[clang] [llvm] [HLSL][SPIRV][DXIL] Implement `WaveActiveSum` intrinsic (PR #112400)
Finn Plummer via cfe-commits
cfe-commits at lists.llvm.org
Fri Nov 1 14:05:46 PDT 2024
================
@@ -85,6 +85,8 @@ 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_getlaneindex : DefaultAttrsIntrinsic<[llvm_i32_ty], [], [IntrConvergent, IntrNoMem]>;
+def int_dx_wave_active_sum : DefaultAttrsIntrinsic<[llvm_any_ty], [LLVMMatchType<0>], [IntrConvergent, IntrNoMem]>;
----------------
inbelic wrote:
No difference in how they are lowered to a DXIL op. They will both be lowered to a `WaveActiveOp` with the same flags.
https://github.com/llvm/llvm-project/pull/112400
More information about the cfe-commits
mailing list