[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [HTCondor-users] Running multiple jobs on the same GPU



On 4/25/2018 9:02 AM, Michael Pelletier wrote:
NVIDIAâs CUDA 9.1 package introduces a new service that partitions GPUs in the driver, so I think weâre starting to get to the point where weâll need to see GPUs as partitionable resources. Iâve been meaning to experiment with that feature to see how one would go about advertising it to the collector.


Indeed! Michael, is this service you refer to above the Volta Mutli-Process Service or something else?

Before nvidia Volta, imho MPS wasn't very interesting / safe to do, but the new MPS hardware assistance in Volta makes it worth considering doing the work to have HTCondor interface with MPS (or whatever the newer service is) to potentially run multiple jobs per device.

regards,
Todd