[llvm] Pre-commit AMDGPU tests for masked load/store/scatter/gather (PR #104645)

[Krzysztof Drewniak via llvm-commits]

================
@@ -0,0 +1,58 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -S %s -passes=scalarize-masked-mem-intrin -mtriple=amdgcn-amd-amdhsa | FileCheck %s
+
+; COM: Test that, unlike on CPU targets, the mask doesn't get bitcast to a scalar,
+; COM: since, on GPUs, each i1 takes up at least one register and so they should
+; COM: be treated separately.
+
+define <2 x i32> @scalarize_v2i32(ptr %p, <2 x i1> %mask, <2 x i32> %passthru) {
----------------
krzysz00 wrote:

Ah, right, wave-uniform, yeah, I can add those

https://github.com/llvm/llvm-project/pull/104645