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
With a query similar to this one we can get the properties of the instance types to make the decision together with the docker image chosen by the user on the system.env file
aws ec2 describe-instance-types --query 'InstanceTypes[?GpuInfo.Gpus!=null].[InstanceType, VCpuInfo.DefaultVCpus, MemoryInfo.SizeInMiB, GpuInfo.Gpus[0].Count, GpuInfo.Gpus[0].Manufacturer, GpuInfo.Gpus[0].MemoryInfo.SizeInMiB, GpuInfo.Gpus[0].Name]' --output text --region us-east-1 | grep NVIDIA | sort
Using VCPUs and other system specs, we should be able to calculate how many workers a given instance can use on a given track
The text was updated successfully, but these errors were encountered: