Open quantumsteve opened 1 year ago
Found while troubleshooting #624.
Did a few tests, I think that 32GB will be enough, but it is still too much for CI.
For some reason, when we create a 6D problem, we are unreasonable amount of memory for something. I wonder if this is an issue with the hash-map taking too much space. If I'm right, this will be an issue across the board, it's just that the problem needs to run 4 copies (since the test uses 4 mpi ranks) and all of our workstations have lots more than that.
Describe the bug A clear and concise description of what the bug is.
To Reproduce Steps to reproduce the behavior:
asgard-unit-mpi-gxx
andasgard-unit-mpi-gxx-scalapack
currently run out of memory on the CI machines. Iincreased the memory size from 15GB to 144GB as a (temporarily?) workaround, suspect there is something over allocating memory or a test that should be scaled down or moved to a different label that runs less often.Expected behavior A clear and concise description of what you expected to happen.
MPI tests pass on a container with 15GB RAM.
System:
module list
]Additional context Add any other context about the problem here.
Reproduced locally with
docker run -m 15000m -it cpu /bin/bash