[llvm] Add IIT_V6 to support 6-element vectors in intrinsics. (PR #88196)
NAKAMURA Takumi via llvm-commits
llvm-commits at lists.llvm.org
Tue Apr 9 15:04:29 PDT 2024
================
@@ -272,28 +272,29 @@ def IIT_F32 : IIT_VT<f32, 7>;
def IIT_F64 : IIT_VT<f64, 8>;
def IIT_V2 : IIT_Vec<2, 9>;
def IIT_V4 : IIT_Vec<4, 10>;
-def IIT_V8 : IIT_Vec<8, 11>;
-def IIT_V16 : IIT_Vec<16, 12>;
-def IIT_V32 : IIT_Vec<32, 13>;
-def IIT_PTR : IIT_Base< 14>;
-def IIT_ARG : IIT_Base< 15>;
-
-def IIT_V64 : IIT_Vec<64, 16>;
-def IIT_MMX : IIT_VT<x86mmx, 17>;
-def IIT_TOKEN : IIT_VT<token, 18>;
-def IIT_METADATA : IIT_VT<MetadataVT, 19>;
-def IIT_EMPTYSTRUCT : IIT_VT<OtherVT, 20>;
-def IIT_STRUCT2 : IIT_Base<21>;
-def IIT_STRUCT3 : IIT_Base<22>;
-def IIT_STRUCT4 : IIT_Base<23>;
-def IIT_STRUCT5 : IIT_Base<24>;
-def IIT_EXTEND_ARG : IIT_Base<25>;
-def IIT_TRUNC_ARG : IIT_Base<26>;
-def IIT_ANYPTR : IIT_Base<27>;
-def IIT_V1 : IIT_Vec<1, 28>;
-def IIT_VARARG : IIT_VT<isVoid, 29>;
-def IIT_HALF_VEC_ARG : IIT_Base<30>;
-def IIT_SAME_VEC_WIDTH_ARG : IIT_Base<31>;
+def IIT_V6 : IIT_Vec<6, 11>;
----------------
chapuni wrote:
You should avoid lower numbering unless it is used often. `<=16` is for short encoders.
https://github.com/llvm/llvm-project/pull/88196
More information about the llvm-commits
mailing list