Sign in with
Sign up | Sign in
Your question

More news about AMD going fabless

Tags:
Last response: in CPUs
Share
June 15, 2007 2:59:46 PM

for the record, I've been saying it for 9 months now, the sooner AMD comes clean, the faster we move on. AMD will never be a 1st rate chipmaker. It was fun while it lasted
http://www.marketwatch.com/News/Story/Story.aspx?guid=%...

More about : news amd fabless

June 15, 2007 3:18:43 PM

Quote:
for the record, I've been saying it for 9 months now, the sooner AMD comes clean, the faster we move on. AMD will never be a 1st rate chipmaker. It was fun while it lasted
http://www.marketwatch.com/News/Story/Story.aspx?guid=%...



Yeah, that Opteron is a piece of crap that should never have been released. Damn them.

Anyway, it says that IF AMD does that it will benefit Intel. Besides, why would they scrap two Fabs that are just ramping?
June 15, 2007 3:40:17 PM

Quote:
for the record, I've been saying it for 9 months now, the sooner AMD comes clean, the faster we move on. AMD will never be a 1st rate chipmaker. It was fun while it lasted
http://www.marketwatch.com/News/Story/Story.aspx?guid=%...


I don't know, that's pretty harsh. Even with AMD behind so far in the performance department, I would still say they are a "first rate chip maker". I mean, compare AMD to Cyrix. And it wasn't all that long ago that they did have the fastest CPU's that money could buy for a home PC. So, behind, yes. Second-Rate, no.
Related resources
June 15, 2007 3:50:01 PM

i think that it is most likely they will try to get more companies in their process development group. If they need it they can even share fabs. But going fabless, I don't think so.
June 15, 2007 3:57:39 PM

Wonder which branch of AMD they are talking about, CPU or GPU? I think a little elboration on what pieces of silicon they're talking about would be helpful in this prediction.

AMD has superior engineers to Intel. Intel has capacity and money on AMD. Don't care how good of an engineer you are, if someone can build more than you, and spend money to market it, your product won't be very successful for far too long. Like Germany in WWII superior tanks, but America could outproduce them.
June 15, 2007 4:19:06 PM

Quote:
I agree I dont think AMD will ever top Intel again. I cant even believe they were able to spank Intel for the 3 years they did. AMD has always been number 2 and always will be. They just dont have the engineers and capacity Intel has. However we dont want AMD to go away because then their will be no competition and with no competition Intel can go back to charging what they want.



AMD has Dirk Meyer, and Alpha engrs, not to mention some Itanium engrs that hired from Intel/HP.

There's nothing wrong with being number two when there are only two competitors. AMD has done and will more than likely continue to do a great job making server procs that trickle down to the desktop.

I really hope they can get high clocks out of Griffin so they will have a more competitive product on the mobile front. It may even make sense to add a single core OC mechanism so that one core can turn off and the other core can OC. Intel says they are doing something like that.

Most mobile SW is single threaded anyway.

I heard from a guy on extremesystems that had an early Barcelona ES that was in his words 10% faster than Kentsfield. Scientia has a link to the thread. This was back in Feb so I think that helps prove that the shipping rev will be "40% faster on a variety of workloads."
June 15, 2007 4:38:17 PM

Quote:
Anyway, it says that IF AMD does that it will benefit Intel. Besides, why would they scrap two Fabs that are just ramping?


I note that the article gives no source concerning AMD going fabless, only a speculation that said that if Goldman Sachs' statment is correct, then "AMD's likely move to an outsourced business model" would benefit Intel. That's a big "if", and it does nothing to tell the possible benefits to AMD, what outsourcing or type of outsourcing or quanity of outsourcing, or anything else. This looks more like a bit of trash reporting than anything else to me.
June 15, 2007 4:43:41 PM

Quote:

Most mobile SW is single threaded anyway.

Could you explain that a little more?

Quote:

I heard from a guy on extremesystems that had an early Barcelona ES that was in his words 10% faster than Kentsfield. Scientia has a link to the thread. This was back in Feb so I think that helps prove that the shipping rev will be "40% faster on a variety of workloads."


I heard from a guy on the THG forums that the early Barcelona ES didn´t work at all and the current spetting works only at speeds below 1.8 Ghz.
June 15, 2007 5:05:59 PM

Quote:

AMD has superior engineers to Intel. Intel has capacity and money on AMD. Don't care how good of an engineer you are, if someone can build more than you, and spend money to market it, your product won't be very successful for far too long. Like Germany in WWII superior tanks, but America could outproduce them.

That is a weird comparison. I somewhat resent it because it does seem a little off. The important characteristics of tanks were speed, maintainablitiy and cost of production. American and even more so russian tanks excelled at maintainability and cost of production. The german machines were to complicated, lacked spare parts and, later in the war, got way to heavy. Germany concentrated on the wrong aspects of their tanks and got what they deserved for that.
June 15, 2007 5:13:25 PM

I'm confused?...

Everybody turns on AMD because they suck at producing/fabrication?...that's the reality of their conundrum.

You can't deny that they(AMD) have put together great processors, so they are competitive in terms of putting ideas on the table.

The issue comes with getting that product to market and then backing it. Intel spent majority of their R&D on fabrication during the netburst era. Now we are seeing the fruits of that labor as well as the maturity of their PIII architecture. The architecture in C2D is just a culmination of several technologies.

AMD never really caught up to Intel, they just capitalized on the void brought about by Intel's R&D for future product.

AMD has been sprinting the last few years in their attempt to catch up in production technology. Intel has been very calculated. It's no accident that C2D over clocks so well. Enthusiasts aren't a key financial market unless you account for the clout they maintain amongst friends and family. Without that egg in their basket I'm sure AMD's financial progress wouldn't have come to such a braking hault.

So let's be real...AMD's achilles is in their fabrication process. They have a great idea on paper, it can't be realized because of their fabrication issues. If they can get some help on that front(maybe from IBM?) then you'll see a huge difference in the market.

I'm gonna cross my fingers and hope that it doesn't kill the edge that has allowed AMD to create a competitive scene with Intel.
June 15, 2007 5:21:09 PM

While this has nothing to do with PCs, the American tanks throughout WWII were junk.

In fact the tanks that the US sent to the Soviet Union via Lend Lease were mostly used to tow various items and not combat vehicles due to their significant inferiority.

Germany lost because of shear numbers.

The Russian tanks, however, were of very good quality and by the mid part of 1942 were often as good or better than their German couterparts. The superiority often swung back and forth.

Germany and Japan combined are tiny and would fit inside a small part of the US. There is no way those two countries could take on the world. Their smaller allies had no real interest in the war from their people.

The massive bombing campaign from the Westen Allies on Germany is what kept their industry in check and prevented the ability to produce enough equipment or spare parts.
June 15, 2007 5:26:48 PM

Quote:

Most mobile SW is single threaded anyway.

Could you explain that a little more?

Quote:

I heard from a guy on extremesystems that had an early Barcelona ES that was in his words 10% faster than Kentsfield. Scientia has a link to the thread. This was back in Feb so I think that helps prove that the shipping rev will be "40% faster on a variety of workloads."


I heard from a guy on the THG forums that the early Barcelona ES didn´t work at all and the current spetting works only at speeds below 1.8 Ghz.


How many people are running Photoshop on a laptop? Maybe a desktop replacement but do you note the different terminology?

This guy at xtremesystems is a trusted member over there and that gives him more credibility.
June 15, 2007 5:35:06 PM

Quote:

How many people are running Photoshop on a laptop? Maybe a desktop replacement but do you note the different terminology?

The question is, don´t they because their processors are limiting them or don´t they because they don´t want to? I´m willing to pay for the option to use it and so are a lot of people.

Quote:

This guy at xtremesystems is a trusted member over there and that gives him more credibility.


That makes his info even more reliable than AMDs official one. Yet, i will doubt his word unless i see proof since everything i´ve read so far points that the new chip will be only 40% faster in a few select benchmarks and may be just able to compete clock for clock.
June 15, 2007 5:57:49 PM

Quote:
While this has nothing to do with PCs, the American tanks throughout WWII were junk.

In fact the tanks that the US sent to the Soviet Union via Lend Lease were mostly used to tow various items and not combat vehicles due to their significant inferiority.

Germany lost because of shear numbers.

The Russian tanks, however, were of very good quality and by the mid part of 1942 were often as good or better than their German couterparts. The superiority often swung back and forth.

Germany and Japan combined are tiny and would fit inside a small part of the US. There is no way those two countries could take on the world. Their smaller allies had no real interest in the war from their people.

The massive bombing campaign from the Westen Allies on Germany is what kept their industry in check and prevented the ability to produce enough equipment or spare parts.


Umm the M1 Sherman tank was a great fighting tank, wasnt on par with the Tiger or the King Tiger, but those tanks where not the main tanks in germany because of there cost to built the Sherman dealt with the Panzer which was evenly matched to the M1 sherman tank. Sheer numbers didnt beat Germany back in WW2, who ever told you that was a fool. It was the will of the troops and the fact we controlled the air. The tanks of WW2 where all rather great, also Russia didnt use our tanks in battle for a totally diffrent reason, it meant a barrle change to use our tanks, because they didnt have shells that fit the Sherman, thus it would be pointless to use.


Now back on topic

AMD isnt gone yet, Intel has nothing in the value segment to compete with AMD, there Core Microarticture OC's well but is cache dependant as seen by the benchmarks of the new E2140 and E2160 where both lagged behind the X2 3600 and X2 3800. In the value world Intel still can not compete, and the Core micro articture needs large cache amounts or insane speed to beat AMD still. Also not with out ocing, the E6600 and the x2 6000 are evenly matched cpus and they cost the same, same goes for the chips down the line, to most users the X2 and the Core2 are evenly matched chips, and in the sub 50 dollar super value chips intel is still touting there Celeron D vs the Sempron which isnt anywhere that Intel stands a chance and Sempron is being sold in Dell, HP, Compaq, eMachines ect, so id say AMD is doing fine, and how soon have we forgotten AMD's glory of the Athlon beating the Pentium 3 and intel never really recovering till the Core2.
June 15, 2007 6:05:37 PM

Quote:
I'm confusedSo let's be real...AMD's achilles is in their fabrication process. They have a great idea on paper, it can't be realized because of their fabrication issues. If they can get some help on that front(maybe from IBM?) then you'll see a huge difference in the market.


Well, let's see. Intel has lot's of Fab capacity. Maybe AMD can contract with Intel to produce their chips. :lol:  Hey, it can happen. Imagine how ironic that would be.
June 15, 2007 6:10:25 PM

Quote:
While this has nothing to do with PCs, the American tanks throughout WWII were junk.

In fact the tanks that the US sent to the Soviet Union via Lend Lease were mostly used to tow various items and not combat vehicles due to their significant inferiority.

Germany lost because of shear numbers.

The Russian tanks, however, were of very good quality and by the mid part of 1942 were often as good or better than their German couterparts. The superiority often swung back and forth.

Germany and Japan combined are tiny and would fit inside a small part of the US. There is no way those two countries could take on the world. Their smaller allies had no real interest in the war from their people.

The massive bombing campaign from the Westen Allies on Germany is what kept their industry in check and prevented the ability to produce enough equipment or spare parts.


Yes, the American tanks were utter crap, when compared with the German ones. They used sandbags fo armour for christ's sake!
The Germans' undoing, however, was their pickyness about quality. The components on their tanks were incredibly well engineered, and would have lasted for a long time. This meant, however, that the tanks were expensive and slow to build and spare parts were in short supply.

I'm not sure about the Russian tanks, but I do know that in the German tanks, fuel would freeze in the lines in Russia. This obviously did not happen to Russian tanks, having been designed by the Russians :lol: 
There is a Russian proverb: "When all our other generals are defeated, we still have generals January and Febuary."

There is way a small country like Germany could achieve domination. Hitler was just too insane and too much of a dictator for that to happen. Attacking Russia was his biggest mistake: 90% of German army casualties were in Russia. He also made several other critical mistakes. In order to attack Russia, he ordered the cessation of the Blitzkrieg attacks on Britain. He halted submarine production, giving Britain naval superiority. He would also insist on the planning of the war, not allowing his generals to conduct it as they saw fit.
Germany nearly had its own atomic bomb by the end of the war.
Europe as we know it only exists because Hitler was a dumbass.
June 15, 2007 6:21:06 PM

Quote:

So let's be real...AMD's achilles is in their fabrication process. They have a great idea on paper, it can't be realized because of their fabrication issues. If they can get some help on that front(maybe from IBM?) then you'll see a huge difference in the market.

I'm gonna cross my fingers and hope that it doesn't kill the edge that has allowed AMD to create a competitive scene with Intel.


That has to be the most negative post in this thread. How can you say they can't fabricate processors? They showed them already (barcelona). Even the mighty Intel had to recall a rev of Core 2 last year because of a problem.

I say again that Anand got AM2 right before it came out and it sucked real bad. It couldn't even equal 939, but when they came out the perf was there and that was only a redesign of the IMC. Barcelona is an entirely new chip basically. I expect it to come out in July at the reported perf levels.
a b à CPUs
June 15, 2007 6:26:21 PM

Quote:
for the record, I've been saying it for 9 months now, the sooner AMD comes clean, the faster we move on. AMD will never be a 1st rate chipmaker. It was fun while it lasted
http://www.marketwatch.com/News/Story/Story.aspx?guid=%...


Oh puh-leez! The linked article says absolutely nothing about AMD's fabrication process and is really about...
Quote:
Intel Corp. (INTC : Intel Corporation) was upgraded to buy from neutral at Goldman Sachs...
Given there is no specific verbiage identifying AMD's fabs, only an allusion to AMD's move to an "outsourced model", tying that back to their Fabs is pure speculation and market talk. From all the news that read, including the transcript to the 4/19 1Q AMD conference call, the idea of outsourcing is a run-on from talk of an "asset light model". Exactly what an "asset light model" is has been the fodder of speculation and assumption since, so take it for what it's worth.

Not only that, wasn't there a big stink last year about AMD opening up a new fab in the Luther Forest Technology Park in Saratoga County in upstate New York. As well as all the hype over AMD converting Fab30 to 300mm and renaming it Fab38. Along with a significant long term investment in all the Dresden Fabs?

All things considered, it seems to me that AMD is committed to their Fabs with their fab process being right in line with their processor roadmap and long term plans to synergize technology from the ATI merger.
June 15, 2007 6:34:30 PM

Yeah people will read something and take it all out of context and flip out about it. Ask a Jehova's witness sometime.

BTW chunky I am dying laughing at turpit's quote in your sig. :lol: 
June 15, 2007 6:37:46 PM

Quote:
Barcelona is an entirely new chip basically. I expect it to come out in July at the reported perf levels.


So you expect it to suck too?
June 15, 2007 6:49:51 PM

Quote:


Yeah, that Opteron is a piece of crap that should never have been released. Damn them.

Anyway, it says that IF AMD does that it will benefit Intel. Besides, why would they scrap two Fabs that are just ramping?

Opty's are crap? Right.... And what happened to your undying love of AMD, huh baron?
June 15, 2007 6:56:50 PM

Quote:


Yeah, that Opteron is a piece of crap that should never have been released. Damn them.

Anyway, it says that IF AMD does that it will benefit Intel. Besides, why would they scrap two Fabs that are just ramping?

Opty's are crap? Right.... And what happened to your undying love of AMD, huh baron?

Never heard of sarcasm huhn?
a c 115 à CPUs
a b À AMD
June 15, 2007 6:58:44 PM

Quote:
for the record, I've been saying it for 9 months now, the sooner AMD comes clean, the faster we move on. AMD will never be a 1st rate chipmaker. It was fun while it lasted
http://www.marketwatch.com/News/Story/Story.aspx?guid=%...


Company shows interest in opening more factories during briefing on $3.2B chip fab proposal
Quote:
Friday, May 4, 2007
SARATOGA SPRINGS -- Company executives for Advanced Micro Devices Inc. meeting in the Spa City to discuss the $3.2 billion computer chip factory planned for Malta announced Thursday that the company may be interested in opening a second, or perhaps third, factory in the Luther Forest Technology Campus.


Quote:
However, even under the best-case scenario, the first shovel wouldn't hit the ground -- at the very earliest -- until spring 2008.... Including tax breaks and infrastructure improvements, the state's package for AMD totals $1.2 billion.
June 15, 2007 7:25:18 PM

I have to admit, this is the first I have even heard speculation of AMD going fabless. And I call it exactly that, speculation. While I have read various stories about AMD redoing fabs, and looking to open more. So if they really had plans to do that, then why go fabless?

Also, what people have to remember is that computers as a whole are still a new technology, I have been into them since before the x86 days, not seriously into them until the 386 days but still. And I have watched AMD go from a completely unheard of company to a name synonymous with Intel. Now they also held the performance crown from the time pc's hit 1 ghz up until the core architecture actually hit the stores. Saying their 15 mins of fame is up is about the most stupid thing I have read on all of the threads. And to be honest, I pray to God that AMD doesn't start believing that because if they end up dying, we ALL pay. And big, there would be absolutely nothing to stop intel from charging $2000 for a new proc, and if you want a new computer, you will pay it. Even if Barcelona and Griffin don't deliver C2D killing performance, if they can match clock for clock, that's a step up, and if later revs can beat the C2D's then great. Or even the next gen of procs after that. As long as they have a competitive product then we all win.
June 15, 2007 7:36:41 PM

Someone correct me if I'm mistaken, but isn't one of the conditions of AMD's x86 license is that they can't outsource more than 20%?
June 15, 2007 7:43:00 PM

Quote:
for the record, I've been saying it for 9 months now, the sooner AMD comes clean, the faster we move on. AMD will never be a 1st rate chipmaker. It was fun while it lasted
http://www.marketwatch.com/News/Story/Story.aspx?guid=%...


I don't know, that's pretty harsh. Even with AMD behind so far in the performance department, I would still say they are a "first rate chip maker". I mean, compare AMD to Cyrix. And it wasn't all that long ago that they did have the fastest CPU's that money could buy for a home PC. So, behind, yes. Second-Rate, no.

In terms of quality, yes, in terms of capacity..well, they are well less than 50% of Intels capacity. But second rate is harsh...Via would be more crrectly termed "second rate".
a b à CPUs
June 15, 2007 9:05:33 PM

Quote:
and how soon have we forgotten AMD's glory of the Athlon beating the Pentium 3 and Intel never really recovering till the Core2.


Man am I sick of people revising history because they can only manage to remember 4 years back. Sure the original Athlon came along and beat the PIII in breaking the GHz barrier. So what, Intel came out with the P4 and they were back in the game. Unfortunately they also dragged RAMBUS along with it which made their systems expensive. Once Intel dropped RAMBUS in favor of DDR though, the Athlon XP was once again the under-performing budget CPU. So I would call that a tit-for-tat race. Intel's problem was trying to spread too small a piece of butter over too large a piece of toast. In otherwords they hung on to NetBurst too long. After Northwood, Intel should have released a new architecture. Unfortunately Intel actually believed that they could ramp NetBurst up to 10GHz. It wasn't until Prescott that they realized this wasn't going to happen (they couldn't refine their process quick enough to make it possible). Of course what happened right around the time of Prescott? A64. This is really the era in which Intel had nothing to offer but the embarrassment that was the Extreme Edition. So it was what 3 years that Intel couldn't compete with AMD, not the 7 years you are talking about. So the moral to the story is don't exaggerate AMD's superiority based on 3 years.

Quote:
Also not with out ocing, the E6600 and the x2 6000 are evenly matched cpus and they cost the same


Yeah, and how long did the AMD boys have to wait for this? It took over 7 months to have an answer to Intel's fourth best (cause QX6700 was out already). Not to mention conveniently removing Intel's OC'ing ability allows you to sidestep the thorny reality that AMD's architecture can't handle anymore clockspeed (and at the moment that's all they have to improve performance on their present architecture), there's no headroom left in this architecture. They got caught flat footed, just like Intel did with A64. Turn around is fair play.

Quote:
there Core Microarticture OC's well but is cache dependant as seen by the benchmarks of the new E2140 and E2160 where both lagged behind the X2 3600 and X2 3800.


Yeah, where have I heard the cache crutch before? HMMM, wasn't it Sharikiou and Scientia the two biggest AMD fanboys ever in the history of ever. Sure Core 2 runs well with 4MB cache, but it doesn't run much different with 2MB of cache either (E6300, E6400). The Allendale CPU's are nearly the match for the Conroes when clockspeeds are the same.

Quote:
and in the sub 50 dollar super value chips Intel is still touting there Celeron D vs the Sempron which isnt anywhere that Intel stands a chance and Sempron is being sold in Dell, HP, Compaq, eMachines ect


I would like to argue this point, but I got nothing. AMD has always been great with the value segment. Don't get me wrong, I love AMD and I think they have the ability to be every bit as good as Intel, but they aren't right now. I am happy for AMD that they can align their name with eMachines and their ilk :roll: . Sadly all the ground they made up in the server market with the Opteron over the last two years, has disappeared in the last 3 quarters. How fickle the IT experts are. It took them a little while to warm up to what was a superior product (Opteron), but as soon as Intel was back with a superior product, sales for 2P servers shoot right back up to where they were. I guess they are just more comfortable with Intel. Not saying it's right, just what happened.

I don't like Intel anymore than AMD, I like superior performance. So Intel is the only game right now. What I dislike is inaccurate information, which is what all the refuted quotes are all about. I have seen so many post lately where just because AMD was the unchallenged performance leader since the Opteron/A64/FX, so people just paint AMD as the best since the first Athlon. Don't be a revisionist, if you can't recall history as it was, don't make unsubstantiated claims.
June 15, 2007 9:38:34 PM

Quote:

So let's be real...AMD's achilles is in their fabrication process. They have a great idea on paper, it can't be realized because of their fabrication issues. If they can get some help on that front(maybe from IBM?) then you'll see a huge difference in the market.

I'm gonna cross my fingers and hope that it doesn't kill the edge that has allowed AMD to create a competitive scene with Intel.


That has to be the most negative post in this thread. How can you say they can't fabricate processors? They showed them already (barcelona). Even the mighty Intel had to recall a rev of Core 2 last year because of a problem.

I say again that Anand got AM2 right before it came out and it sucked real bad. It couldn't even equal 939, but when they came out the perf was there and that was only a redesign of the IMC. Barcelona is an entirely new chip basically. I expect it to come out in July at the reported perf levels.



...in what way was that negative? Baron, the third quarter is only a little over a week away with no real engineering samples available. Although assumptions are subjective, you'd have to assume that they're having issues with ramping up. Everybody knows they've had issues with SOI at 65nm. Why would it be different with the K10 architecture? Simply put, their fabrication process is holding them back at this point.

They would have been much more competitive with C2D if they had been able to ramp up processor speeds with K8. However, their fab. process held them back. The biggest advantage C2D has is their ability to run sse in one processor cycle. Obviously, this is what holds back K8, reguarding competition, in majority of benchmarks. The fix?...ramp up processor speeds, but can they?...again, no.

So this time I won't be so indirect. Fab technology is what killed AMD's ability to compete with C2D, assuming Intel wouldn't have ramped up their processor speeds in response. Fab technology is also the center piece surrounding K10 and whether or not it will compete directly with Penryn/any other portion of Intel's product line. Point being, they have a great idea on paper, but it won't be worth anything if Intel runs it into the ground through a Mhz war.


EDIT: grammatical, I'm horrible at proof reading.
June 15, 2007 9:42:26 PM

Quote:

Anyway, it says that IF AMD does that it will benefit Intel. Besides, why would they scrap two Fabs that are just ramping?


Sometimes it´s not a question of would or want. Fabs cost money and that´s something AMD doesn´t have. While i´m certain that the Fabs would start to generate money at some point, it is a possibility that AMD needs the money NOW and not in a few months.
a c 115 à CPUs
a b À AMD
June 15, 2007 9:43:01 PM



Yeah . . gotta believe those analyists at Goldman Sachs

Quote:
Goldman Sachs report on Enron, October 24, 2001:

"We do not believe that management has done anything wrong..."

Goldman's analysts continued to keep Enron on its "recommended list," Goldman's highest rating....

"...you ain't seen nothin' yet. Enron's growth businesses all appear exceptionally well positioned to generate major and perhaps breakout earnings..."


http://manage.kenlayinfo.com/upload/exbln036098_15152.p...
June 15, 2007 9:48:42 PM

Quote:


Sometimes it´s not a question of would or want. Fabs cost money and that´s something AMD doesn´t have. While i´m certain that the Fabs would start to generate money at some point, it is a possibility that AMD needs the money NOW and not in a few months.




It almost makes sense from a financial stand point.

1) Sell Fabs for $XXXXX
2) Use $XXXX for product development to maintain competition
3) Lower short term margins through use of 3rd party fabs
4) Rebuild assets for long term development


That's the best logic I can put together...provided they do their financial research in terms of how this will pan out



EDIT: Fixed Quote
June 15, 2007 9:53:41 PM

Quote:

It almost makes sense from a financial stand point.

1) Sell Fabs for $XXXXX
2) Use $XXXX for product development to maintain competition
3) Lower short term margins through use of 3rd party fabs
4) Rebuild assets for long term development


That's the best logic I can put together...provided they do their financial research in terms of how this will pan out



EDIT: Fixed Quote


Nice idea. There is one problem though. If they sell something - heck, if they sell anything at all - Stanley & Morgan get first dibs. So if they sell it, it´s mostly to reduce their debt.
June 15, 2007 10:07:31 PM

good point...I forgot about the original terms of agreement.

Seems like somebody really dropped the ball signing that contract. So there it is, all this speculation is just horse shiat. :) 


I really don't believe AMD would consider getting rid of those fabs. I think K10 will just compete with Intel, nothing great, just compete.

That should be enough to give them the time to sink into R&D with future product.
June 15, 2007 11:11:25 PM

Quote:


Yeah, that Opteron is a piece of crap that should never have been released. Damn them.

Anyway, it says that IF AMD does that it will benefit Intel. Besides, why would they scrap two Fabs that are just ramping?

Opty's are crap? Right.... And what happened to your undying love of AMD, huh baron?

Never heard of sarcasm huhn?
No, I have, I just don't like what has been going on surrounding you.
June 16, 2007 12:34:51 AM

Quote:
for the record, I've been saying it for 9 months now, the sooner AMD comes clean, the faster we move on. AMD will never be a 1st rate chipmaker. It was fun while it lasted
http://www.marketwatch.com/News/Story/Story.aspx?guid=%...



Yeah, that Opteron is a piece of crap that should never have been released. Damn them.

Anyway, it says that IF AMD does that it will benefit Intel. Besides, why would they scrap two Fabs that are just ramping?

this is long term planing, not overnight. Fabs will go on for the next 5 years but there be no new ones to be built...that;'s the meaning, not shutting down now!
June 16, 2007 2:25:42 AM

"Umm the M1 Sherman tank was a great fighting tank, wasnt on par with the Tiger or the King Tiger, but those tanks where not the main tanks in germany because of there cost to built the Sherman dealt with the Panzer which was evenly matched to the M1 sherman tank. Sheer numbers didnt beat Germany back in WW2, who ever told you that was a fool. It was the will of the troops and the fact we controlled the air. The tanks of WW2 where all rather great, also Russia didnt use our tanks in battle for a totally diffrent reason, it meant a barrle change to use our tanks, because they didnt have shells that fit the Sherman, thus it would be pointless to use. "

You're just wrong here the main German tank was the Panther which was vastly superior to any Sherman during the European part of the war. In practice we would loose 4-5 Sherman's trying to take out 1 Panther. The Russian T-34 was also better tank.

I don't know where you got you facts but any US vet that was there will tell you the same thing along with many other books and online sources. Until the Pershing came about we didn't have a solid tank on the battlefield. Oh yeah there was no M1 Sherman I think what you meant was the M4A1.
June 16, 2007 2:59:51 AM

The original article said (live data removed):
Quote:


INTC was upgraded to buy from neutral at Goldman Sachs, which said rival Advanced Micro Devices Inc.'s AMD likely move to an outsourced business model will create significant benefits for Intel over the longer-term. Goldman said such a move by AMD would make it easier for Intel to "retain a sustainable product advantage," given the difficulty AMD will face in optimizing technology. "Capacity driven cyclical downturns should be much more muted given that only one player would need to cut capital expenditure to fix excess supply," the broker added.


Firstly, nowhere is it defined in this article what they mean by 'an outsourced business model'. It is quite often the case that it is sensible to outsource certain aspects of your business. Everyone seems to have jumped to the conclusion that it is fabs that will be outsourced, and so it might be, but they could also be outsourcing, say, packaging, test engineering, etc.

Of course, given that they don't quote a source on their view that AMD are likely to move to an outsourced business model, you can't really tell whether they have understood and how much they have understood.

Equally, given the 'AMD likely to Intel a favour' tone of one sentence, you might have expected some discussion of why a company felt moved to do a big competitor a favour. I'm guessing that they would have had to have continued 'because the cashflow implications of not doing are too scary', but they should have stated that, with figures and references, not have me guess.

Given the sentence "Capacity driven cyclical downturns should be much more muted given that only one player would need to cut capital expenditure to fix excess supply," you might be tempted to conclude that the analyst who wrote that had limited knowledge of the semi industry. It is one that has managed to bounce around the over/undercapacity tradeoff for years, without showing much evidence of being able to control cyclical overshoots, whichever model (outsourcing or not) has been used.

The trouble is that these are big numbers and a fundamentally unstable system and if you are really clever in outsourcing you just succeed in bamkrupting the company that you outsource to. If you are not, you just end up paying over the odds.

Quote:
Someone correct me if I'm mistaken, but isn't one of the conditions of AMD's x86 license is that they can't outsource more than 20%?


May well not be relevant: Nowhere does it say that x86 products are involved. They could well (& it would probably make more sense to) outsource other stuff that doesn't need the highest of high tech fabs.

If you try to outsource the fab of 'bleeding edge' parts, there is always the difficulty of finding someone that you can outsource to. If you look at people like Chartered they tend to be that bit later in introducing the latest, smallest geometry, parts. If you then throw in a very non-standard process like SOI, you may well find none of the merchants can deal with it. So, you have the choice of finding someone, probably one of your nominal competitors, who already produces parts in a similar process, redesigning your parts for a different process, getting even further behind the curve or bankrolling process development at a merchant. Not one of these is attractive, but maybe what you are looking at currently is even less attractive.

Quote:

"Capacity driven cyclical downturns should be much more muted given that only one player would need to cut capital expenditure to fix excess supply,"


Hmmm, it would be interesting to see if anyone from AMD would sign up to the position that "one of the problems that we had had is that we have been supplying way too many chips". My feeling is that is the exact opposite of the way AMD see things and the AMD POV is also the way people at the end of the channel see things. The other (bizarre) interpretation is that Intel, with the massive investment, would want to cut their chip supply and mothball some fabs to make life easier overall. I don't think that is likely either.

Quote:
AMD isnt gone yet, Intel has nothing in the value segment to compete with AMD, there Core Microarticture OC's well but is cache dependant as seen by the benchmarks of the new E2140 and E2160 where both lagged behind the X2 3600 and X2 3800. In the value world Intel still can not compete, and the Core micro articture needs large cache amounts or insane speed to beat AMD still. Also not with out ocing, the E6600 and the x2 6000 are evenly matched cpus and they cost the same, same goes for the chips down the line, to most users the X2 and the Core2 are evenly matched chips, and in the sub 50 dollar super value chips intel is still touting there Celeron D vs the Sempron which isnt anywhere that Intel stands a chance and Sempron is being sold in Dell, HP, Compaq, eMachines ect, so id say AMD is doing fine, and how soon have we forgotten AMD's glory


Well AMD hasn't gone yet, fortunately, I'd agree with you on that. As for the rest, I think you are overstating the case slightly.

I would say the E2140 and E2160 do compete with the lower end of the X2s. Just because they are outbenchmarked at standard speeds doesn't mean that they don't compete. There are other factors such as overclocking and platform stability and availability of motherboards and chipsets to take into account and while I can see why a person might choose either one or the other, I think that describes a competitive marketplace.

The Celeron D is still there, but not much longer. It seems from a marketing point of view that Pentium is the new Celeron and Celeron will be quietly killed. And not before time!

The real problem comes with the statement "AMD is doing fine". Currently AMD is selling low to mid range chips at what, by historical standards, are fire-sale prices. Now you can argue (I would) that while Intel was pushing the Cr*pburst architecture there wasn't real, intense, competition and so in that period, prices stayed too high. Therefore, this recent period when prices came down was more like a readjustment to a more normal pricing regime.

The question is, however, is it a regime in which AMD is making enough money to plow back into R & D, process development and capacity and techonology enhancement? My guess is that they are not:

- The low end chips are, well, as cheap as chips, and its hard to make much on cheap chips, unless you really sell mega volumes.

- The high end chips are having a difficult time selling against the revised Intel architecture chips. You don't make any money from the chips that you don't sell. And, if you cut the prices too far, you still don't make money.

This leaves AMD in a bit of a bind: Until they get Barcelona and related chips out of the door and in volume, they won't have a product for which they can charge a premium price and sell in huge numbers. They not only need it now, and need it to be good, they have needed for more than the last six months. Yes, AMD stay in the business selling stuff off at fire sale prices, but that doesn't allow them to invest in the future.

Looked at historically (and this might change, but I have no reason to think that it will) AMD have done well against Intel when AMD has had the superior architecture in Instructions per Clock. This has meant that, in spite of Intel being ahead in processing, AMD's superior architecture has enabled them to compete well. Right now, I don't have the confidence that Barcelona and related parts' ICC is beyond that of Conroe and related parts, because I don't see the evidence. Sure, AMD say it will all be all right, but numbers would convince me more and useful numbers (not 4 cores against two, not number that just apply to SSE special cases and have limited applicability to general computing) so I'd really like to see good benchies from AMD parts. But all I hear is 'marketing' and vague rumours about initial problems that will be corrected shortly limiting the ability of current parts to truly perform.

We shall see! These certainly are interesting times.
June 16, 2007 4:26:28 AM

Quote:
"Umm the M1 Sherman tank was a great fighting tank, wasnt on par with the Tiger or the King Tiger, but those tanks where not the main tanks in germany because of there cost to built the Sherman dealt with the Panzer which was evenly matched to the M1 sherman tank. Sheer numbers didnt beat Germany back in WW2, who ever told you that was a fool. It was the will of the troops and the fact we controlled the air. The tanks of WW2 where all rather great, also Russia didnt use our tanks in battle for a totally diffrent reason, it meant a barrle change to use our tanks, because they didnt have shells that fit the Sherman, thus it would be pointless to use. "

You're just wrong here the main German tank was the Panther which was vastly superior to any Sherman during the European part of the war. In practice we would loose 4-5 Sherman's trying to take out 1 Panther. The Russian T-34 was also better tank.

I don't know where you got you facts but any US vet that was there will tell you the same thing along with many other books and online sources. Until the Pershing came about we didn't have a solid tank on the battlefield. Oh yeah there was no M1 Sherman I think what you meant was the M4A1.


Eh, interesting how the conversation turned to WW2 tanks. I love it.

You know, Guderian (the father of modern tank warfare) wrote that the three aims of tank design were:

1) great mobility
2) heavy armament
3) heavy armour

Of these three aims, mobility was the most important; and heavy armour was the least important. That's because Guderian imagined tanks as fulfilling a primarily offensive role -- i.e., they were supposed to run roughshod over enemy communications after initial break through had been achieved.

On this account, Guderian states over and over again in his book that German tank development was very far from satisfactory from a tactical standpoint, mainly because German designers starting with the Tiger I and even to some extent with the PzKpfw V had decided to sacrifice mobility for firepower and thick armor.

The Panther and Tigers I and II were prone to mechanical failure, mainly because of stress to transmission and suspension systems. They also had very limited range; and no machine guns (at least initially). For all of these reasons, these weapons were very unsuitable in an exploitation role, which was shown time and time again during offensive operations on the Eastern front.

As for the high loss figures for M4s vs German armour in the Western theatre, this can largely be attributed to the way Tigers and Panthers were utilized in the West. That's to say, they were mostly used in a defensive role. In this role, they were quite successful since their high velocity L75 and 88mm guns could rip thru the frontal armour of the M4 even at great distances. M4 losses to towed weapons, especially towed 88s, were also very high, and for exactly the same reason. The point is, defensive warfare masked the Panther's and Tiger's main weakness: its lack of sufficient mobility. In open field engagements, on the other hand, even Panthers were prone to being flanked and disabled by their more nimble adversaries.
June 16, 2007 5:34:42 AM

So, does this mean AMD is going to start making tanks? 8O

:lol:  :lol:  :lol: 
June 16, 2007 12:44:29 PM

They would ship when the war was over :D 
June 17, 2007 8:23:43 AM

Quote:
the new E2140 and E2160 where both lagged behind the X2 3600 and X2 3800. In the value world Intel still can not compete,.


This is simply flat out factually incorrect.

Here is a review of the x2-3600 x2-3800 head to head with the two cheap Dual Core Intels, and the 2160 wins virtually every benchmark.

http://www.xbitlabs.com/articles/cpu/display/pentium-e2...

The x2-3xxx do win modestly in manygames, but in virtually everything else the 21xx series wins, often by fairly large margins...

Oh and yes, the Intels overclock well over 3 ghz, something unlikely on an x2-3800.
June 17, 2007 10:42:30 AM

It's funny to read how people think outsourcing must mean that AMD will close their own factories. It isn't possible that they just expect a greater demand than they can fullfil and to not get into this situation again?

After all this might also explain the wish to build more factories, so they can get more of the profit in the future, when they don't need to outsource that much of their production.

Why do we have this much treads of speculation that AMD might go out of bussiness, after all they are still competitive.
June 17, 2007 3:30:09 PM

Although I like your point and hope that it is true, it doesn't line up with their actions in the last financial quarterly conference call.

In that call, they stated that the curent conversion to 300mm of an older fab would be slowed. They also stated that they would invest less into R&D. Finally, they made special notice of the fact that they CAN outsource much more to their partners under current agreements (a good indicator that they will - at a time when they don't have a huge reason to see more demand). Maybe things will change after Barcelona - we'll have to wait and see.

If AMD is fully outsourced in terms of mfg (probably a long ways off, who knows), my question is what are they? IBM does a huge bit of their research for them (and Intel does as well to some degree), they don't do very strong marketing, and if they aren't even making their own chips...what are they? Are they even needed at that point?
!