Gaming has been the only reason for discrete graphics cards for many years already.
It's up to AMD and Nvidia to create new markets which Nvidia is doing with Tegra and CUDA. AMD doesn't have the same high profile non-gaming GPU activity going on, and here is the result. Seems this 10% cut is hitting the GPU side hardest.
Intel would need to ramp up their driver department quite a bit. And AMD has quite a nice offering concerning integrated graphics already on the market. Also, GPUs can be used for heavy parallel computation as well. Is Intel considering Quadro/FireGL style graphics solutions? I think both AMD and nVidia will be fine for many years to come.
Best wishes to those who are departing AMD. It's never easy being booted. In life when one door closes, often another opportunity presents itself.
Let's hope Read knows what he is doing and can get AMD sorted out. He can't do it alone so there will be new people coming into AMD as well. AMD has some good products now and more in the pipeline. They need to execute better at GF and TSMC for AMD to move forward in big strides.
Its a shame they sold all their mobile technology a few years back, what they could really do with right now is a good ARM SoC. I think they should still be able to make one though, ARM will licence out a core to anyone who pays, and AMD has both CPU and graphics building expertise. Few sectors are growing as rapidly as ARM SoC's. Its clear that continuing to compete with Intel directly will be difficult for them and they will have to fight tooth and nail for every bit of profits, I think they could build a formidiable ARM SoC though, if only they started three years ago. That's one of the reasons their last CEO got the boot, right? I hope they still do it though, better late then never, and Windows 8 on ARM should be pretty big.
Their technology is successful at Qualcomm and Broadcom at least. That is actually better then something failing in house. That's more then you can say about nVidias tech that has been marginalized by being tied to their own products now rather then an addition/companion to another SoC/platform. Their survival in that field is dependent on that it makes it in the tablet market, but it got tough competition from TI, Samsung, Qualcomm, Freescale and other now. GF already does ARM SoC's for mayor vendors and semi-companies there really are no reason for AMD to cook a standard ARM IP macro or soft SoC. GF already has in-house developed ARM IP tech of their own. But they got other clients. Competition would be far grater in this field though also. Windows 8 tablets should pretty much exclusively be x86 I guess.
but the last AMD cpu i bought was Athlon64, and it now seems that my last GPU will be the Radeon5770 that is still in my system.
i guess that the great talents of both companies did a good job together, but now it seems that BOTH camps are empty. AMD has no gpu and no cpu talents from old times.
I think we can safely say that there will be continued strive for smaller GPU-architecture design thanks to their current plans and products with integrated gpus and APUs where they just can't spend a billion transistors on the graphics. They will have to re-think again to pull it off in our mobile world where mobile gpus for laptops and Fusion APUs will be their biggest graphics silicon market. They sold off the tech ATi had acquired in the field though, but a fresh start might be good. However the mobile and embedded players will get a larger role in this I would guess, their gpus and drivers have to be pretty good and suitable for all types of multimedia. I guess there will be some convergence there at least all platforms will support pretty much the same features though AMD probably won't go for the handset market but I can see them delivering low-power chips for Tablet PCs or some such. Intel hasn't really come up with a GPU for their embedded market either so. But they will have to do more with less (transistors) definitively. CPUs and custom logic gets faster, to be a compute engine they will have to do more with what they got or they will simply be framebuffers plus lots of drivercode and not the high-tech industry it can be.
I think high-end gpus will be pretty much out even when it comes to games in the future though. Games have been pretty stuck as well as the tech under it all it's simply a matter of targeting now, even integrated graphics is pretty competent now. Mid-end should have a great future in the notebook computer space for some time to come though. You still need an architecture that scales though. But just not so badly downwards smaller versions/platforms.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
19 Comments
Back to Article
silverblue - Friday, November 4, 2011 - link
I wonder if JF is still working for AMD?cfaalm - Friday, November 4, 2011 - link
Not good news indeed. I hope they can find their strength and a new opportunity to show it.JF? Just check the blogs. If he blogs again after today, he's still there ;-)
mmatis - Friday, November 4, 2011 - link
"<i>Hopefully we'll see some of these folks appear at Apple, Intel, NVIDIA and Qualcomm.</i>"Or hopefully they'll start their own business and come back to body shop at AMD and the others you mentioned...
jeremyshaw - Friday, November 4, 2011 - link
well, the nVidia CEO used to work at AMD, :pvision33r - Friday, November 4, 2011 - link
As Intel improves it's GPU performance, it's slowly killing Nvidia and ATI's dedicated graphics solutions.In a few years, other than high end games. Most people will stick to integrated graphics.
Taft12 - Friday, November 4, 2011 - link
Gaming has been the only reason for discrete graphics cards for many years already.It's up to AMD and Nvidia to create new markets which Nvidia is doing with Tegra and CUDA. AMD doesn't have the same high profile non-gaming GPU activity going on, and here is the result. Seems this 10% cut is hitting the GPU side hardest.
magnetik - Monday, November 7, 2011 - link
You're quite uninformed about AMD GPU compute performance.The AMD OpenCL implementation provides much better fixed-point math performance than Nvidia.AMD GPUs perform 4x-6x faster per Watt than Nvidia GPUs when MD5 hashing, as is used for Bitcoin mining.
If I were an enterprise customer looking to do some serious fixed-point math, I would buy AMD GPUs. The Nvidia "Tegra" brand is just a marketing ploy.
Th-z - Tuesday, November 8, 2011 - link
One of reasons. Graphic and design folks also buy discrete card. GPGPU is the new market that people start to buy discrete card.Yes AMD would need better product in GPGPU market. Let's hope their Graphic Core Next can better compete with NVIDIA's CUDA architecture.
Not sure if the layoff impacts their GPGPU plan going forward.
Death666Angel - Friday, November 4, 2011 - link
Intel would need to ramp up their driver department quite a bit. And AMD has quite a nice offering concerning integrated graphics already on the market. Also, GPUs can be used for heavy parallel computation as well. Is Intel considering Quadro/FireGL style graphics solutions?I think both AMD and nVidia will be fine for many years to come.
Arnulf - Friday, November 4, 2011 - link
"The relevant exerpt from ..."Should really be excerpt.
Beenthere - Friday, November 4, 2011 - link
Best wishes to those who are departing AMD. It's never easy being booted. In life when one door closes, often another opportunity presents itself.Let's hope Read knows what he is doing and can get AMD sorted out. He can't do it alone so there will be new people coming into AMD as well. AMD has some good products now and more in the pipeline. They need to execute better at GF and TSMC for AMD to move forward in big strides.
FATCamaro - Friday, November 4, 2011 - link
I can't believe you would go up on stage with jeans that show your unit....ezekiel68 - Friday, November 4, 2011 - link
Funny, I hadn't noticed. I guess it all depends on what you're looking for.tipoo - Friday, November 4, 2011 - link
Its a shame they sold all their mobile technology a few years back, what they could really do with right now is a good ARM SoC. I think they should still be able to make one though, ARM will licence out a core to anyone who pays, and AMD has both CPU and graphics building expertise. Few sectors are growing as rapidly as ARM SoC's. Its clear that continuing to compete with Intel directly will be difficult for them and they will have to fight tooth and nail for every bit of profits, I think they could build a formidiable ARM SoC though, if only they started three years ago. That's one of the reasons their last CEO got the boot, right? I hope they still do it though, better late then never, and Windows 8 on ARM should be pretty big.Penti - Saturday, November 5, 2011 - link
Their technology is successful at Qualcomm and Broadcom at least. That is actually better then something failing in house. That's more then you can say about nVidias tech that has been marginalized by being tied to their own products now rather then an addition/companion to another SoC/platform. Their survival in that field is dependent on that it makes it in the tablet market, but it got tough competition from TI, Samsung, Qualcomm, Freescale and other now. GF already does ARM SoC's for mayor vendors and semi-companies there really are no reason for AMD to cook a standard ARM IP macro or soft SoC. GF already has in-house developed ARM IP tech of their own. But they got other clients. Competition would be far grater in this field though also. Windows 8 tablets should pretty much exclusively be x86 I guess.marc1000 - Friday, November 4, 2011 - link
but the last AMD cpu i bought was Athlon64, and it now seems that my last GPU will be the Radeon5770 that is still in my system.i guess that the great talents of both companies did a good job together, but now it seems that BOTH camps are empty. AMD has no gpu and no cpu talents from old times.
mxnerd - Saturday, November 5, 2011 - link
AMD really sholud have licensed ARM three years ago. But with 10% workforce and lead engineer gone, really wonder AMD's future.Penti - Saturday, November 5, 2011 - link
I think we can safely say that there will be continued strive for smaller GPU-architecture design thanks to their current plans and products with integrated gpus and APUs where they just can't spend a billion transistors on the graphics. They will have to re-think again to pull it off in our mobile world where mobile gpus for laptops and Fusion APUs will be their biggest graphics silicon market. They sold off the tech ATi had acquired in the field though, but a fresh start might be good. However the mobile and embedded players will get a larger role in this I would guess, their gpus and drivers have to be pretty good and suitable for all types of multimedia. I guess there will be some convergence there at least all platforms will support pretty much the same features though AMD probably won't go for the handset market but I can see them delivering low-power chips for Tablet PCs or some such. Intel hasn't really come up with a GPU for their embedded market either so. But they will have to do more with less (transistors) definitively. CPUs and custom logic gets faster, to be a compute engine they will have to do more with what they got or they will simply be framebuffers plus lots of drivercode and not the high-tech industry it can be.I think high-end gpus will be pretty much out even when it comes to games in the future though. Games have been pretty stuck as well as the tech under it all it's simply a matter of targeting now, even integrated graphics is pretty competent now. Mid-end should have a great future in the notebook computer space for some time to come though. You still need an architecture that scales though. But just not so badly downwards smaller versions/platforms.
fdfsxcvdcfdh - Monday, November 7, 2011 - link
ONLINE STOR good shopping