[clang] [llvm] [RISCV][P-ext] Add packed multiply high accumulate with byte/halfword index intrinsics (PR #224261)

[via cfe-commits]

================
@@ -8116,3 +8128,162 @@ define i64 @test_maccsu_w11_i64(i64 %rd, <2 x i32> %a, <2 x i32> %b) {
   %r = call i64 @llvm.riscv.maccsu.11.i64.v2i32(i64 %rd, <2 x i32> %a, <2 x i32> %b)
   ret i64 %r
 }
+
+; Packed multiply high accumulate with byte index (RV32 only form)
+define <2 x i16> @test_pmhacc_b0_v2i16(<2 x i16> %rd, <2 x i16> %rs1, <4 x i8> %rs2) {
----------------
sihuan wrote:

This (and test_pmhacc_b1_v2i16 / test_pmhaccsu_b0_v2i16 / test_pmhaccsu_b1_v2i16 below) is also defined in rvp-simd-32.ll with identical checks. Both files run on RV32 and RV64, so the coverage is duplicated.  Please drop the v2i16 copies here. 

https://github.com/llvm/llvm-project/pull/224261