tqchen opened a new pull request, #20392: URL: https://github.com/apache/tvm/pull/20392
Primitive bitwise and shift expressions currently use generic intrinsic calls, requiring separate matching logic in analysis and lowering. This change introduces LShift, RShift, BitwiseAnd, BitwiseOr, BitwiseXor, and BitwiseNot nodes and migrates compiler consumers to typed operands. Public builders and operators retain their validation and constant-folding behavior. The migration covers reflection and traversal, Python and TVMScript, symbolic analysis including Z3, Relax and TIR transforms, and C/LLVM/SPIR-V/WebGPU code generation. Existing tests receive representation-only updates. Validation: 2,046 existing tests passed, 462 skipped, and 8 expected failures across C++, symbolic analysis, IR/script, transforms, and available backend suites. LLVM, CUDA, WebGPU codegen and Z3 compiled successfully; GPU runtime tests skipped because no GPU was exposed. SPIR-V source was reviewed but not compiled or executed. Applicable formatting and lint checks passed. -- 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]
