblock afloat.
Ed at Overclockers guesstimates $0.5B in actual losses as well, so AMD's Q4 could set some sort of negative value record for them, this Thursday.
The AMD fans at Tom's were predicting "billions" in "losses" for Intel for Q4, so when I posted their actual results (as fazers_on_stun) showing Intel made a small profit, they changed their story to mean "reduced revenue". Funny how they move the goalposts around like that, sorta like Sci when his credibility gets challenged :).
I find it hilarious that literally one day after Scientia's latest blog post assessing the cost competitiveness of Phenom II, Intel largely invalidates his conclusions by announcing major price cuts that had been widely expected anyway following the introduction of Core i7.
AMD needs to drop Phenom II prices by roughly $70 to remain competitive and to keep Scientia's latest article remotely relevant. Currently, Core 2 Quad is easily the better value.
Also, Scientia's endless fixation with Anandtech is beginning to border on frothing-at-the-mouth psychotic. Aren't there many other review sites that arrived at essentially the same conclusions as AT did over the last few years? Also, AT's review of Phenom II was decidedly positive, from the title right through to the conclusion, making claims of Intel bias pretty questionable. Anyway, even if AT is the most visited enthusiast site we should not overestimate the impact of enthusiast sites on the general public. They don't read these reviews and never will.
Axel, ya beat me to it.
If there was any reason to consider a Pheromone system, INTC just put the kibosh on that.
$316 Q9650! Where the hell is Q9550 going to fall?
This is the one that will KILL PhII.
SPARKS
That's interesting... Intel is also introducing a line of 65w CPUs that are priced at a premium compared to the same models rated at 95w. Is this the fruits of a new process or refinement of their 45nm process? Is the pricing a reflection of low initial volume or is it a shot across AMD's bow? I know that AMD has made some noise about performance per unit of power, and so it seems that pricing in the Phenom 2 niche is now something like:
Pricing tier 1: Intel 95w Core2
Pricing tier 2: Phenom 2
Pricing tier 3: Intel 65w Core2
If that's the case, it's a shrewd approach from Intel. It seems as if they smell blood in the water.
Intel is also introducing a line of 65w CPUs that are priced at a premium compared to the same models rated at 95w.
Frankly,this is probably just an attempt to eke out more revenue. I highly doubt there would be a 30Watt delta between them (not that you were implying that); and when you factor in that idle power will probably be pretty close... Well I wonder how much of a premium Intel can command on these. This is probably an attempt to get a little more revenue for those willing to pay for a little extra efficiency (and cash in on the green bandwagon)
Of course you could just undervolt the 95Watt chip just a little bit and make up the difference on your own (of course you could also undervolt the 65Watt chip too...).
I'm not sure about the additional binning cost (if any); but just like I railed on AMD for having what seemed 1000 SKU's over a mere 500MHz range, I think Intel runs the risk of having too many product bins. Unless they will be phasing out the 95Watt models and will ultimately end up with one power bin at each clockpeed?
Tonus - do you have a link on the 75Watt plans? I'd be curious to see if it is a new stepping or just more aggressive binning/process refinements.
Axel: Also, Scientia's endless fixation with Anandtech is beginning to border on frothing-at-the-mouth psychotic.
It's a well-known fact that Sci The Pimpi is flat-out jealous of Anand Lal Shimpi, because Anand has a real website and advertising dollars and can afford to build a pretty neat home theater. Sci has trouble scraping up $500 every 3 years for a new computer :)
"Sci has trouble scraping up $500 every 3 years for a new computer :"
LOL, I thought the same thing.
SPARKS
Guru, this is the link at Dailytech.
Sci has trouble scraping up $500 every 3 years for a new computer
The thing I found amusing was how motherboard costs and the "high" cost of DDR3 were brought into the picture to quickly discard the i920.
Sure, I will concede that these are legitimate costs and you can make use them as part of the argumnet. But the pricing criteria he pretended to lift from Anandtech was strictly CPU pricing. That is until it didn't suit where he was going with his post.
At the very least he should have subtracted the additional cost of the motherboard and memory of the "acceptable" systems back out of the equation. Then he could pretend it was a CPU cost and that he was making an apples-to-apples comparison. But he didn't even try.
Talk about the pot calling the kettle black.
I won't even get into the slight-of-hand regarding cache size.
I suppose I should post a thank you for providing me a chuckle, but I'm sure he would ban it as offensive.
Tonus - thanks for the link.
The pricing is a bit "curious", a 65Watt, 2.83GHz C2Q is $53 MORE than a 3.0GHz, 95 Watt part? It is also $103 more than the 2.83, 95 Watt part - it makes sense that it is more, but a 40% premium for it?
This makes me thinking this is a binning thing (and not a major stepping revision) and I'd bet the 65Watt'ers are in short supply. I still think you'd be better off buying a 3.0GHz part and undervolting it... save $50 and probably end up near the same power consumption at a slightly higher clock (especially if you factor idle time in).
I'm looking at upgrading a dual core to quad (Gigabyte P35 board), so the cuts are coming at a good time for me! Debating between the 9550 and 9400 (not sure how big a deal the extra 6MB makes)
Axel: "AMD needs to drop Phenom II prices by roughly $70 to remain competitive and to keep Scientia's latest article remotely relevant. Currently, Core 2 Quad is easily the better value."
They appear to have cut them by $40. I am sure that scientia can still find a way to justify his Phenom II purchase, and the Phenom II is still a very good deal at the new prices. The problem for AMD is that they're forced to slash prices shortly after they released their newest CPUs, in order to keep up with Intel's midrange offerings.
ITK: "But the pricing criteria he pretended to lift from Anandtech was strictly CPU pricing. That is until it didn't suit where he was going with his post."
That post confuses me somewhat. His point seems to be that Anand was being critical of AMD's CPU pricing, but decided to overlook similar pricing from Intel.
Yet the quotes he provides indicate that Anand's criticism was that AMD's cheapest dual-core at the time was costlier than Intel's. Today, Intel does have CPUs in the $300-500 (and much higher) price range, but they also have dual/quad core CPUs in the low price ranges as well.
If the Intel 9650 was their lowest-priced quad-core, I can understand calling out Anand. Or if Anand was wrong about AMD's CPU pricing (ie, if they had a lower-priced dual-core at the time and he missed or ignored it for any reason). But if his point is that Anand treated them differently, his post doesn't make sense. Anand treated them differently because the circumstances were different.
Are people really reading that blog post and not recognizing his mistake? Because it's pretty obvious, unless there is something that I am missing?
It is rubbish comments like this that drive me nuts about Scientia's posts.
There actually is no problem with the current prices for PII in terms of profitability.
The only way you can make a statement like this is if you know AMD's costs. So either he works for AMD's accounting department, he is a high level AMD executive, or he is just full of it. I'll let you decide which.
Anonymous AMD buyer/seller. I must give you credit for an excellent, well timed turnover. AMD is back down to $2.00. You bought at 2 and sold at ~2.75. Well done, know when to get in, know when to get out. Brilliant, ya got some set of calzones there papa, much bigger than mine.
SPARKS
"AMD reduced their pricing late yesterday on the Phenom II X4 940 from $275 to $235 and the 920 model from $235 to $195."
Now this was in response to INTC's cuts. I wonder what the master finacier 'Dementia' feels about those profitable AMD chips now?
The perfect storm gets WORST!
'Avast! Man the life boats. Leave the deck chairs, take the cannolies!'
SPARKS
sparks: "Now this was in response to INTC's cuts. I wonder what the master finacier 'Dementia' feels about those profitable AMD chips now?"
Unless I am mistaken, his comment about profitability was made after he learned of the price cuts.
So let's see if I've got this right. AMD dropped their price by about $40 on the PhII. But there are no issues with AMD selling the PhII at a profit. So doesn't that mean those blood sucking leaches at AMD were gouging us for their processors? :)
So doesn't that mean those blood sucking leaches at AMD were gouging us for their processors? :)
It means in the wannabe financial analyst/market analyst world there are apparently only quad desktops!
Desktops are now <50% of CPU's (with revenue declining). Now take that <50%, and ask how much of that is quad desktop? It really doesn't matter how much they charge.
The other thing - DEMENTIA is probably only considering the production cost (fab and assembly costs) and comparing that to price. He is ignoring all the other expenses... salary, taxes, sales/marketing, distribution, debt payments, payouts to NY politicians. Ultimately you need NET PROFIT (not just operating profit).
The bottom line - he doesn't have a clue how profitable these are for AMD, nor do I know (though I don't pretend to). There is just no public data (yield, wafer cost, packaging cost, bin splits) to understand the profitability of a specific product segment and what the true cost of a specific product is.
You want the reasons for AMD's world of hurt?
http://www.fudzilla.com/index.php?option=com_content&task=view&id=11606&Itemid=1
What was on the original roadmaps for Q4'07 (dual core K10's), now might actually make it out Q3'09! AMD has spent 2 years battling clock and power issues and decided to do this internal battle on quads instead of what would likely have been much easier in dual cores. This at a time when desktop quads was still a niche market, and not a budget market (i.e. the people buying quads were willing to spend the money for performance and were not looking for performance per watt per dollar per IGP trying to play FPS)
So AMD spent (/will spend) nearly an extra 2 years servicing the highest volume desktop market by simply cutting prices on K8... not exactly a "strategic plan". And now that they are getting around to cranking out some duallies, the migration to quad is much further along and the impact of a fast dual core will be much less. Oh and notebooks have overtaken desktops.
The other huge reason for there troubles is the notebook sector. Next thing you know AMD will be going after the netbook segment :)
Man you said it.
From Fuddie...
"AMD is finally getting ready to launch some competitive dual-cores..........."
".............it should ship in very early Q3 2009"
"..........but at this point AMD doesn’t want to reveal its highest 45nm dual-core clock. If it wants to fight Intel it will have to go much higher than 3GHz."
Fuddie amazes me. He swallows/reports this "scoop" as if it's some great break though. On closer examination what it reveals is a sorry state of affairs for the Scrappy Little Company.
First, he says AMD is "getting ready" in 9 months. Hey, GURU can beat the shit out of me for being a newbie, but I do know a little something (as I'm sleeping at the wheel), 9 months (3 Quarters) is friggin enernity in this business!
Second, isn't bringing out a dual core solution in 3Q 2009 kind of late to the party when quads are getting as cheap as dualies? Ok, I'm a performance slut, but at these prices I don'i even think about duelies anymore. The movement towards programers compiling multitreaded applications/programs has gone into full swing, not to mention nine months from now. So are the OS's. I see this as a monumental back track.
Third, higher than 3 GHz!?! to go up against what, An E8600, perhaps? That is one chip I would HATE to tangle with now, or nine months from now. Guys have been blasting along with this thing at 4.5+. (remember GIANT) Does anyone at AMD think can do this, and what speed?
Fourth, what in hell are they going to charge for these things when they're releasing/selling quads at $235? Again, too many SKU's and very little range in price.
Finally, isn't this something that should have been done two years ago. They should wake up and smell the coffee. Intel has a lock on the dual core market, and AMD simply miissed the boat. When (if) AMD does release these things, the dual core market shall have passed into history.
Fuddie thinks this is all marvelous, I see a dual core solution dead by 2010. (Personally I see it dead now, I'm spoiled) It's all a wasted effort, too little and too late.
SPARKS
Third, higher than 3 GHz!?! to go up against what, An E8600, perhaps?
I chuckled at this one because I'm running an "old" dual core 6700 (65nm, one of the very early steppings) at over 3GHz, while I undervolt it! This thing is several years old and AMD is getting around to releasing something that may get close to it later this year? In fairness it will use less power at stock (though with my undervolting the old 65nm part will probably give this newfangled 45nm AMD technology a run for its money).
Sign me up! Gots to gets me one of these!
BTW sparks - dual core will not be dead in 2010.... there will be very little need for quads in the corporate space, so unless these are somehow forced onto corporate America (which is unlikely in the near term economic environment)
Things are getting ugly for Intel. It looks like they have just announced layoffs of 5-6K employees. The details are here. This after vigorously claiming they weren't planning any layoffs just last month.
I can't tell you how bad I feel for the poor sod's that are being shown the door in this economy.
The scary thing is, I don't think we have hit bottom yet.
The only amusing thing I have seen regarding this news is some poor myopic soul over at amdzone that thinks that Intel's problems stem from the recent price war with AMD. He even seems to insinuate that somehow Intel caused the crash. I'd love to see the logic behind that one.
Keep in mind that this is a group that, for the most part, believe that AMD is the one putting pricing pressure on Intel, and not the other way around.
Things are getting ugly for Intel. It looks like they have just announced layoffs of 5-6K employees.
According to the article Intel is closing some 5 plants around the world - anybody have an idea what these are? I thought the Malaysia location was a packaging plant.
Anyway Intel should probably close some of those old fabs, or else sell them to the Foundry :)
Blogger InTheKnow said...
Things are getting ugly for Intel. It looks like they have just announced layoffs of 5-6K employees. The details are here.
yes and no. the number given are those that affected, not the number of layoffs. anyway, quite some % of those would be really laid off if not able/willing to redeploy.
A Nonny Moose said...
I thought the Malaysia location was a packaging plant.
Not really, besides assembly and test, Intel Malaysia also has Si design center, failure analysis, tool design, packaging, regional IT, marketing and a bit over everything, except Fab. in this event, 2 of the assembly and test plants of Intel Penang will be shut down.
Let's hope they can all return to work when the ecomomy turns around.
Good luck all.
SPARKS
Look at what Fuddie is reporting now.
http://www.fudzilla.com/index.php?option=com_content&task=view&id=11617&Itemid=1
If anyone thought 4Q 2008 was bad for AMD, 1Q 2009 will be murder. Duelies in 3Q 2009? They can't give away $110 quads. Who's gonna eat these things?
SPARKS
Joe Citarella has picked up where Ed Stroligo left off, his latest look at AMD and Intel is decidely glum on AMD. I think his analogy ("AMD is cutting into muscle") is apt, if it's true that they're going as far as to cut wages across the board and even "suspending some contributions to employee's retirement accounts."
Intel's actions and outlook for Q1 indicate that they're looking at a bad quarter and a down year. AMD's actions indicate that they're in desperate straits. You can talk about how great it is that they spun off their foundry business (um, lol?) but when you're cutting wages and holding off on financial perks, you are no longer concerned with remaining healthy, you're trying to stave off starvation by eating yourself.
but when you're cutting wages and holding off on financial perks
On the positive (?) side, they probably don't have huge concerns with people jumping ship as there is no place to jump to in this economy (unless it is someone with real solid credentials).
Also the salary cuts are highest at the higher levels which are positions that are harder to move from in this type of an economy. As long as they minimize the mid/low level engineering loss, it's probably not too bad (the people DOING stuff as opposed to the people MANAGING STUFF)
That said things have to be pretty bad as they have done a few rounds of cuts (and probably realize they can't cut anymore people without having some "real" impacts). But it's no like they really have any other options at this point
And the spinoff of the foundry in this environment is actually not a good thing! This will require extra resources to coordinate between 2 companies when in effect the 2 companies are only doing one thing (producing chips for AMD). They have no external customers until ~2010 (and even then that will be minimal volumes), so now you have just effectively added headcount, or taken from elsewhere, to manage communication/logistics between the companies. And for those that consider this to be minimal - keep in mind you now will have 2 CEO's, 2 sets of VP's, 2 boards...and all those 20% cuts end up meaningless as you probably have 2X'd the # of really senior management between the 2 companies.
I don't think people understand until ~2011 (or later...when non-AMD orders become significant revenue for the foundry) this is really one company posing as 2 companies. If one fails they both fail. If AMD fails, the foundry loses its entire customer base and has no fab utilization. If the foundry fails, AMD has nowhere to produce chips (they are not allowed to outsource more than 20% of CPU production per the licensing agreement). The spinoff only "helps" when the foundry can stand on its own (or at least get substantial revenue from elsewhere); and for the next 2 years it is merely a business unit for AMD under the guise of a new company.
Also I don't read much into the Intel plant closings... but normally these people would have been re-deployed elsewhere in the company. In this environment (and considering Intel has been cutting headcount over the last 2 years) I would suspect the # of redeployment positions will be minimal. The plant closing seem like an excuse to reduce headcount - I suspect it will have minimal impact on Intel's manufacturing capacity.
To me the stock dividend is a key flag - if Intel cuts/lowers this then that is a sign that things are truly brutal. And I'm a bit confused how Nvidia has 3X the market cap of AMD (surely this is a debt thing). Sure Nvidia is competing against a weaker company in graphics, but soon Intel will be joining the fray and Nvidia will be in a worse position than AMD.
AMD's Q4 results are out:
AMD AMD today reported fourth quarter 2008 revenue from continuing operations1 of $1.162 billion. Fourth quarter 2008 revenue decreased 35 percent compared to the third quarter of 2008 and 33 percent compared to the fourth quarter of 2007. Fourth quarter 2008 revenue was down 28 percent sequentially, excluding third quarter 2008 process technology license revenue of $191 million.
In the fourth quarter of 2008, AMD reported a net loss of $1.424 billion, or $2.34 per share. For continuing operations, fourth quarter 2008 loss was $1.414 billion, or $2.32 per share, and the operating loss was $1.274 billion. The results for continuing operations include an unfavorable impact of $996 million, or $1.64 per share as described in the table below. Loss from discontinued operations was $10 million, or $0.02 a share.
For the year ended December 27, 2008, AMD achieved revenue of $5.808 billion. Fiscal 2008 net loss was $3.098 billion. AMD reported revenue of $5.858 billion and a net loss of $3.379 billion for fiscal 2007.
From http://news.moneycentral.msn.com/ticker/article.aspx?symbol=US:AMD&feed=BW&date=20090122&id=9538124
Looks like much worse performance than the estimated 54 cents per share loss the street expected..
A Nonny Moose
For the year ended December 27, 2008, AMD achieved revenue of $5.808 billion. Fiscal 2008 net loss was $3.098 billion. AMD reported revenue of $5.858 billion and a net loss of $3.379 billion for fiscal 2007.
So AMD lost a staggering $6.5 billion dollars over two years. About half of that was non-cash goodwill impairment, but still...
I'm sure there'll be a lot of reading between the lines for the earnings call that's in progress.
What I'd like to see are the marketshare reports for DT, mobile and server. I suspect AMD lost share in each segment.
I had thought AMD's revised guidance said something like 20% down - turns out it was more like 33% down.
Post a Comment