Dissappointing Year So Far for AMD

[Snowdog]

Yields that low are simply not possible to sustain.

I think it's pretty clear to all that defect ratio is inversely proportional to yield... The more defects the lower the yield, then taking into consideration die size it's clear that is directly proportional to yield. The larger the die at a given defect ratio, the lower the yield. This is common sense, but a defect ratio that high would require an unsustainably low yield, and I dont buy it. It's BS.

Like I said from the very beginning it assumes a ridiculous defect ratio. And I dont buy that number.

Here you ago yet again with this circular illogic. As I have stated numerous times and you ignore each time. There is a simple fact of the semiconductor industry:
Defect rate decreases as a process matures.
One more fact:
Intel has a much more mature 65nm process as they have been producing much longer.
Therefore:
Intel will have a lower defect rate.

Now in your absolutely twisted unreasoning, AMD can't have a higher defect rate because it would be terrible for AMD!?! This has got to be the most transparently faulty thinking you exhibit. Just because you don't like the outcome of facts, you can't work backwards and insist that the facts must change. I think this is known as insanity.

Now combine Intels lower defect rate with smaller die size Intel will have a much higher yield rate. Certainly we don't know the numbers. But this is a big part of the reason Intel makes tons of money and AMD loses it. That and Intel commands higher ASP. Lower yields and lower ASP is the recipe to lose hundreds of millions of dollars.

Just like you knew that P4 was going to be the greatest thing since sliced bread, and that Itanium was going to revolutionize the industry... Only people with your level of know how knew that.

No that is because unlike you, most of us are not blind rabid fanboys. I knew the P4 sucked that is why I used Athlons and never owned one. When the K8 architecture was first shown, it is why I said it would kick Intels ass and faced rabid Intel fanboys just like you are a rabid AMD fanboy. I just like the best tech, so I get attacked by fanboys on both sides.

I am knocked off my feet with a nasty virus this weekend. What is your excuse for blowing your weekend in front of the computer inviting people to attack you by saying ridiculous, irrational things?

 

[savantu]

..

Your "in depth" explanation doesnt even address what the bug is, and only it's side affects. As such it is of no value to any one here.

Do you realize how illogical your arguments are?

The 2.4Ghz and higher K10s were affected by the bug ( which appears in some corner cases ) .We can see the IPC on lower clocked parts which aren't affected by the bug.Strangely they are 10% slower on average than a Kentsfield at the same clock.

 

[Scali2]

Now in your absolutely twisted unreasoning, AMD can't have a higher defect rate because it would be terrible for AMD!?! This has got to be the most transparently faulty thinking you exhibit. Just because you don't like the outcome of facts, you can't work backwards and insist that the facts must change. I think this is known as insanity.

Now combine Intels lower defect rate with smaller die size Intel will have a much higher yield rate. Certainly we don't know the numbers. But this is a big part of the reason Intel makes tons of money and AMD loses it. That and Intel commands higher ASP. Lower yields and lower ASP is the recipe to lose hundreds of millions of dollars.

He doesn't seem to understand that it is indeed bad for AMD, but they still have to go through with it because no quadcores is even worse for AMD. Don't forget, their top models before Phenom were the 6400+, with 2x1 mb cache and built at 90 nm, clocked extremely high. No doubt those were expensive to manufacture aswell, and I bet they don't actually bin a lot of 6400+ at 90 nm.

 

[dexvx]

and that Itanium was going to revolutionize the industry... Only people with your level of know how knew that.

Itanium has been profitable for the last few quarters. It actually did revolutionize Big Iron - killed off Alpha, MIPS, and PA-RISC. Quite an accomplishment, I'd say.

 

[duby229]

Here you ago yet again with this circular illogic. As I have stated numerous times and you ignore each time. There is a simple fact of the semiconductor industry:
Defect rate decreases as a process matures.
One more fact:
Intel has a much more mature 65nm process as they have been producing much longer.
Therefore:
Intel will have a lower defect rate.

Now in your absolutely twisted unreasoning, AMD can't have a higher defect rate because it would be terrible for AMD!?! This has got to be the most transparently faulty thinking you exhibit. Just because you don't like the outcome of facts, you can't work backwards and insist that the facts must change. I think this is known as insanity.

Now combine Intels lower defect rate with smaller die size Intel will have a much higher yield rate. Certainly we don't know the numbers. But this is a big part of the reason Intel makes tons of money and AMD loses it. That and Intel commands higher ASP. Lower yields and lower ASP is the recipe to lose hundreds of millions of dollars.

No becouse it would be impossible to maintain yeilds that low. I'm sure they have lower yeilds then Intel, but I'm also sure they arent nearly as bad as your made up numbers make them seem. You dont know what yilds actually are. I'm sure Intel does have a higher yeild then AMD, prolly about 90% or somewhere in that range, and AMD's is prolly around 80% or somewhere in that range. This is based entirely off of how many chips there are capable of producing compared to how many chips they are producing. This really is the best that you can assume becouse you dont have the information that you need to perform any hard calculations,.... The data does not exist.

So what your doing is saying that AMD --must-- have a low yeild becouse I say so? How does that work?

 

[brucedeluxe169]

He doesn't seem to understand that it is indeed bad for AMD, but they still have to go through with it because no quadcores is even worse for AMD.

i honestly dont think a lack of quads would hurt AMD *that* badly. IMHO, quads hurt more than they help because they eat up so much of AMD's fab capacity, AND they are such a bitch for AMD to make, in addition to the fact that they just dont clock all that well....

really, AMD woulda done well to stick to dual cores for their entire consumer line and quads for the opteron line until at least 65nm is *very* mature, or 45nm is ready. I mean, even if AMD can get quads out in quantity, most high end consumers (the kind that want 4 cores in the first place) are probably running to the intel side anyway.... 95% of the market just wants a cheap dually, and a 2 core K10 woulda filled that niche nicely. Yea I know its all about ASPs anyway, but quad core K10s will make up such a small part of AMD's consumer shipments anyway.... besides, they would still have the quad core opterons on sale....


one last thing.... I thought griffin was a mobile, low power core? Why would anyone expect it to clock to 4ghz? I think AMD will be targeting the 2ghz range with that core and focus on power consumption and cost..... makes sense anyway, and the success of the EEE pc has shown that the ultra cheap/ultra mobile market wants cheap and low power, performance really is an afterthought (I would know.... me and 4 other friends bought EEE pcs.... not a single one of us cared about the performance.... why? all we use them for is email/spreadsheets)

 

[duby229]

Itanium has been profitable for the last few quarters. It actually did revolutionize Big Iron - killed off Alpha, MIPS, and PA-RISC. Quite an accomplishment, I'd say.

I guess that depends on how convoluted your viewpint is. Considering that the P4 was never supposed to be released to the public, and IA64 was supposed to be on the desktop many years ago, it was a total and utter faulure. It cost billions of dollars that'll take years to recover, and prolly never really will.

 

[duby229]

one last thing.... I thought griffin was a mobile, low power core? Why would anyone expect it to clock to 4ghz? I think AMD will be targeting the 2ghz range with that core and focus on power consumption and cost..... makes sense anyway, and the success of the EEE pc has shown that the ultra cheap/ultra mobile market wants cheap and low power, performance really is an afterthought (I would know.... me and 4 other friends bought EEE pcs.... not a single one of us cared about the performance.... why? all we use them for is email/spreadsheets)

We all know that mobile parts OC better thendesktop parts becouse they are hand picked for low power. We also already know that we have K8's running at 3.4 ghz on 65nm right now. Given a more mature process and a lower power architecture it makes perfect sense that Griffin may well be able to clock to those frequencies. The question is whether or not anybody will bother trying.

 

[Dan_D]

Boy is that statement going to come and bite you in the ass later. Ooooooooohhh goodie.. I cant wait for it.

I doubt it. Scali2 knows what he is talking about. I wouldn't say he is always right, but he uses reason and logic to arrive at an opinion. You do not.

Well, when there are only two competitors, there's not much of a gray area between good and bad.
Thing is, AMD needs a *better* core architecture than Intel, because they are behind on manufacturing.
This is why the Athlon and Athlon64 were successful. An architecture that is 'almost as good' as Intels means they're losing hard. Not only manufacturing, but also on architecture.
K10 was a year later than Conroe/Kentsfield, and should have been better, not 'almost as good'. AMD can't recover from this, because they won't ever catch Intel on manufacturing.

I agree completely. The only point I disagree with is that AMD can't recover from this. I think it's possible, though it is hard to visualize any way for them to do so.

Well you're the expert on statements biting you in the ass later

Currently AMD is still struggling with 65 nm, while Intels 45 nm is already doing well. By the time AMD gets 45 nm on the market, Intels 45 nm will be very mature already... Not very likely that AMD will beat Intel there.
So when exactly will AMD beat Intel on manufacturing? 32 nm? Heck, at this point I wouldn't even assume that AMD will survive that long. They might, but it certainly isn't guaranteed at this point.

Again, an opinion formed by logic and reason. I think that you are again right on the money, and I've thougth the same thing myself.

Boy you just keep diggin that hole deeper dont you?

If you took the time to read what I wrote throughout this thread, you'll know exactly what I think.

1: Barcelona is a stop-gap.

Ok.

2: AMD bought ATi, for it's R600 architecture.

There you go again. Even if that were true, it's obviously not helping much. If that's the case it's simply another in a list of bad decisions made by AMD.

3: Some future gen chip will incorporate a lot of ATi's IP in the CPU pipeline.

Perhaps they will, but that doesn't validate the approach or even prove that it will work.

4: Once this happens AMD will finally have a replacement for the --original-- K10.

Maybe.

5: Ruiz kicked Intels ass for how many years.

WTF are you smoking? Not once did AMD ever eclipse Intel's marketshare nor did they overtake them in overall sales. A few wins in the retail sector here and there does not equate to Ruiz kicking Intel's ass. This is a clearly biased fanboi opinion based on ZERO facts. AMD has had better performing parts here and there but over the last 20 years Intel has kicked AMD's ass in overall performance, sales and marketshare more often than not.

6: Becouse of that Meyer is going to have a more powerful company when he takes over.

No, I think AMD is at a particular low point. They are probably about as well off as they were during the K5 days, but there is no Nextgen Systems to buy out and help them this time. Maybe they though ATI would help them in this regard?

7: Meyers big contributions will be in manufacturing.

We will see. Do you have any data to prove that this will be the case? Is there any particular tidbits of information to substantiate this thinking or is it more blind loyalty?

That pretty much sums it up. Theres more of course, those are pretty much just the talking points.

Wow.

Yup, I even linked to a post where Dubious229 hinted that the Phenom would beat C2D, but he soon forgot that and keept trolling along.

He can't handle facts apparently.

You mention Intel yourself, and point the finger at me?
Hell, even Dan thinks you are a fanboi...did that come out of the thin air?

/ignore

Yes I do. I arrived at this conclusion based on the evidence. Much of the time Duby229 simply ignores facts and technical data and only posts crap to support his undying love for AMD/ATI in what can only be described as an irrational manner.

Except that he recognizes that he too is a fanboy.... And that's your problem. I know damn wel;l that AMD needs to get there shit on the ball, that they are indeed behind on manufacturing, and that K10 is not performing up to par. But you see I'm an optimist, and a fan of the company.

I recognize that I am a fanboy? Hardly. Currently Intel has the superior performing hardware. So that's what I bought. When AMD's Athlon 64 was kicking ass, I had one of those in my box. Later I bought two dual processor Opteron machines on workstation/server motherboards to serve my needs. I've spent more money on AMD processors in the last two years than I have on Intel processors in the same amount of time. My loyalties lie with the company providing me with the most absolute performance. Price is secondary to me.

This is a clear example of you twisting words and statements to make your point. A point or conclusion you arrived at without any logical thought or application of reason.

And so you know what I do? I stay on this side of the forum... You on the other hand dont recognize what you are, and come over here to bad mouth me, becouse you dont like AMD? Tell me how that makes any sense?

Some people....... I tell ya.......

Erm, sure they're going to try to catch up.
The thing I don't agree with is you preaching that it's inevitable that they are going to catch up, nay beat Intel in manufacturing, in the near future at that!
AMD is far too small and in far too much debt to even dream of catching up with Intel in the next 5 years or so. They tried to reduce the gap from 18 to 12 months this time with 65 nm, and look at that, not only is their K10 6 months late (making any theoretical gain null and void), but it's not even a good, mature product at this time. They'll likely need another 3 to 6 months to get it to the maturity that Intel had right at the launch of Core2.
Did they try to catch up? Yes.
Did they succeed? No, if anything they're further now than they were a few years ago. The only way to catch up is to cut corners and develop a new architecture while your new production process isn't mature yet. This is a huge risk, because you'll end up having to fix too many things at a time, often not even knowing whether the cause is design or manufacturing.
I think AMD learnt its lesson, and won't attempt to try this again.

Agreed. I think it's possible that AMD could catch up, and even release a vastly superior product to anything Intel has. However, this seems unlikely currently. I also agree that AMD catching up being inevitible is totally untrue.

Nor does it give you the right to patronize me becouse of the choices I make. If you arent willing to discuss the technology,and bash those who are, then your trolling and should leave. Just like everyone was taught from a young age, if you dont have anything nice to say, dont say anything at all.

On the other hand if your willing to discuss the pitfalls, and drawbacks of the technology, that's one thing, but "AMD sucks and your stupid" has no place here.

This is you ignoring facts and twisting words AGAIN. Many people in this thread have tried to discuss the technology with you and point out how they have arrived at logical conclusions that formed the basis for their opinions on the subjet matter. You refuse to acknowledge or even read many of these posts.

We are discussing the pitfalls and drawbacks of the technology. I didn't read a lot of posts that said "AMD sucks and your stupid" in this thread. That is simply the statement you gleened from the posts as you skimmed through them failing to comprehend or acknowledge the content of those posts.

AMD is clearly in a bad position. Clearly, they have an architecture that is inferior in most respects including absolute performance as of this moment. These are undeniable facts. AMD's process is clearly not mature, they aren't getting the yields they need, and even if they were, they are fighting an uphill battle against a larger company with much greater manufacturing capacity that has a superior product at this time. The ATI business unit doesn't appear to be helping them in the short term.

That doesn't mean that AMD sucks, or that you suck as you might have gathered from the above paragraph. What it means is that AMD is in a bad position and they are going to have to come up with an architecture that is superior to anything Intel has for some time and they'll have to be able to deliver that product in the quanties necessary. Then they will have to follow up on that product, with more products that don't simply rival Intel's but beat them. Then they will have to keep those reasonably priced, and they need to meet the demand for them, and repeat the cycle accordingly. They'll have to maintain not only product superiority but sales as well. Even if they get the first part right with a superior product, the sales part will be even more difficult.

You don't seem to understand the situation here. Realizing AMD has issues right now (some of which seem insurmountable) doesn't make me or anyone else a fanboy. This opinion is formed based on the facts presented, and understanding them in the correct context they are presented.

 

[ICE_9]

We all know that mobile parts OC better then desktop parts becouse they are hand picked for low power.

WTF? What makes you think that chips on one part of a wafer are lower powered than the others. That completely defeats the purpose of standardizing manufacturing technologies.

Mobile chips are made on mobile chip wafers. Desktop chips are made on desktop chip wafers.

The difference, some chips scale better than others on the wafer. Better binning at higher GHz.

 

[brucedeluxe169]

We all know that mobile parts OC better thendesktop parts becouse they are hand picked for low power. We also already know that we have K8's running at 3.4 ghz on 65nm right now. Given a more mature process and a lower power architecture it makes perfect sense that Griffin may well be able to clock to those frequencies. The question is whether or not anybody will bother trying.

well, i guess if you count overclocking.... but for stock speeds and the mass market, griffin wont get out of the low 2ghz range... nor should it, spreadsheet jockeys, youtube addicts, and programming nerds like me simply do not need/really desire anything more than something a 1.4ghz K8 class IPC core could give us in a *cheap* (emphasis on cheap), ultra portable like the EEE or slightly bigger... and that is the market that griffin and silverthorne are marketed towards....

i know i'll get flamed for this.... but the situation is NOT like is was in the 90s!!!! wake up people, AMD CAN cater to the low end and if they can get their costs low enough be SUCCESSFUL. The difference between now and then was that back then there was a real need and drive for faster processors because we *needed* them... today, anything northwood class and higher is pretty much all you need to be successful for the mass market, and AMD can, and *should* address this market. Sure their ASPs would be lower... but so would *anyone* looking to sell to 95% of the computing consumers out there. Seriously, I see no problem with AMD ceding the high end, and they probably should for now until they can get their debt under control.....

 

[Atech]

And Intel's 65nm can reach 5.3GHz:
http://news.softpedia.com/news/Intel-Core-2-Duo-X6800-Overclocking-Record-5-3GHz-51236.shtml

But taking that as guideline for how most people will run their CPU, or what Intel/AMD will be offering is so stupid I don't know a english word for it...

 

[pxc]

5: Ruiz kicked Intels ass for how many years.

WTF are you smoking? Not once did AMD ever eclipse Intel's marketshare nor did they overtake them in overall sales. A few wins in the retail sector here and there does not equate to Ruiz kicking Intel's ass.

I read a funny way of putting Ruiz's "success" in perspective a few days ago. Ruiz was paid more than the net of all AMD profits for the last 2 (or more) years combined. He gets multi-million dollar bonuses even while he's bombing, along with pretty much useless options. Ruiz is doing pretty well... for himself.

But back to the point, under Ruiz, AMD never even got back to the high market share point that Jerry Sanders earlier achieved. Ruiz became CEO in April 2002, far into the development of K8. He inherited a killer K8 and squandered it over 4.5 years after release. The K10 and ATI debacles are his failures. He is a poor leader and has failed to execute.

 

[Scali2]

i honestly dont think a lack of quads would hurt AMD *that* badly. IMHO, quads hurt more than they help because they eat up so much of AMD's fab capacity, AND they are such a bitch for AMD to make, in addition to the fact that they just dont clock all that well....

Well, thing is that they need to produce them in reasonable quantity in order to find and cure problems with the design and manufacturing.
It just takes time for them to get their quadcores to a mature state. So they can't wait that much longer now. They need to get hands-on experience with quadcores before quadcores become as commonplace as dualcores are now.
I don't think it needs to cost all that much fab capacity. They can just gradually move over their production lines from Athlons to Phenoms, starting with just one or two at first.

 

[Scali2]

I agree completely. The only point I disagree with is that AMD can't recover from this. I think it's possible, though it is hard to visualize any way for them to do so.

What I mean is that AMD can't make a 'recovery' with K10. It has lower IPC and clocks lower than Core2 now, and this will probably always be the case.
Their next chance at a better architecture is their future architecture. It's impossible to turn K10 around into a winner now. Especially at the pace that Intel is tick/tocking now.
So Shanghai and Sandtiger and all that... they are probably already a lost cause, because they use the K10 architecture.

 

[dexvx]

What I mean is that AMD can't make a 'recovery' with K10. It has lower IPC and clocks lower than Core2 now, and this will probably always be the case.
Their next chance at a better architecture is their future architecture. It's impossible to turn K10 around into a winner now. Especially at the pace that Intel is tick/tocking now.
So Shanghai and Sandtiger and all that... they are probably already a lost cause, because they use the K10 architecture.

Not really.

Willamette (first p4) was pretty horrible in comparison to the Athlon XP. The Northwood, which was just a die shrink with increased cache stomped the Athlon XP.

 

[Battleneter2]

I've always sad that K10 was a stop gap. I had --hoped-- that t would perform better then it does. I made a few bets that I lost, but I had always known that K10 was a stop gap, as I always said it was.

K10 has never been a stop gap, it was always touted as a imported product that will take on Intel's offerings, suggesting it is some sort of stop gap even sounds ridiculous.

I guess it depends how you define "stop gap", all products are stop gaps as new products at some point are immanent.

 

[harpoon]

Not really.

Willamette (first p4) was pretty horrible in comparison to the Athlon XP. The Northwood, which was just a die shrink with increased cache stomped the Athlon XP.

That's because Athlon XP clockspeeds stagnated, and it took until Northwood 'C' before Intel truly had the performance crown against Athlon XP, and that was around 2 years after Willamette was introduced.

You are deluding yourself if you think K10 will make such a comeback.

 

[savantu]

Well, thing is that they need to produce them in reasonable quantity in order to find and cure problems with the design and manufacturing.
It just takes time for them to get their quadcores to a mature state. So they can't wait that much longer now. They need to get hands-on experience with quadcores before quadcores become as commonplace as dualcores are now.
I don't think it needs to cost all that much fab capacity. They can just gradually move over their production lines from Athlons to Phenoms, starting with just one or two at first.

K10 was rushed to the market by about a year.It should have launched around mid 08.The problems we see now are inherent to such a complex design when you lack manpower and experience , but I'm sure few would have surfaced if K10 had a rigorous validation period.

Secondly , IMO they simply aimed too low with K10.The improvements made weren't enough to beat Core , but even so they probably exhausted AMD's design team.

 

[Scali2]

Not really.

Willamette (first p4) was pretty horrible in comparison to the Athlon XP. The Northwood, which was just a die shrink with increased cache stomped the Athlon XP.

Willamette was mostly horrible because DDR wasn't available, and it didn't clock very high.
With Rambus memory it performed pretty reasonably, but it was very expensive.
With SDR, it was terrible.
If you look at comparisons of Northwood and Willamette at the same speed, both on Rambus, the difference in IPC is far from impressive: http://www.digit-life.com/articles/northwood22ghz/index.html

Athlon XP was mainly stomped by brute force clockspeeds, not because Northwood had much higher IPC than Willamette.

Putting that into K10-perspective...
Large IPC-jumps never occur from respins of an existing architecture.
Northwood allowed Intel to ramp up the clockspeeds, something the Pentium 4 was custom-made for.
K10 is not made for superhigh clockspeeds. Most importantly it doesn't have the superdeep pipeline that allowed the Pentium 4 to reach clockspeeds of nearly 4 GHz even on 90 nm.

 

[savantu]

Not really.

Willamette (first p4) was pretty horrible in comparison to the Athlon XP. The Northwood, which was just a die shrink with increased cache stomped the Athlon XP.

Well , the problems with Willy were :
-cache size ( misses were awfull )
-slow memory ( low speed FSB and crappy memory )

Northwood which was basically the same doubled the cache while maintaining latency and doubled memory BW while lowering the memory latency.

 

[duby229]

WTF? What makes you think that chips on one part of a wafer are lower powered than the others. That completely defeats the purpose of standardizing manufacturing technologies.

Mobile chips are made on mobile chip wafers. Desktop chips are made on desktop chip wafers.

The difference, some chips scale better than others on the wafer. Better binning at higher GHz.

No two dies will be quite the same. Some perform better then others. Gene4rally the dies closer to the center perform better then the dies closer to the edge. Usually it is the better center dies that get binned for the mobile market. AMD has had a unified architecture through out the entire K8 generation. Griffin will be the first AMD product that was --designed-- and manufactured specifically for the mobile market.

Besides the guy asked if K8 would ever hit 4.2ghz, and I replied that ifd it ever does then Griffin will be the most likely candidate.

EDIT:
Will you guys stop confessing your love for the P4? If you guys want to wack your willy, then do it in the Intel section please... Lets try to keep this clean for the kiddies.

 

[duby229]

K10 has never been a stop gap, it was always touted as a imported product that will take on Intel's offerings, suggesting it is some sort of stop gap even sounds ridiculous.

I guess it depends how you define "stop gap", all products are stop gaps as new products at some point are immanent.

AMD never admitted it of course, why would they? But we have the benefit of hind sight from the outside. Just look at the events leading up to the announcement of K8L, and the ATi buy out and you'll see that it makes perfect sense now that it is done and over with.

 

[duby229]

...................

There really isnt much in this post that is worth replying to,. I just thought that since you took the time to write it, you deserve a reply.

Lets face the facts.....

AMD's K10 doesnt compete as well as it should. WE ALL KNOW THAT. Stop gloating about it. Put your d**k away, and stop squeezing out the lube. It's not time to party just yet.

I dont know how else to put it. I just spent the last 100 posts or so trying to say in every way that I can think that you folks are --WAAAAY-- overreacting. I tryed being polite. I tryed giving my views to contrast with yours. I tryed offering my perspective as a fan. And what happens, pretty much everybody who frequents the Intel forum comes over here to flame me. I think most of you guys need to just chill out and wait to see what happens. We --know-- that AMD is in a bad spot. We dont need you to gloat about it.

So I'm done in this thread. You guys can have your fun gloating in here, but like I said before come this time next year I'll bet that I'm closer to the mark then you guys are.

EDIT: And you "really" not a fan huh? I guess that just makes you a "selective" pessimist eh?

 

[pxc]

Who wants to start the disappointing 2008 thread early?

Have a sad look at the near future:
http://www.pcsforeveryone.com/Product/AMD/OS2350WAL4BGE

That is the only Opteron 2350 listing on pricegrabber. They have none and look at the ETA:
Low Availability
( ETA: Unknown - Best Case Q1 2008 )

Google shopping is the same. A few no-name storefronts and a couple larger online stores have no 2350s and no ETA.

IBM had their SPEC scores pulled for non-compliance (for not shipping the tested Barcelona 1.9GHz server within 90 days of submission) and the newly "channel friendly" AMD can't even get parts launched nearly 3 months ago out yet.

No major OEM (HP, Dell, IBM) has shipped a Barcelona server at any speed.
Even Tigerton servers are already shipping.

edit: does anyone know where these "hundreds of thousands" of quad core Opterons are being sold this quarter? That statement is really going to come back to bite AMD in January.

 

[vista_blista]

Those quad-core Opterons are being sold to large corporations for use in server farms. They probably don't have much use for storefronts like pcsforeveryone.com. I wouldn't doubt the majority of Phenoms will go to major OEMs first. Those are the two markets you definitely don't want to disenfranchise if there's any plan to stay in business.

 

[dudewth]

Those quad-core Opterons are being sold to large corporations for use in server farms. They probably don't have much use for storefronts like pcsforeveryone.com. I wouldn't doubt the majority of Phenoms will go to major OEMs first. Those are the two markets you definitely don't want to disenfranchise if there's any plan to stay in business.

This happens a lot in the hardware industry. The earliest adopters are companies with a relationship with the manufacturer...nothing new.

 

[pxc]

Those quad-core Opterons are being sold to large corporations for use in server farms.

Wow, you're saying that the traditional 80% of AMD server sales done by the tier one OEMs has suddenly changed with the launch of Barcelona? You should alert the trade magazines about this radical shift. LOL

Your research is second to none, or you're just making up a convenient excuse. I'll go for the second option.

 

[Mr. Bluntman]

Stop gloating about it. Put your d**k away, and stop squeezing out the lube. It's not time to party just yet.

Okay, you can stop sucking Hector's cock about now, and stop with all the Intel hating and spewing forth nonsense and insults. You're making the rest of us AMD fans look really bad. Just stop already. Please.

 

[HAF72]

meh i think this thread is turning into a flaming war, so a mod should lock this thread up

 

[vista_blista]

Wow, you're saying that the traditional 80% of AMD server sales done by the tier one OEMs has suddenly changed with the launch of Barcelona? You should alert the trade magazines about this radical shift. LOL

Your research is second to none, or you're just making up a convenient excuse. I'll go for the second option.

Then don't be so disingenuous, and skip the wiseguy rhetorical questions. Otherwise I'm going to mistake you for a troll with a lot of misdirected hostility.

 

[Manny Calavera]

Who wants to start the disappointing 2008 thread early?

Have a sad look at the near future:
http://www.pcsforeveryone.com/Product/AMD/OS2350WAL4BGE

That is the only Opteron 2350 listing on pricegrabber. They have none and look at the ETA:
Low Availability
( ETA: Unknown - Best Case Q1 2008 )

Google shopping is the same. A few no-name storefronts and a couple larger online stores have no 2350s and no ETA.

IBM had their SPEC scores pulled for non-compliance (for not shipping the tested Barcelona 1.9GHz server within 90 days of submission) and the newly "channel friendly" AMD can't even get parts launched nearly 3 months ago out yet.

No major OEM (HP, Dell, IBM) has shipped a Barcelona server at any speed.
Even Tigerton servers are already shipping.

edit: does anyone know where these "hundreds of thousands" of quad core Opterons are being sold this quarter? That statement is really going to come back to bite AMD in January.

I posted months ago about a longtime friend of mine, who works for Ingram Micro,and what he said about the K10 in the channel.

He is saying nearly the same now about Phenom.K10 Opty's are still scarce months arfter launch,for some of the biggest groups in the sales channel.

Is AMD sandbagging like Intel and Nvidia are ?, just in a very different way ?

 

[savantu]

Those quad-core Opterons are being sold to large corporations for use in server farms. They probably don't have much use for storefronts like pcsforeveryone.com. I wouldn't doubt the majority of Phenoms will go to major OEMs first. Those are the two markets you definitely don't want to disenfranchise if there's any plan to stay in business.

Really ?

Since Dell , IBM , HP and the rest of Tier 1 aren't capable to deliver K10 systems and have ETAs for Q1 2008 , who are the system providers for those large server farms ?
Your local average Joe computer shop ?

 

[Darakian]

Really ?

Since Dell , IBM , HP and the rest of Tier 1 aren't capable to deliver K10 systems and have ETAs for Q1 2008 , who are the system providers for those large server farms ?
Your local average Joe computer shop ?

Didn't some texas company make a large order back when it launched?

Edit:
I can't seem to find a link anywhere but I seem to recall something like that.

 

[APOLLO]

Didn't some texas company make a large order back when it launched?

Edit:
I can't seem to find a link anywhere but I seem to recall something like that.

http://www.crn.com/white-box/202402138

 

[vista_blista]

Until AMD fully discloses this information we can only speculate. They're saying tens of thousands have been shipped in Q3, likely to heavy-hitting Tier 1s such as Sun, HP, and IBM. And hundreds of thousands are being prepared for between December and January. I believe the Texas Advanced Computing Center recieved an order for Sun Blade X8440 supercomputer containing over 15k Barcelona processors. I also read where APPRO has been promised 48k Barcelonas in January 08. And they're a 2nd Tier OEM.

It also doesn't take much searching to find plenty more information. That's probably closer to providing a source than anything to the contrary. OR maybe some of you insiders have a direct line to Dirk Meyer and Randy Allen's office.

 

[duby229]

Okay, you can stop sucking Hector's cock about now, and stop with all the Intel hating and spewing forth nonsense and insults. You're making the rest of us AMD fans look really bad. Just stop already. Please.

Your kidding me right? Wll, I guess everybody has a different boat to float, and if you guys dont like it, then move along.

 

[darktiger]

Wow this thread has gotten long real fast. I think this year has been pretty disappointing for AMD/ATI. Intel and NVidia has out preformed AMD/ATI in almost every aspect. Even me, a diehard AMD fanboy is now running 8 Core Duo 2 or quad and only 1 X2 out of my 9 rigs....

 

[savantu]

Until AMD fully discloses this information we can only speculate. They're saying tens of thousands have been shipped in Q3, likely to heavy-hitting Tier 1s such as Sun, HP, and IBM. And hundreds of thousands are being prepared for between December and January. I believe the Texas Advanced Computing Center recieved an order for Sun Blade X8440 supercomputer containing over 15k Barcelona processors. I also read where APPRO has been promised 48k Barcelonas in January 08. And they're a 2nd Tier OEM.

It also doesn't take much searching to find plenty more information. That's probably closer to providing a source than anything to the contrary. OR maybe some of you insiders have a direct line to Dirk Meyer and Randy Allen's office.

That's not what you've said , so stick to the subject.

This are your words :

Those quad-core Opterons are being sold to large corporations for use in server farms. They probably don't have much use for storefronts like pcsforeveryone.com. I wouldn't doubt the majority of Phenoms will go to major OEMs first. Those are the two markets you definitely don't want to disenfranchise if there's any plan to stay in business.

Corporations=commercial use , sometimes HPC too.
A university is not a corporation.

IBM was forced to pull down its Spec scores because it couldn't supply the system in the 3 month period required.
=> IBM doesn't have K10s to sell.
HP has ETA of Q1 2008 => HP doesn't have K10s to sell.
Dell -> same story.

The major wins for K10 were , guess what , HPC : Cray and Sun.
The difference ? Massive.

HPC contracts , especially big ones , are excellent publicity and have very strict deadlines.
OTOH , selling hundreds of small servers to commercial use isn't really anything to brag about.

Basically ,most of the K10s were allocated to a few wins which are headline news.The rest was insignificant for IBM,HP, DELL which had to postpone their server shipments.

 

[vista_blista]

The major wins for K10 were , guess what , HPC : Cray and Sun.
The difference ? Massive.

HPC contracts , especially big ones , are excellent publicity and have very strict deadlines.
OTOH , selling hundreds of small servers to commercial use isn't really anything to brag about.

Basically ,most of the K10s were allocated to a few wins which are headline news.The rest was insignificant for IBM,HP, DELL which had to postpone their server shipments.

AMD admitted to selling a paltry *tens* of thousands in Q3... not *hundreds* of thousands. But are you saying any differently? And more important, do you have any proof?

Otherwise, it doesn't take much analysis to tell us AMD is doing very poorly with this launch. Dirk Meyer has already stated that the delays are due to "tuning the design to the technology." Right now IBM is quoting a wait of 16 to 18 days from order to delivery. And yet, I still expect them to break even this quarter, or come very close, Barcelona and Phenom notwithstanding.