tqchen opened a new pull request, #19955:
URL: https://github.com/apache/tvm/pull/19955

   ## Summary
   
   - split regular CPU and GPU pytest selection with explicit `gpu` marker 
expressions while preserving native pytest exit behavior
   - phase out four obsolete testing modules and narrowly xfail only the known 
DISCO compile-time uncomputed-expression failure
   - repair seven independently audited CPU failure categories with localized 
TE, TIR transform, codegen, tensor-intrinsic, and schedule regressions
   - preserve the exact seven-category commit history so each root cause 
remains independently reviewable
   
   The seven repaired categories cover packed raw strings (7 failures), 
CUDA/Metal matrix access-pointer types (7), internal TE extern offsets (1), a 
typed DLTensor fixture expectation (1), MetaSchedule scalar annotations (1), 
opaque workspace ABI casts (12), and storage-scope-independent 
auto-tensorization matching (19).
   
   ## Validation
   
   - all 35 locally available original failing nodeids pass together
   - the 13 dependency-gated CBLAS/TFLite cases have focused compiling TE and 
generated-C regressions
   - all mapping and tensor-core sketch cases pass without changing expected 
scripts
   - focused marker selection proves 32 non-GPU and 8 GPU selections
   - compiler build, branch-wide pre-commit, and Git whitespace checks pass


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to