Closed mathomp4 closed 9 months ago
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged.
I'm going to block this until I can actually test on Milan. Probably good not to break things. 😄
Well, testing with the Bill Branch shows it crashes with any optimization it seems. znver2, znver3, native...
I'll try stock GEOSgcm...
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged.
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged.
This PR is being prevented from merging because you have added one of our blocking labels: Contingent - DNA, Needs Lead Approval, Contingent -- Do Not Approve. You'll need to remove it before this PR can be merged.
This PR adds support for AMD Milan. Testing at NAS shows that:
matches the Romes and that:
matches the Milan processors on AWS.
We mainly do this to use
-znver3
on AWS.UPDATE: Turns out tests on the Milans at NCCS show there is no difference between
-znver2
and-znver3
. Same speed, zero-diff. Thus we do not add complication to ESMA_cmake and close this PR.