Hi Chris,
That works fine, but each GPU has to have enough memory to do the processing (i.e., you can’t use it as a way to “stitch together” slow cards with limited memory”).
Cheers,
Erik
> On 19 Apr 2017, at 00:06, Christopher Lilienthal <[log in to unmask]> wrote:
>
> Hello everyone,
>
> Is it possible to run the GPU accelerated portions of RELION on multiple nodes with several GPUs installed in each? I tried once around the end of December and remember there being an issue, but conveniently I forgot what the issue was…
>
> Just to clarify what I am asking; we have a cluster that has 4 nodes each containing 2 GPUs and I was wondering if it was possible to submit a job to utilize all 4 nodes and their 8 GPUs.
>
> Thank you,
>
> Chris
|