Has anybody here experience with using CUDA with the 9400M integrated GPU?
I have a couple questions about it that hopefully someone knows the answers to:
What’s the maximum amount of system memory that can be used as device memory (as a % of the installed total and any hard limit)
Are CUDA device-host memory copies in any way optimised to take advantage of the fact that the device and host memories are common, or do copies still happen over the logical PCIe connection?
Could someone with such a system post, or send me, the results of the pinned bandwithTest, please?
Correct me if I’m wrong, but I thought the 9400M was still a discrete GPU. It might be ‘integrated’ into a laptop in the sense that it’s not an add on card, but it still has separate memory from the host.
At least on the macbooks/macbook pros, I’m pretty sure the graphics memory is actually shared RAM with the host. In fact, a big complaint of many pro owners is that even when using the 9600 (which does have its own memory), the system still blocks off 256MB of your RAM for the 9400. Pretty annoying.
Where can I determine if the 9400M supports SATA PM.
I’ve googled mainly, but am unable to find any info at present.
I’m aware it has 6 ports but don’t know if they support PM (port mulitplier)