tqchen opened a new pull request, #20249: URL: https://github.com/apache/tvm/pull/20249
Lift primitive expression nodes into a shared `ir/prim` layer so arithmetic and other core consumers can use them without treating expression structure as TIRX-owned. Move reduction construction into TE as opaque pre-lowering state, retain TE validation of nested reductions, and keep the primitive builtin surface limited to the nine operations required by core IR consumers. Validation: - LLVM-enabled full C++ build - C++ tests: 130/130 passed - Focused and regression Python tests: 1,036 passed, 13 xfailed, 1 xpassed - clang-format, diff, namespace/name, comment/header, and Python byte-compilation audits No new test cases were added. -- 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]
