AMD Bulldozer and Llano going to be delayed? GF 32nm troubles?

[veri745]

Bulldozer tapes out

http://www.xbitlabs.com/news/cpu/di...apes_Out_First_Bulldozer_Microprocessors.html

Not really news. Both the "BD tapes out" info and the "32nm delayed" info (that started this thread) were gleaned from the AMD quarterly conference call and stated by Dirk.

 

[taltamir]

Charlie Demerjian from Semiaccurate had hinted about process issues, but from what I concluded, it didn't seem to be a big thing.

but charlie from semi accurate is an vindicative and clueless idiot who just runs his mouth.

Idontcare on the other hand, has legitimate sources.

anyways, it always comes back to manufacturing... superior manufacturing saved intel during the P4 days, and give them a huge advantage when their designs are actually solid. Intel seems to be head and shoulders above anyone else in the industry when it comes to manufacturing.

 

[jvroig]

Intel seems to be head and shoulders above anyone else in the industry when it comes to manufacturing.

That is clearly true today. Of course, even if they weren't, they wouldn't be in the hot seat that TSMC is, considering that their fabs are their own. They will have no customers running their mouth at a press conference demanding for something impossible such as zero via defects, even when such defects are known and ways around them stated in the guidelines they issue to their fab customers.

Their failures (if it turned out that way) would still be known (no doubt through product delays), but they would not be as consistently derided as TSMC has been since most everything will be kept internally, "on the down low", if I use that phrase correctly.

 

[taltamir]

That is clearly true today. Of course, even if they weren't, they wouldn't be in the hot seat that TSMC is, considering that their fabs are their own. They will have no customers running their mouth at a press conference demanding for something impossible such as zero via defects, even when such defects are known and ways around them stated in the guidelines they issue to their fab customers.

Their failures (if it turned out that way) would still be known (no doubt through product delays), but they would not be as consistently derided as TSMC has been since most everything will be kept internally, "on the down low", if I use that phrase correctly.

its not just today... look back on the past ten years, intel was in the lead in terms of manufacturing tech.

 

[jvroig]

its not just today... look back on the past ten years, intel was in the lead in terms of manufacturing tech.

Yes, of course, I did not mean to imply otherwise. I was fully agreeing with the reality that they are leading the industry. I only meant to say they wouldn't receive the hot-seat treatment TSMC has been getting today even if they weren't the manufacturing tech leader.

 

[Scali]

Makes you wonder if AMD will ever catch up with Intel of process technology. Seems like they been behind by about a year for a while now (65nm, 45nm and now 32nm).

AMD used to be about 18 months behind, and then decided they wanted to decrease the gap to only 12 months.

It seems that ironically the more they try to close the gap, the more problems they run into, and the more they get delayed.
Although technically AMD has been able to get to the production phase on a new process in about 12 months after Intel... the production didn't really reach maturity until about 6 months later. Often when AMD releases products on a new process, initially they aren't even better than the old process in terms of power consumption, clock speed etc.

With Intel on the other hand, as soon as they release products on a new process, it generally results in significant power savings out of the box, and although Intel hasn't really scaled up clockspeed in years, the overclockability of their parts have reached almost legendary status.
Their 32 nm process was another tour-de-force, debuting a 6-core CPU at 3.33 GHz.

 

[Scali]

I remember people saying that back in the friggin K6 days... Yeah. AMD is still around.

Just because AMD is still around doesn't make it any less true.
AMD has been holding on by the skin of their teeth for most of their existence actually.
The reason why they survived the past few years is not because of the success of their own products, but mainly the Arab investment group, and the Intel settlement.
Most of the time, AMD reports quarterly losses. Even in their best period, the Athlon XP/64 days, they barely turned in a decent profit.
AMD is not exactly a healthy company, never was, and by the looks if it, it never will be.
It is still around though. As long as people want to invest in AMD, it will continue to stay around, even if it doesn't make a decent profit.

Read this article from 2003 for example:
http://news.cnet.com/AMDs-elusive-path-to-profits/2010-1071_3-5063556.html

How bizarre is that?

Just like Intel.

Uhh no... Intel may have had less competitive products at times, but Intel has never been anywhere near bankrupcy. In fact, I can't recall the last time Intel turned in a loss in their quarterly results... if they ever operated at a loss at all.

 

[IntelUser2000]

AMD used to be about 18 months behind, and then decided they wanted to decrease the gap to only 12 months.

When were they 18 months behind? As far as I know, the last few years have been consistent 12 months with smaller gap in the Pentium II/III days.

 

[Scali]

When were they 18 months behind? As far as I know, the last few years have been consistent 12 months with smaller gap in the Pentium II/III days.

Heck I dunno.
I just remember that some years ago AMD set this goal, afaik to reduce the gap from 18 to 12 months.
Impossible to google that up now.
Does it matter? No it doesn't.

 

[taltamir]

AMD is not exactly a healthy company, never was, and by the looks if it, it never will be.

Well... they certainly stand a better chance of becoming healthy since they dumped hector ruin.

 

[Scali]

Well... they certainly stand a better chance of becoming healthy since they dumped hector ruin.

I don't think so, to be honest.
Ever since Core2, Intel has basically had AMD in a stranglehold. They let AMD live, but only just... AMD doesn't make any profit, so they cannot invest in R&D, and as such, they cannot develop breakthrough technology for future products.
If Bulldozer turns out to be any good, Intel can just squeeze a bit harder, so AMD still can't keep prices high enough to get a decent profit. If Bulldozer turns out to be another Barcelona... well, Intel can just sit back and do nothing.

It's pretty much a status quo... Intel has the huge marketshare and the superior technology. So they have products that are cheaper to make, yet are sold at higher prices. and larger volumes. These huge profits can be invested in R&D to make sure they maintain their lead... tick-tock, tick-tock...

It doesn't matter who runs the company, AMD doesn't stand a chance until Intel makes another misstep like Pentium 4. AMD didn't use that situation to their advantage last time though...

 

[taltamir]

mmm, well, intel DID create a massive gap... intels current "legendary overclocking" is because they simply massively underclock their latest and greatest in order to squeeze more money out of the market.

that being said, AMD has the advantage of not having to reinvent the wheel, they can simply create their own implementation of intel innovation. EX: turbo boost.

 

[extra]

Uhh no... Intel may have had less competitive products at times, but Intel has never been anywhere near bankrupcy. In fact, I can't recall the last time Intel turned in a loss in their quarterly results... if they ever operated at a loss at all.

You completely misread my post. I wasn't talking anything about Intel's financial status there. I'm talking about their products. Intel and AMD have both had some good products and some not so good products. I predict that we'll see more of the same in the future.

 

[Scali]

I predict that we'll see more of the same in the future.

And I predict that we won't.
Ever since Core2 Duo, the only good thing about AMD's products was their price (pretty much the same story as before the Athlon really). And that comes at the cost of AMD's future as a company, as I've already covered above.

 

[jvroig]

And that comes at the cost of AMD's future as a company, as I've already covered above.

And eventually it may reach a point where industry/market stagnation or "harm to consumers" will be decided by the US Government and break Intel up into pieces, probably focused on taking the manufacturing arm away from the company, in order to revitalize the industry (just like the AT&T Break-up / Bell System Divestiture in 1982).

Due to the tremendous R&D and fab costs required, I don't think that would be a good move myself (not that I'd get a say, I'm not even American), since the industry isn't one where a startup can just realistically start to compete. But having a monopoly also isn't exactly the best, and Intel holding back a little (or a lot) shows it (and yeah, the Dell bribery thing... but that didn't happen, even AMD 'forgot' about it, so we all must have collectively imagined it).

Well, they'll sort it out eventually. The best that could happen for consumers is that AMD and Intel eventually get a more balanced share of the pie (say, 40-60 in favor of Intel, that might be enough). Other than that, AMD just performing poorer and poorer will lead to no good, Intel getting broken up will also lead to no good (by "no good", I mean not foreseeably better than today).

 

[Dark4ng3l]

I think AMD needs to focus on something if they are to survive long term. I mean when you look at the x86 market today vs 10 years ago you see that it has evolved into different segments that require completely different products to fill.

You need high performance and reliability on the enterprise side. Desktop pc market has sort of split into cheap systems and high end performance(both always existed but now its no longer a question of using higher/lower clocked versions of the same chips or just a chip with less cache, etc.) and in the laptop market you have regular notebooks and the ultra low power atom kind of segment.

Really AMD is just spread too thin to cover everything well. They can't seem to break into the enterprise market and their mobile lineup is by far the weakest part of their offerings. Unfortunately those are the two most important segments in terms of revenue going forward.

Ontario seems to be targeting a market somewhere between atom and ULV cpus and that might be a great product for AMD if they can capitalize on the opportunity. Right now the only edge AMD has is massively superior graphics tech and they need to find a way to use this to differentiate their products from intel in a good way.

 

[Scali]

And eventually it may reach a point where industry/market stagnation or "harm to consumers" will be decided by the US Government and break Intel up into pieces, probably focused on taking the manufacturing arm away from the company, in order to revitalize the industry (just like the AT&T Break-up / Bell System Divestiture in 1982).

I doubt it. It's AMD failing to make competing products. In most lines of business, such companies just go out of business, and nothing is ever heard of them. You can't compare it with AT&T, where there was no way for competitors to enter the market, since AT&T owned the entire phone infrastructure.
Intel isn't doing anything illegal... In fact, if Intel would be even more friendly to consumers, it will only kill AMD sooner, which may harm consumers more, because it would remove a budget option from the market.

I hope the US government isn't dumb enough to reward AMD for failing and punish Intel for being successful.

 

[taltamir]

but if the government breaks up intel, you will end up with intel north, south, etc...
you will not end up with a magically revitalized AMD.

it may very well be AMD's final failure and closing down that would prompt the US government to split up intel.

 

[Dark4ng3l]

I doubt it. It's AMD failing to make competing products. In most lines of business, such companies just go out of business, and nothing is ever heard of them. You can't compare it with AT&T, where there was no way for competitors to enter the market, since AT&T owned the entire phone infrastructure.
Intel isn't doing anything illegal... In fact, if Intel would be even more friendly to consumers, it will only kill AMD sooner, which may harm consumers more, because it would remove a budget option from the market.

I hope the US government isn't dumb enough to reward AMD for failing and punish Intel for being successful.

The problem here is that market conditions are basically preventing anyone from competing with Intel. Nobody is going to launch a company that is going to try and compete in the x86 cpu market because the barrier of entry and the level of risk involved are so huge that nobody else will ever be able to compete. For better or worse(at least until process tech stagnates to even up the competition more) its Intel and AMD. Just the investment in building fabs that can produce enough chips to put any kind of market pressure on Intel would be many billions of dollars.

 

[Scali]

I think AMD needs to focus on something if they are to survive long term. I mean when you look at the x86 market today vs 10 years ago you see that it has evolved into different segments that require completely different products to fill.

Exactly.
If a company cannot be successful with its current product line, it has to reinvent itself, and come up with products that ARE successful.
AMD missed quite a few opportunities... they missed the netbook market completely... same with the flash/SSD market (they actually had a flash division, but sold it just before flash got big, great timing guys... AMD's biggest enemy is not Intel, it's AMD's own management).

 

[Scali]

it may very well be AMD's final failure and closing down that would prompt the US government to split up intel.

And a few years later they'll realize what they've done, when they can no longer find decent affordable processors for their servers, supercomputers etc, because all the small Intels aren't capable of delivering what the big Intel once could, they're too busy fighting over budget products.

 

[Scali]

The problem here is that market conditions are basically preventing anyone from competing with Intel. Nobody is going to launch a company that is going to try and compete in the x86 cpu market because the barrier of entry and the level of risk involved are so huge that nobody else will ever be able to compete. For better or worse(at least until process tech stagnates to even up the competition more) its Intel and AMD. Just the investment in building fabs that can produce enough chips to put any kind of market pressure on Intel would be many billions of dollars.

It's a fact of life, but why is that a problem?
I mean, there's Pepsi and Coca Cola aswell, but it's virtually impossible for a third party to take a significant share of the worldwide cola/soft drink market. Again, barrier of entry and level of risk for setting up a worldwide operation on such a large scale makes it impossible.

It's better to come up with something different... Case in point: Red Bull. They entered the market relatively recently, and managed to cut out a good niche in the market.

The same goes for the GPU market by the way... there's AMD and nVidia, but virtually impossible for anyone else to enter the market... But I've never heard anyone making a big deal out of that.

 

[Dark4ng3l]

It's a fact of life, but why is that a problem?
I mean, there's Pepsi and Coca Cola aswell, but it's virtually impossible for a third party to take a significant share of the worldwide cola/soft drink market. Again, barrier of entry and level of risk for setting up a worldwide operation on such a large scale makes it impossible.

It's better to come up with something different... Case in point: Red Bull. They entered the market relatively recently, and managed to cut out a good niche in the market.

The same goes for the GPU market by the way... there's AMD and nVidia, but virtually impossible for anyone else to enter the market... But I've never heard anyone making a big deal out of that.

Oh please. There are plenty of other companies producing cola. Hell you could probably do that out of your own home. Of course your operation would be small but its not comparable at all. You need to spend 10 billion in RND before even starting to think about making a cpu. Then if you make it you might have to sell it for 50$. All this investment needs to be made 5+ years before you can even start to sell your first product.

 

[Scali]

Oh please. There are plenty of other companies producing cola.

There are plenty of companies producing CPUs aswell...
But it's the scale at which they operate, and the marketshare they manage to acquire.
Nobody is anywhere near the scale that Pepsi/Coca Cola or Intel/AMD are.

In other words, it's just not realistic to expect to be able to compete with companies of this magnitude. As AMD continues slipping, this goes for them more and more aswell.

 

[Dark4ng3l]

There are plenty of companies producing CPUs aswell...
But it's the scale at which they operate, and the marketshare they manage to acquire.
Nobody is anywhere near the scale that Pepsi/Coca Cola or Intel/AMD are.

In other words, it's just not realistic to expect to be able to compete with companies of this magnitude. As AMD continues slipping, this goes for them more and more aswell.

Look your comparison does not work at all because cola is super easy to produce. Sure nobody is as big as the big2 but when I go to my supermarket and can buy a couple of cheap brands of cola off the shelf. The difference here is that in the x86 cpu market Via and Transmeta are more or less completely irrelevant. They might exist on paper but even if I want to buy one of their products I have no idea where to go.

You can't compare high tech business where the products you buy have required hundreds of billions of dollars in RND over the last 30 years to make versus something that boils down to mixing 12 ingredients and marketing. The fixed start up costs to make a cpu that is somewhat competitive are many orders of magnitude greater than pretty much any other consumer level product. The only things that might be harder to get into would be military hardware or something.