================
@@ -120,7 +120,21 @@ llvm::Type *CodeGenTypes::ConvertTypeForMem(QualType T) {
}
return llvm::ArrayType::get(IRElemTy, MT->getNumElementsFlattened());
}
-
+ if (T->isCooperativeMatrixType()) {
----------------
asudarsa-qti wrote:
The ConvertTypeForMem handling is intended to provide the canonical
QualType-to-LLVM-type lowering for cooperative matrices. It should not
duplicate the type construction performed by CGBuiltin.
I will centralize construction of the spirv.CooperativeMatrixKHR TargetExtType
in CodeGenTypes and make CGBuiltin reuse that helper. This keeps CGBuiltin
responsible only for builtin-specific call emission and avoids maintaining two
independent cooperative-matrix type mappings
https://github.com/llvm/llvm-project/pull/221328
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
