================ @@ -141,6 +141,16 @@ emitAMDGCNImageOverloadedReturnType(CIRGenFunction &cgf, const CallExpr *e, return callOp.getResult(); } +static mlir::Value emitAMDGPUPredicate(CIRGenFunction &cgf, const CallExpr *e, ---------------- skc7 wrote:
Small consistency nit: Helper is `emitAMDGPUPredicate` matching classic `GetAMDGPUPredicate` and rest of the code in CIR. (`cir.amdgpu_*`, `emitAMDGPUBuiltinExpr`). The op created is `cir.amdgcn_feature_predicate`. Could the op be renamed to `cir.amdgpu_feature_predicate` so all the names are sync. Just a thought since recently triple have also moved to amdgpu*-amd-amdhsa. https://github.com/llvm/llvm-project/pull/229330 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
