================ @@ -83,6 +83,7 @@ def int_dx_umad : DefaultAttrsIntrinsic<[llvm_anyint_ty], [LLVMMatchType<0>, LLV def int_dx_normalize : DefaultAttrsIntrinsic<[LLVMMatchType<0>], [llvm_anyfloat_ty], [IntrNoMem]>; def int_dx_rsqrt : DefaultAttrsIntrinsic<[llvm_anyfloat_ty], [LLVMMatchType<0>], [IntrNoMem]>; def int_dx_wave_is_first_lane : DefaultAttrsIntrinsic<[llvm_i1_ty], [], [IntrConvergent]>; +def int_dx_wave_read_lane_at : DefaultAttrsIntrinsic<[llvm_any_ty], [LLVMMatchType<0>, llvm_i32_ty], [IntrConvergent]>; ---------------- inbelic wrote:
My understanding was that `IntrNoMem` denotes there are no side-effects in the operation and that cross-lane communication would be a side-effect. https://github.com/llvm/llvm-project/pull/111010 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
