xla
xla copied to clipboard
PR #59936: [NVIDIA XLA] Disable TF32 evaluation for SelfAdjointEigTest cases.
PR #59936: [NVIDIA XLA] Disable TF32 evaluation for SelfAdjointEigTest cases.
Imported from GitHub PR https://github.com/tensorflow/tensorflow/pull/59936
When run on Ampere and newer GPUs some of the self adjoint eigenvalue test cases may fall outside of the float32-calibrated tolerances. This PR disabled TF32 for these tests. Copybara import of the project:
-- d2f4bee27b53b2071abd0a5fae3a1096e7f01134 by Nathan Luehr [email protected]:
Disable TF32 evaluation for SelfAdjointEigTest cases.
Merging this change closes #59936
FUTURE_COPYBARA_INTEGRATE_REVIEW=https://github.com/tensorflow/tensorflow/pull/59936 from nluehr:adjoint-eig-tf32-fix d2f4bee27b53b2071abd0a5fae3a1096e7f01134