Open ilumsden opened 1 year ago
As recommended by @hariharan-devarajan, we should add timeouts to the UCX communication routines to help us detect failures.
As a starting point for this, we can look to see in UCX supports timeouts in the tag-matching API. If not, we can add manual timeouts.
As recommended by @hariharan-devarajan, we should add timeouts to the UCX communication routines to help us detect failures.
As a starting point for this, we can look to see in UCX supports timeouts in the tag-matching API. If not, we can add manual timeouts.