AMD CPU speculation... and expert conjecture

[-Fran-]

since there is no GPU thread:

https://d24w6bsrhbeh9d.cloudfront.net/photo/5911191_700b.jpg

Hahhaa, I remember that with the 4200ti a friend gave to me. The card was bathed in liquor and busted the ram. I cleaned it up and worked fine after that.

Fun times.

Cheers!

 

[amdfangirl]

The HD 4890 I'm using right now a friend threw away (at me). I picked it up and saved myself the $229 it would have cost to get a GTX 660 . It had "RGB sand". One application of the fridge trick and it was fine.

 

[fazers_on_stun]

They might get Thunderman as the jury

AMD4LIFE?

Thunderpants couldn't find his trousers with both hands, let alone a courthouse

Besides, unless he sproinged a new alias in the last 2-3 years, he hasn't been on THG with one of his drive-by flamethreads ..

 

[fazers_on_stun]

isnt this how antittrust/ monopoly laws work ?
Was it MS fault that its OS is so popular in europe, and that customers demand a web browser if they pay for a OS ? But MS got hit by antitrust laws.

Heh, I wouldn't base US anti-trust law on what the EC does - they are pretty much in the 'fine the big US corporations' to max out their funding, plus encourage Euro companies to compete by discouraging US companies..

 

[fazers_on_stun]

xbitlabs said 'amd cancelled/delayed steamroller excavator as we know it'. so it's possible that amd might redesign them... may be for tsmc. as for hsa, since both arm and amd are members, amd might even (re)design future apus as arm cpus cores with hsa igpus....

slightly off topic but related to the topic:
something has been bugging me. i read the earlier articles on how nvidia ended up making much more money than amd (in dgpu sector) by designing their flagship asic smaller than amd's - this was attributed as a big factor. does this mean that implementing a smaller-but-similar-or-higher-perf design helped them offset yield issues and cost? then.. what about 32 nm llano, bd and pd cpus vs sandy bridge, did intel make more money partly because of smaller die? iirc sb dies were smaller than bd dies.

As far as AMD CPUs go, to start with - back in 2005 SOI wafers cost around 3X what strained silicon wafers went for, at least at 300mm size: http://www.forbes.com/home/free_forbes/2005/0919/072.html

Of course that is just the starting material, but my understanding is that the processing steps are somewhat different for SOI than strained silicon, not to mention the design. And AMD or GloFlo was unable to find any other customers for SOI, which meant AMD had to bear the fab costs all alone.

2nd, lower yields as a result of a bigger die size means that an already more expensive fab cost gets magnified even further. So Intel has a lower production cost than AMD does, also GF has to make some sort of profit which means another expense to AMD..

 

[mayankleoboy1]

Does that part of Intel called "fabrication and Foundry" has to make a profit when doing business for that part of intel called "CPU business"

 

[-Fran-]

Does that part of Intel called "fabrication and Foundry" has to make a profit when doing business for that part of intel called "CPU business"

It really depends on how Intel accounts each department of the company, but if they use the "internal partners" ideology (one sells a product or service to the other), then yes. Experian uses this model/way, but I don't know the exact name for this way of doing it. Maybe its just "business", hahaha.

Each dept has its own budget and income, etc; but in the big picture, I'd say they don't need to "make a profit", just cover costs when selling internally.

Cheers!

 

[gamerk316]

isnt this how antittrust/ monopoly laws work ?
Was it MS fault that its OS is so popular in europe, and that customers demand a web browser if they pay for a OS ? But MS got hit by antitrust laws.

Only if they INTENTIONALLY abused their market position. Monoplies based on TECHNICAL advantages (EG: Simply being better) are still legal under the Sherman/Clayton Anti-trust acts anyway.

Now, if Intel actually took a loss to undercut AMD, THEN AMD would have a case.

Secondly, Intel makes CPU's. Microsoft makes OS's. If the fact Windows/OSX 98% share benefits the X86 architecture, well, good for X86 as far as the law is concerned. Feel free to drag MS/Apple into a courtroom to explain their love affair with X86, but legally, thats not Intel's problem.

 

[hansmoleman1981]

could be that AMD tried to do too many things with Tahiti : Graphics and compute.

DING DING DING DING DING

BINGO

Compute performance that no one cares about, or asked for in a mainstream consumer GPU.

ROFL.

someone will undoubtedly say "DURR, I am a super crazy developer / animator who does a lot of rendering, and needs compute"

Ok, you have a big ****, no one else cares. You alone are not responsible for driving GPU revenues. Its the kid who wants 90fps on BF3 that drives GPU revenues.

 

[hansmoleman1981]

AMD fanboys and those with a myopic viewpoint are propagating a myth that AMDs survival is necessary to keep Intel/NVidia in check. This myth is at odds with reality.

1. The HPC era is winding down, from a mainstream consumer perspective. Nobody cares about gigabertz, and cores anymore, its not 2005 (coincidentally the year AMD stopped being relevant). Dwindling PC sales, and virtually every market forecast indicates that the desktop PC will be highly irrelevant to the average consumer.

2. The future is mobile. The future is SoC. "IF AMD NOT THERE, WHO WILL STOP INTEL/NVIDIA 111!!". AMD is just the first casualty. NVidia will also go the way of the DoDo. The future is SoC. No one needs to "stop" NVidia. If they don't adapt they will also die.

3. You don't NEED a GTX 670 to game at playable framerates on most games. http://store.steampowered.com/hwsurvey
A small minority of users have the latest and greatest in their machines. Intel HD Graphics 3000 tops out at 4.25%. The only reason GTX 670s/Radeon 7990s sell is because nerds want e-peen, and frames they'll never see....Almost every PC game is optimized to run on cards from two generations ago.

The industry is changing in a big way. The future is SoC. AMD is just the first casualty, and them dying does not mean you are forced to buy NVidia cards for $700....NVidia will also go the way of the Dodo because dGPUs are legacy, and not the future.

4. DX9 is still the standard. Windows 8 only brings DX11.1. Nothing significant.

Basically I'm making the point that you don't need the latest and greatest ***, to play a video game. The bulk of the market spends less than $200 on dGPus. http://www.xbitlabs.com/articles/graphics/display/nvidia-geforce-gtx-460.html Just because AMD doesn't exist, doesn't mean NVidia can magically start charging $$$ for GPUs.....people will just stop buying them if the price is too much to bear.

So...AMD can burn in hell and nobody will notice.

 

[esrever]

^Nope.

There is a reason everyone is getting mobile solutions out.

 

[griptwister]

AMD fanboys and those with a myopic viewpoint are propagating a myth that AMDs survival is necessary to keep Intel/NVidia in check. This myth is at odds with reality.

1. The HPC era is winding down, from a mainstream consumer perspective. Nobody cares about gigabertz, and cores anymore, its not 2005 (coincidentally the year AMD stopped being relevant). Dwindling PC sales, and virtually every market forecast indicates that the desktop PC will be highly irrelevant to the average consumer.

2. The future is mobile. The future is SoC. "IF AMD NOT THERE, WHO WILL STOP INTEL/NVIDIA 111!!". AMD is just the first casualty. NVidia will also go the way of the DoDo. The future is SoC. No one needs to "stop" NVidia. If they don't adapt they will also die.

3. You don't NEED a GTX 670 to game at playable framerates on most games. http://store.steampowered.com/hwsurvey
A small minority of users have the latest and greatest in their machines. Intel HD Graphics 3000 tops out at 4.25%. The only reason GTX 670s/Radeon 7990s sell is because nerds want e-peen, and frames they'll never see....Almost every PC game is optimized to run on cards from two generations ago.

The industry is changing in a big way. The future is SoC. AMD is just the first casualty, and them dying does not mean you are forced to buy NVidia cards for $700....NVidia will also go the way of the Dodo because dGPUs are legacy, and not the future.

4. DX9 is still the standard. Windows 8 only brings DX11.1. Nothing significant.

Basically I'm making the point that you don't need the latest and greatest ***, to play a video game. The bulk of the market spends less than $200 on dGPus. http://www.xbitlabs.com/articles/graphics/display/nvidia-geforce-gtx-460.html Just because AMD doesn't exist, doesn't mean NVidia can magically start charging $$$ for GPUs.....people will just stop buying them if the price is too much to bear.

So...AMD can burn in hell and nobody will notice.

You're forgetting about the Enthusiasts... If FPS and MAXing out games wasn't an issue, then why the Hell is Intel leading the CPU market?

Nvidia should buy out AMD to compete with Intel... Intel is creating GPUs here soon... should be interesting to see if they get in the Gamer's market.

Okay, To max GAMES out, ALL eyecandy to still get a good framerate, you need a decent GPU. Hence, the GTX 670 is a great value.

DX11.1 Has a noticeable Graphic difference compared to DX11. DX11.1 WILL be used in the next Gen consoles. And Windows 7 Will be (or is already) DX11.1 compatable... Look it up.

Games are getting more demanding... they're not reaching their full potential on consoles, how ever, PC's are a different scenario.

 

[noob2222]

Yeah, read that yesterday and pretty decent analysis of what went wrong with AMD. In particular paying twice as much for ATI as they should have, then shortsightedly selling off the handheld graphics unit which now is a big part of ARM-based tablets & smartphones..

Unfortunately for AMD's management, nobody could ever fix stupid..

It cracks me up when people think AMD overpaid by 2.5B on ATI when it was AMD that put the 2.5B loan against ATI assets and not AMD assets. One company or the other was going to drop by 2.5B, and only one did.

5B company with a 2.5B loan has a value of 2.5B. ATI DIDN'T TAKE OUT THE LOAN TO BUY ITSELF, AMD DID.

 

[-Fran-]

DING DING DING DING DING

BINGO

Compute performance that no one cares about, or asked for in a mainstream consumer GPU.

ROFL.

someone will undoubtedly say "DURR, I am a super crazy developer / animator who does a lot of rendering, and needs compute"

Ok, you have a big ****, no one else cares. You alone are not responsible for driving GPU revenues. Its the kid who wants 90fps on BF3 that drives GPU revenues.

You really think AMD or nVidia design their GPUs just for consumers? You do have a point in that regard though; the people that actually want/needs compute on a video card is little to say the least, but it is a market and you have choice. OTOH, rendering farms (or HPC) do use these GPUs and nVidia has been very successful there, that's why AMD had to come up with a solution to compete in that market, since it pays really well. Not volume, but pure profit.

AMD fanboys and those with a myopic viewpoint are propagating a myth that AMDs survival is necessary to keep Intel/NVidia in check. This myth is at odds with reality.

1. The HPC era is winding down, from a mainstream consumer perspective. Nobody cares about gigabertz, and cores anymore, its not 2005 (coincidentally the year AMD stopped being relevant). Dwindling PC sales, and virtually every market forecast indicates that the desktop PC will be highly irrelevant to the average consumer.

2. The future is mobile. The future is SoC. "IF AMD NOT THERE, WHO WILL STOP INTEL/NVIDIA 111!!". AMD is just the first casualty. NVidia will also go the way of the DoDo. The future is SoC. No one needs to "stop" NVidia. If they don't adapt they will also die.

3. You don't NEED a GTX 670 to game at playable framerates on most games. http://store.steampowered.com/hwsurvey
A small minority of users have the latest and greatest in their machines. Intel HD Graphics 3000 tops out at 4.25%. The only reason GTX 670s/Radeon 7990s sell is because nerds want e-peen, and frames they'll never see....Almost every PC game is optimized to run on cards from two generations ago.

The industry is changing in a big way. The future is SoC. AMD is just the first casualty, and them dying does not mean you are forced to buy NVidia cards for $700....NVidia will also go the way of the Dodo because dGPUs are legacy, and not the future.

4. DX9 is still the standard. Windows 8 only brings DX11.1. Nothing significant.

Basically I'm making the point that you don't need the latest and greatest ***, to play a video game. The bulk of the market spends less than $200 on dGPus. http://www.xbitlabs.com/articles/graphics/display/nvidia-geforce-gtx-460.html Just because AMD doesn't exist, doesn't mean NVidia can magically start charging $$$ for GPUs.....people will just stop buying them if the price is too much to bear.

So...AMD can burn in hell and nobody will notice.

AMD is relevant for the market, not because will keep in check prices alone, but because makes nVidia and Intel try harder in each gen. You think Intel would have come up with Conroe without AMD's pressure? We'd be stuck with Pentium 4's design to this day, I'm sure.

This is economic theory, not fanboism.

You're forgetting about the Enthusiasts... If FPS and MAXing out games wasn't an issue, then why the Hell is Intel leading the CPU market?

Nvidia should buy out AMD to compete with Intel... Intel is creating GPUs here soon... should be interesting to see if they get in the Gamer's market.

Okay, To max GAMES out, ALL eyecandy to still get a good framerate, you need a decent GPU. Hence, the GTX 670 is a great value.

DX11.1 Has a noticeable Graphic difference compared to DX11. DX11.1 WILL be used in the next Gen consoles. And Windows 7 Will be (or is already) DX11.1 compatable... Look it up.

Games are getting more demanding... they're not reaching their full potential on consoles, how ever, PC's are a different scenario.

Ugh, "gamers" don't drive markets; or at least, not "PC gamers". Big publishers just target profitable platforms that are wide spread and in control (consoles), so they turn a profit in the short/medium range (quarter by quarter mentality). I do agree that PCs now drive innovation, but I don't really see profit for developers making big innovations in the PC sector; just incremental refinements of proven mechanics. I've talked about this with a friend of mine and I always tell him that Nintendo is the only innovative company left for gamers. They try to appeal new ways of making interaction with games putting their neck when they can. MS and Sony just used proven tech and do "leaking" innovation (not big improvements, but incremental or copies).

Anyway, big and expensive video cards are some-what justifiable because of bad porting or bad coding from consoles. People with regular computers can't max out most games not because of hardware limitations, but problems with some settings across the board. Too much hardware makes it a coding hell for devs. Even more when publishers don't give money to polish games down the road (bug fixing, basically) and just want to turn a quick buck.

Cheers!

 

[fazers_on_stun]

It cracks me up when people think AMD overpaid by 2.5B on ATI when it was AMD that put the 2.5B loan against ATI assets and not AMD assets. One company or the other was going to drop by 2.5B, and only one did.

5B company with a 2.5B loan has a value of 2.5B. ATI DIDN'T TAKE OUT THE LOAN TO BUY ITSELF, AMD DID.

?? When you buy a company like ATI, you buy its assets and outstanding liabilities. If you don't have enough cash to pay for it, you take out a loan against the merged assets (which is what the senior notes are - IIRC they are "senior" or take precedence over other AMD debt such as outstanding stock that they sold to the stockholders). Immaterial whether it was "ATI assets" used to float the loan, as those became AMD assets once the buyout occurred.

If you read the article, or at all knowledgeable about the deal, you should remember that AMD wrote down the book value of ATI (i.e., the so-called "goodwill" tax writeoff) by over half the $5.4BN, which is basically an admission by AMD that the current value of ATI is lower than what AMD paid for it.

Think of it as buying a house for say $500K, using both cash and a loan, and then finding out 2 years later the house is only worth $200K. Your net worth decreased by $300K, and it doesn't matter if you took out a mortgage loan against the house or hocked the family jewels at a pawn shop .

 

[fazers_on_stun]

On a positive note for AMD: http://seekingalpha.com/article/1027391-amd-s-sleeper-agent-temash-for-tablets?source=msn

What Is Temash?

Simply put, "Temash" is the codename for AMD's next generation low power X86-64 APU based on the next-generation low power "Jaguar" core. The following slide from the company's presentation at "Hot Chips" reveals a number of key details about the product:

Basically, this new chip is a very serious improvement over its already successful "Bobcat" APU. It will run at higher clock speeds, do more work per clock, come packed with more cores, have an enhanced instruction set, and sport a wider floating point unit. Part of this is due to a new design, and part of it is due to a shrink from TSMC's (TSM) 40nm process to its 28nm process.

In terms of performance and performance-per-watt, the chip should be a significant improvement over the last gen "Hondo" tablet part, which has seen very few design wins due to the fact that it is power hungry and not a fully integrated SoC. "Temash" changes that by being a full SoC (meaning lower platform power).

Simply put, this chip could be what AMD needs to try to revive itself. With a ground-up, resdesigned low-power X86 core, AMD has a real chance of threatening Intel (INTC) and the various ARM (ARMH) licensees such as Qualcomm (QCOM) at Nvidia (NVDA) at the low end of the market (tablets, netbooks, cheap notebooks). A power-efficient, low-power, cheap-to-make X86 SoC that combines AMD's graphics expertise and X86 license could work out well for the company, especially as the firm restructures itself.

This could be the kind of focused product that AMD needs to swing back to profitability.

 

[-Fran-]

Ah, sounds like good news indeed. Now... The things is "when"?

According to that article, it should be out by Q4-2013, and that means a lot of wait. Intel will have the next gen Atom by then on market. It better be one hell of a product, because at that time, ARM will also have a lot of designs out the door competing. Samsung should be playing with the next Exynos at 28nm as well. Maybe a quad core or something, since Exynos 5 is 32nm and dual A15's.

Cheers!

 

[gamerk316]

Heres the problem with X86 in mobile: How many apps are written for Android/iOS for the ARM architecture? Its going to be REALLY hard for X86 to break in now. X86 in mobile is now basically reliant on Windows 8 for its success.

EDIT

Basically, its the same exact situation with ARM on the desktop: With so many apps written against X86, what are the chances it can make inroads? Not very good.

 

[mayankleoboy1]

^

IIRC, Medfield had a sorta compatibility layer that made using android ARM apps easy. At the time of launch, from the AT review, quite a large percent of the apps he tested worked.

 

[hansmoleman1981]

You really think AMD or nVidia design their GPUs just for consumers? You do have a point in that regard though; the people that actually want/needs compute on a video card is little to say the least, but it is a market and you have choice. OTOH, rendering farms (or HPC) do use these GPUs and nVidia has been very successful there, that's why AMD had to come up with a solution to compete in that market, since it pays really well. Not volume, but pure profit.

you have no idea what you're talking about. Either provide proof of these phantom "rendering farms", and millions and millions of HPC clients that need the all this GPU compute power (The titan super computer doesn't count, provide proof of mainstream GPU compute needs) or shut up. Also, if you do use Titan as an example, you'd be wrong because Titan uses Tesla GPUs, not GTX 670s. Specialized products for specialized markets.

AMD is relevant for the market, not because will keep in check prices alone, but because makes nVidia and Intel try harder in each gen. You think Intel would have come up with Conroe without AMD's pressure? We'd be stuck with Pentium 4's design to this day, I'm sure.

This is economic theory, not fanboism.

--------------------------------------------

This is like saying Fiat keeps Ferrari in check because both of them make cars based on internal combustion engines.

You're ignorant of market forces, and the relevance of AMD to the new market.

Get lost

 

[amdfangirl]

Heres the problem with X86 in mobile: How many apps are written for Android/iOS for the ARM architecture? Its going to be REALLY hard for X86 to break in now. X86 in mobile is now basically reliant on Windows 8 for its success.

EDIT

Basically, its the same exact situation with ARM on the desktop: With so many apps written against X86, what are the chances it can make inroads? Not very good.

There will be people like me who will buy into AMD's Jaguar-based APU on the simple premise that I can buy a $400 ultra-portable that runs Windows 7 for Photoshop, Illustrator and Libreoffice (I hate change, and as a consequence tablets because I'm an old grump )

Here's a guaranteed sale for AMD :3.

All they need to do is to make the Lenovo x120e (Brazos) I had right now about 300g lighter and I will buy a new one next year with this chip in it .

 

[-Fran-]

you have no idea what you're talking about. Either provide proof of these phantom "rendering farms", and millions and millions of HPC clients that need the all this GPU compute power (The titan super computer doesn't count, provide proof of mainstream GPU compute needs) or shut up.

The broader HPC market is nearly $19B.

http://www.hpcwire.com/hpcwire/2011-06-21/idc_shares_hpc_market_figures,_trends,_predictions_at_isc.html

If you really think Intel is getting in there for penny change, you need to think things a little harder.

This is like saying Fiat keeps Ferrari in check because both of them make cars based on internal combustion engines.

You're ignorant of market forces, and the relevance of AMD to the new market.

Get lost

Good thing you brought up that particular example, because is interesting to actually see how FIAT manages Ferrari as a brand. Although, Ferrari doesn't compete with FIAT directly, they do share technology. So, in that, it would be saying that Atom is fighting the I7 market share... Which it actually is; that's why Intel is going to lower power envelopes going forward, haha.

You should have used Lambo and Ferrari, or Porche and Ferrari. Now, AMD is not into extreme or enthusiast markets, so you can say FIAT and Volkswagen or FIAT and Toyota. If you take a look, Toyota gets out something, then everyone notices then starts doing so in a similar fashion. Audi, Toyota, Volvo and BMW are developing really interesting engines tech that make all other vehicles old.

Well, long story short, the "market" is driven by the variety of the companies populating the different needs. If you take out companies so that 1 or 2 remain, it only deteriorates innovation and competition as a whole. Unless, you want to try a Communist POV in that regard, I'd say it doesn't work with few driving the industry.

Also, why so raging? Did anything bad happen to you today?

Cheers!

 

[mayankleoboy1]

Also, why so raging? Did anything bad happen to you today?

Cheers!

+1

@ hansmoleman1981 : Calm down, big lad.

 

[hansmoleman1981]

The broader HPC market is nearly $19B.

http://www.hpcwire.com/hpcwire/2011-06-21/idc_shares_hpc_market_figures,_trends,_predictions_at_isc.html

If you really think Intel is getting in there for penny change, you need to think things a little harder

GTX 670 is not used in HPC. Go learn about Quadro/Tesla and why one of those cost 100x a GTX 680.

Try a little harder please. Don't give me useless links. Specialized products for specialized markets. Show me Tahiti 7970 being used in HPC/GPGPU compute. No one would buy a FirePro if 7970 was so awesome in GPGPU.

Prove that any significant % of 7970 buyers, buy it for its compute needs. Yea you can't, you're just blowing hot air.

You missed the point entirely. AMD is not relevant to Intel. Intel doesn't care what AMD does. AMD is to computing, what Saab was to automotive.

If you take out companies so that 1 or 2 remain, it only deteriorates innovation and competition as a whole.

Intel owns 81% of x86. What competition?

 

[-Fran-]

The broader HPC market is nearly $19B.

http://www.hpcwire.com/hpcwire/2011-06-21/idc_shares_hpc_market_figures,_trends,_predictions_at_isc.html

If you really think Intel is getting in there for penny change, you need to think things a little harder

GTX 670 is not used in HPC. Go learn about Quadro/Tesla and why one of those cost 100x a GTX 680.

Try a little harder please. Don't give me useless links. Specialized products for specialized markets. Show me Tahiti 7970 being used in HPC/GPGPU compute. No one would buy a FirePro if 7970 was so awesome in GPGPU.

Prove that any significant % of 7970 buyers, buy it for its compute needs. Yea you can't, you're just blowing hot air.



You missed the point entirely. AMD is not relevant to Intel. Intel doesn't care what AMD does. AMD is to computing, what Saab was to automotive.

Uhm... I actually agreed with you regarding about the market of regular video cards and heavy computing. K104 was not meant for GPGPU, nVidia already said that a long time ago; they had big Kepler for that (Quadros based on K100). That's why they slapped 256bit GDDR5 in there for the little Kepler.

Also, I think you got confused. I told you AMD didn't design GCN with regular people in mind, they needed something to counter nVidia in the HPC market. That's why they went and got the GCN arch; too bad they have to use that for both consumer and enterprise level cards, but nVidia did the same with Fermi. And I don't know why HPC folks are not buying (AFAIK) GCN based FirePros, but they do compete with nVidia now; you can go check the benchmarks from Anand and Toms for the new FirePros and Quadros. Maybe it's a porting from CUDA issue or inmmature support for OpenCL or something, who knows really; I don't attend to HPC conferences that touch the GPGPU part (or HPC conferences at all, for that matter, hahaha).

Regarding AMD not being relevant to Intel, yes I agree to some extent. That doesn't mean however that AMD going poof is not going to affect us. You could name me another company that will force Intel to innovate if AMD disappears. Do you think Intel will license x86 to another company out there? I really don't think VIA will step up to the game either, haha. Point is, AMD disappearing will affect how the market evolves and moves. If AMD declares chapter 11, I'm sure Intel shares will skyrocket; maybe nVidias too.

And it's funny you mention SAAB, because they actually had the best of the best security for cars (at least until the 90's IIRC). Too bad they were over-engineered for the time. Go see Top Gears take on SAABs history. Quite informative.

Cheers!