Try running Mooncake linalg tests with CuTensorMaps#461
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. 🚀 New features to boost your workflow:
|
lkdvos
left a comment
There was a problem hiding this comment.
For runtime purposes, do you think it makes sense to instead of calling the full test_rule, we could just compare the values with the CPU versions? Or do we care a bit more about the interface tests as well (and can we check only those?)
|
I think it can make sense to check that the memory is unchanged considering sometimes stuff allocs a new array on CUDA that doesn't on the CPU. Let me see if that is handled in the interface or correctness part of the Mooncake test utils. |
|
Looks like the memory address checking happens in the interface tests for reverse rules. Let's then restrict to interface-only on CI, we can keep the full suite when running locally. |
|
Your PR no longer requires formatting changes. Thank you for your contribution! |
No description provided.