Both Intel hereand AMD here are putting the GPU on the CPU chip,I have some concerns about this, as with Intel’s first attempt, even after this amount of time, we see no genuinely decent drivers, open or closed source. But I have read some articles that suggest things are on the improve, AMD having an open source driver before the chip is on the market for example is nice. Anyone have some thoughts or information on this that they don,t mind sharing?
It is good news for me and for all amd owners. I am waiting for buldoser.
My understanding is this typically only refers to the decoding of videos (being passed from the CPU to the GPU) and NOT to encoding of videos. Further IMHO this is most useful for PCs that do not have powerful CPU’s. For example a PC with a Core i7 CPU typically does not need to offload the decoding from CPU to GPU. But PCs with slower CPUs can benefit from such.
Presumeably the CPU’s that Intel and AMD are offering with this such a capability are not up to Core i7 speed levels.
I assume in all Intel and AMD cases where a GPU on the same motherboard as the CPU, that the GPU can be disabled in BIOS, if one wishes instead to use a separate PCI-e graphic card’s GPU.
The state of Linux drivers would always be a concern of mine. Thus far nVidia for over a year have been successful in providing proprietary Linux graphic driver that allows the decode of selected video formats to be offloaded from the CPU to the GPU.
ATI, on the other hand, have mostly not been successful in such Linux driver implementation. There have been some recent help request threads by users on our forum asking for assistance to setup CPU to GPU offload on ATI hardware and no one has pitched in to help. Thus for ATI at least, it appears this does not function with openSUSE (I have read stories of Ubuntu users succeeding with 3rd party packaged apps).
When considering Intel graphics, I have read of users with PCs that have relatively slow (compared to Core i7) Intel CPUs that can play HD videos with no problem, suggesting that the latest Intel Graphic drivers for Linux (such as 2.12.0 or newer) DO support offloading of the deoding from CPU to GPU. The latest 2.6.38 kernel with the latest 2.14.0 Intel graphic driver, is puported to be a big improvement for the Sandybridge architecture, and hence it appears that Intel have surpassed ATI in providing this Linux driver CPU-to-GPU HD decoding functionality, and indeed Intel may be passing nVidia soon in this same decoding functionally.
The handy thing about the Intel Linux drivers is one typically does not need to download and install a proprietary Intel driver to obtain the CPU to GPU offload functionality, while one MUST for both ATI and nVidia.
As a person who does lots of encoding of videos (in addition to decoding) I am looking forward to the day when the CPU to GPU offloading (for Linux) will also include encoding and improve the encoding speed.
Oldcpu is right about ati drivers. Ati drivers must be improved for linux users. About nvidia drivers I do not know because in Hellas subforum we have problems with nvidia drivers. These processors will usefull from laptops. Laptops have not as good graphics card as desktops.
It is good news for me and for all amd owners. I am waiting for buldoser.
I admit to being a bit of an AMD fan, I feel they play a large part in keeping computers affordable, just by offering an alternative to intel.
Integrated graphics chips are important for laptops/notebooks/netbooks and other mobile devices. As well as taking up less space, they typically run cooler than discrete/dedicated graphis devices, and cost less.
About cost I do not know. In Hellas anything technological new produce itself, it is always expencive.
ATI, on the other hand, have mostly not been successful in such Linux driver implementation. There have been some recent help request threads by users on our forum asking for assistance to setup CPU to GPU offload on ATI hardware and no one has pitched in to help. Thus for ATI at least, it appears this does not function with openSUSE (I have read stories of Ubuntu users succeeding with 3rd party packaged apps).
If I understand correctly the AMD/ATI reasoning for not supplying open source drivers that use the GPU to do video encoding/decoding has to do with licensing, that does not excuse the lack of a closed source driver that does though.
When considering Intel graphics, I have read of users with PCs that have relatively slow (compared to Core i7) Intel CPUs that can play HD videos with no problem, suggesting that the latest Intel Graphic drivers for Linux (such as 2.12.0 or newer) DO support offloading of the deoding from CPU to GPU. The latest 2.6.38 kernel with the latest 2.14.0 Intel graphic driver, is puported to be a big improvement for the Sandybridge architecture, and hence it appears that Intel have surpassed ATI in providing this Linux driver CPU-to-GPU HD decoding functionality, and indeed Intel may be passing nVidia soon in this same decoding functionally.
I am typing this from a laptop with an intel graphics chip, and the linux driver is every bit as good as the windows driver, but intel’s support seems patchy. I can’t just say intel’ s graphics chips are a good choice for linux.
I notice nvidia are producing some CPU/GPU combo’s for lower powered applications, It would be nice to see that go main stream, competition has to be a good thing.
Integrated graphics chips are important for laptops/notebooks/netbooks and other mobile devices. As well as taking up less space, they typically run cooler than discrete/dedicated graphis devices, and cost less.
I agree, my concern is how well they support linux as they make all the changes to the CPU/GPU package, and the supporting hardware.
In reality, it doesn’t quite work like that. 
Hardware changes for several reasons like new technology, enhancement, competitition, etc. The question or concern is then “how well does/will linux (or any other OS) support that new/changed hardware?”, assuming that the h/w technology works, is taken up by the marketplace, and its interface data is made available to OS developers.
So am I, the linux driver (2.12.0) is good, and I have no reason to dispute your claim in comparison to W7’s driver. Intel’s support for my chipset (> one year old) doesn’t seem to be patchy. The main bug (full-screen Flash freeze) was cured by a linux kernel update or a simple kernel upgrade. I could claim that’s patchy linux support, but it would be a claim without knowing all the underlying facts. Even with the much newer Arrandale chipset, the latest user results on our forum suggest the intel driver isn’t the main issue, but a significant kernel upgrade is required to make it work. Thus far I would say that intel’s graphics chipsets are a good choice for linux, providing you don’t expect instant support for the latest chipsets.
I notice nvidia are producing some CPU/GPU combo’s for lower powered applications, It would be nice to see that go main stream, competition has to be a good thing.
I agree that competition is usually “a good thing”, but too much could stretch linux resources and delay delivery of kernel/xorg/driver support for some graphic chipsets.
The question or concern is then “how well does/will linux (or any other OS) support that new/changed hardware?”,
Point taken, but still, some go to the trouble of offering decent drivers, ( as opposed to relying on the open source community to write them ) these are the companies I prefer to deal with.
So am I, the linux driver (2.12.0) is good, and I have no reason to dispute your claim in comparison to W7’s driver. Intel’s support for my chipset (> one year old) doesn’t seem to be patchy. The main bug (full-screen Flash freeze) was cured by a linux kernel update or a simple kernel upgrade. I could claim that’s patchy linux support, but it would be a claim without knowing all the underlying facts. Even with the much newer Arrandale chipset, the latest user results on our forum suggest the intel driver isn’t the main issue, but a significant kernel upgrade is required to make it work. Thus far I would say that intel’s graphics chipsets are a good choice for linux, providing you don’t expect instant support for the latest chipsets.
I would like to see them getting things working somewhat quicker though. (It’s on my wish list!)
I agree that competition is usually “a good thing”, but too much could stretch linux resources and delay delivery of kernel/xorg/driver support for some graphic chipsets.
That’s a good point, but again, if they put a lot of the work in to the driver?
I do understand that the majority of sales go to windows users, but hey, we can still hope!
Indeed, … but an annoyance for Linux users of the proprietary ATI graphic driver is even the proprietary driver offloading of video decoding from the CPU to GPU is not implemented in any sort of user friendly manner. As I noted, I have read of a VERY small number of Ubuntu users getting this to work (with custom 3rd party packaged apps installed) but to date I have not read of any openSUSE users who claim success. I have seen help requests on our forum to which no one has really replied much (other than my feeble efforts - which went unanswered indicating my suggestions were likely no good).
Nvidia with the VDPAU (Linux equivalent of nVidia’s MS-Windows Pure Video) implementation works reasonably well with mplayer/smplayer combination. I’ve read of openSUSE mythTV users getting this to work with nVidia graphic cards. Both xine and vlc are purportedly working on trying to implement access to VDPAU. With VDPAU (w/proprietary nVidia graphic driver) on an OLD 32-bit athlon-2800 and a 512MB nVidia GeForce 8400GS PCI (not PCI-e) card, I can play HD videos that a much newer and more powerful 64-bit P8400 Dual Core w/ATI Radeon HD3450 (and ATI proprietary catalyst driver) can not play smoothly. IMHO thats a good indication as to how offloading of the video decoding from CPU to GPU can really help slower CPUs. Of course not all nVidia cards support VDPAU (although many do) so one must choose their card carefully.