[clang] [Clang][AMDGPU] Add builtins for instrinsic `llvm.amdgcn.raw.buffer.store` (PR #94576)

[Matt Arsenault via cfe-commits]

================
@@ -0,0 +1,264 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu verde -emit-llvm -o - %s | FileCheck %s --check-prefixes=VERDE
+// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu tonga -emit-llvm -o - %s | FileCheck %s --check-prefixes=GFX8
+// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx1100 -emit-llvm -o - %s | FileCheck %s --check-prefixes=GFX11
+
+#pragma OPENCL EXTENSION cl_khr_fp16 : enable
+
+typedef short v2i16 __attribute__((ext_vector_type(2)));
+typedef int v2i32 __attribute__((ext_vector_type(2)));
+typedef half v2f16 __attribute__((ext_vector_type(2)));
+typedef float v2f32 __attribute__((ext_vector_type(2)));
+typedef short v4i16 __attribute__((ext_vector_type(4)));
+typedef int v4i32 __attribute__((ext_vector_type(4)));
+typedef half v4f16 __attribute__((ext_vector_type(4)));
+typedef float v4f32 __attribute__((ext_vector_type(4)));
+
+// VERDE-LABEL: @test_amdgcn_raw_buffer_store_i8(
+// VERDE-NEXT:  entry:
+// VERDE-NEXT:    tail call void @llvm.amdgcn.raw.buffer.store.i8(i8 [[VDATA:%.*]], <4 x i32> [[RSRC:%.*]], i32 0, i32 0, i32 0)
+// VERDE-NEXT:    ret void
+//
+// GFX8-LABEL: @test_amdgcn_raw_buffer_store_i8(
+// GFX8-NEXT:  entry:
+// GFX8-NEXT:    tail call void @llvm.amdgcn.raw.buffer.store.i8(i8 [[VDATA:%.*]], <4 x i32> [[RSRC:%.*]], i32 0, i32 0, i32 0)
+// GFX8-NEXT:    ret void
+//
+// GFX11-LABEL: @test_amdgcn_raw_buffer_store_i8(
+// GFX11-NEXT:  entry:
+// GFX11-NEXT:    tail call void @llvm.amdgcn.raw.buffer.store.i8(i8 [[VDATA:%.*]], <4 x i32> [[RSRC:%.*]], i32 0, i32 0, i32 0)
+// GFX11-NEXT:    ret void
+//
+void test_amdgcn_raw_buffer_store_i8(char vdata, v4i32 rsrc) {
----------------
arsenm wrote:

We are trying to move to using real pointers (i.e. use llvm.amdgcn.raw.ptr.buffer.store instead of llvm.amdgcn.raw.buffer.store) 

https://github.com/llvm/llvm-project/pull/94576