[llvm] [AMDGPU][NFC] Pre-commit tests for f16<->i32 conversions (PR #176630)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,116 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
+; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1010 < %s | FileCheck %s --check-prefix=GFX10
+; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1150 < %s | FileCheck %s --check-prefix=GFX11
+; RUN: llc -global-isel=0 -mtriple=amdgcn -mcpu=gfx1200 < %s | FileCheck %s --check-prefix=GFX12
+
+define amdgpu_ps void @s_fptosi_f16_to_i32(half inreg %x, ptr addrspace(1) %out) {
----------------
arsenm wrote:

If you want to test scalar outputs, you should not be using a pointer output. Return in SGPR by using integer type with amdgpu_ps 

https://github.com/llvm/llvm-project/pull/176630