This pull request updates the handling of CUDA synchronization and trigger calls in the PDL lowering and warp specialization code, and updates the TVM submodule. The changes simplify and clarify how these synchronization points are detected and marked in the codebase.

Refactoring of CUDA Sync/Trigger Call Detection:

• Updated MarkCudaSyncCalls in lower_pdl.cc to directly check for tl::pdl_trigger() and tl::pdl_sync() instead of matching external call names, improving robustness and maintainability.
• Simplified the logic in WarpSpecializedRoleMarker in warp_specialized_rewriter.cc to directly check for pdl_sync() and pdl_trigger() instead of string-based matching, ensuring more reliable role assignment.

Testing Improvements:

• Enhanced the test_pdl_sync test in test_tilelang_language_pdl.py to assert that "__restrict__" is not present in the generated code, in addition to checking for "cudaGridDependencySynchronize".

Dependency Update:

• Updated the TVM submodule to a newer commit, ensuring compatibility with recent upstream changes.

Summary by CodeRabbit

Release Notes

• Improvements

  • Refined handling of CUDA PDL operations in the compilation pipeline.
  • Updated external dependencies to latest stable versions.

• Tests

  • Enhanced kernel generation test coverage with additional validation.

_{✏️ Tip: You can customize this high-level summary in your review settings.}