You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When using Batch, we would like to be able to run some jobs on the CPU and some on GPU. To do this, we will need to modify the RV profile configuration to specify CPU and GPU resources. Currently a config looks like:
Then the aws_batch runner needs to be modified so that it picks the correct job queue and job def for the command. Each command needs to have a mechanism for deciding which queue should be used (analyze, chip, and eval seem like good candidates for CPU). When jobs are submitted, the required number of CPU cores needs to be specified. It should stay at the current value for GPUs, but CPU jobs should only require one core.
When using Batch, we would like to be able to run some jobs on the CPU and some on GPU. To do this, we will need to modify the RV profile configuration to specify CPU and GPU resources. Currently a config looks like:
but it should become something like
Then the
aws_batch
runner needs to be modified so that it picks the correct job queue and job def for the command. Each command needs to have a mechanism for deciding which queue should be used (analyze, chip, and eval seem like good candidates for CPU). When jobs are submitted, the required number of CPU cores needs to be specified. It should stay at the current value for GPUs, but CPU jobs should only require one core.A prerequisite for this issue is #634
The text was updated successfully, but these errors were encountered: