So for me as well, I suspect, as the vast majority of gamers the Sandy Bridge integrated graphics processor would be best served as a co-processor to take heavy computational work away from the general purpose cores such as physics, image processing etc a la IBM's Cell chip. Intel will be releasing OpenCL for Sandy Bridge's IGP and obviously DirectX compute shaders will also work.
However for some absoultely barmy reason Intel has disabled the IGP when a discrete GPU has been plugged in to the host motherboard and so stopped any such co-processor work going on! Ahhhhh!!!!
So as far as I know Intel is planning to change this in the future but I don't know whether they mean with Ivy Bridge (or beyond), a new motherboard chipset or simply by a motherboard firmware update. Has anybody any idea (technical or otherwise) why Intel has decided to do this and how Intel or someone else might enable the IGP whilst a discreet GPU is plugged in through one of the above methods or otherwise?
One ray of hope I can see is that the motherboard must tell Sandy Bridge that a discreet GPU has been plugged in so it doesn't seem to be a far stretch that a motherboard manufacturer provides a way through say the BIOS (or UEFI) to simply switch off this signal and thus have the IGP still active. Also the IGP is connected directly to the L3 cache via the ring bus all of which must still be active when the discreet GPU is in so there doesn't seem to be any special bus that can only be used by one GPU at a time which is also promising.
Anybody any information/thoughts about this?
However for some absoultely barmy reason Intel has disabled the IGP when a discrete GPU has been plugged in to the host motherboard and so stopped any such co-processor work going on! Ahhhhh!!!!
So as far as I know Intel is planning to change this in the future but I don't know whether they mean with Ivy Bridge (or beyond), a new motherboard chipset or simply by a motherboard firmware update. Has anybody any idea (technical or otherwise) why Intel has decided to do this and how Intel or someone else might enable the IGP whilst a discreet GPU is plugged in through one of the above methods or otherwise?
One ray of hope I can see is that the motherboard must tell Sandy Bridge that a discreet GPU has been plugged in so it doesn't seem to be a far stretch that a motherboard manufacturer provides a way through say the BIOS (or UEFI) to simply switch off this signal and thus have the IGP still active. Also the IGP is connected directly to the L3 cache via the ring bus all of which must still be active when the discreet GPU is in so there doesn't seem to be any special bus that can only be used by one GPU at a time which is also promising.
Anybody any information/thoughts about this?
Facebook
Twitter