[llvm] [AMDGPU] Add new llvm.amdgcn.wave.shuffle intrinsic (PR #167372)

[via llvm-commits]

================
@@ -0,0 +1,76 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
+; RUN: llc -global-isel=0 -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 < %s | FileCheck -check-prefixes=GFX11 %s
+; RUN: llc -global-isel=0 -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1200 < %s | FileCheck -check-prefixes=GFX12 %s
+
+; RUN: llc -global-isel=0 -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX11-64 %s
+; RUN: llc -global-isel=0 -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1200 -mattr=+wavefrontsize64 < %s | FileCheck -check-prefixes=GFX12-64 %s
+
+declare float @llvm.amdgcn.subgroup.shuffle.float(float, i32)
+
+define float @test_subgroup_shuffle_scalar(float %val, i32 %idx) {
+; GFX11-LABEL: test_subgroup_shuffle_scalar:
+; GFX11:       ; %bb.0: ; %entry
+; GFX11-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-NEXT:    v_lshlrev_b32_e32 v1, 2, v1
----------------
saxlungs wrote:

Per below conversation, we're just going with cpp implementations due to the lowering logic being complicated

https://github.com/llvm/llvm-project/pull/167372