Hello. I have a PC with two GeForce GTX 1080 installed in PCI-E slots 2 and 4 (both at x16 mode) and want to run a CUDA program using the two GPUs as one, so I can use the 5120 cores (2560x2) at once. I am new to CUDA so after installation, the first step was to run some samples. The simpleMultiGPU sample seems to work fine:
$ ./simpleMultiGPU
Starting simpleMultiGPU
CUDA-capable device count: 2
Generating input data…
Computing with 2 GPUs…
GPU Processing time: 14.015000 (ms)
Computing with Host CPU…
Comparing GPU and Host CPU results…
GPU sum: 16777280.000000
CPU sum: 16777294.395033
Relative difference: 8.580068E-07
But when running the simpleP2P sample this is the result:
$ ./simpleP2P
[./simpleP2P] - Starting…
Checking for multiple GPUs…
CUDA-capable device count: 2
GPU0 = “GeForce GTX 1080” IS capable of Peer-to-Peer (P2P)
GPU1 = “GeForce GTX 1080” IS capable of Peer-to-Peer (P2P)
Checking GPU(s) for support of peer to peer memory access…
Peer access from GeForce GTX 1080 (GPU0) → GeForce GTX 1080 (GPU1) : No
Peer access from GeForce GTX 1080 (GPU1) → GeForce GTX 1080 (GPU0) : No
Two or more GPUs with SM 2.0 or higher capability are required for ./simpleP2P.
Peer to Peer access is not available amongst GPUs in the system, waiving test.
I saw a couple of similar topics in this forum, but their solutions do not apply to my specific case. So my question is: the fact that P2P access is not allowed between the cards represents a problem for the type of use I want?
Thanks.