You are not logged in.

#1 2014-12-10 18:27:10

labotsirc
Member
Registered: 2013-08-21
Posts: 108

Does CUDA multi-GPU peer memory access work in Arch ?

Hello,

Recently I've been facing problems with CUDA multi-GPU specific functionality. It is the P2P memory access between one GPU and another, of course both installed in the same workstation.
For the people who have 2 or more GPUs and Arch linux, can you test the CUDA example called "simpleP2P", because mine becomes freezed
at some point in the execution.

I posted this in more in detail at this link: http://stackoverflow.com/questions/2740 … cation-why
And they have replied me that Arch linux is not officially supported for CUDA. If at least one person can make the "simpleP2P" example work
in Arch Linux, then que can know that the distro is not the problem. But to know this I really need the help from all of you who have 2 or more GPUs.

Thanks in advance

Last edited by labotsirc (2014-12-10 18:42:58)

Offline

Board footer

Powered by FluxBB