derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
136 lines
5.8 KiB
Common Lisp
136 lines
5.8 KiB
Common Lisp
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --check-globals smart
|
|
// REQUIRES: amdgpu-registered-target
|
|
// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx950 -emit-llvm -o - %s | FileCheck %s -check-prefixes=GFX,GFX950
|
|
// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx9-4-generic -emit-llvm -o - %s | FileCheck %s -check-prefixes=GFX,GFX9_4_GENERIC
|
|
// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx1250 -emit-llvm -o - %s | FileCheck %s -check-prefixes=GFX,GFX1250
|
|
// RUN: %clang_cc1 -triple amdgcn-unknown-unknown -target-cpu gfx12-generic -emit-llvm -o - %s | FileCheck %s -check-prefixes=GFX,GFX12_GENERIC
|
|
|
|
typedef __attribute__((__vector_size__(4 * sizeof(unsigned int)))) unsigned int v4u32;
|
|
typedef v4u32 __global *global_ptr_to_v4u32;
|
|
|
|
//------------------------------------------------------------------------------
|
|
// Store
|
|
//------------------------------------------------------------------------------
|
|
// GFX-LABEL: @test_amdgcn_global_store_b128_00(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: tail call void @llvm.amdgcn.global.store.b128(ptr addrspace(1) [[PTR:%.*]], <4 x i32> [[DATA:%.*]], metadata [[META8:![0-9]+]])
|
|
// GFX-NEXT: ret void
|
|
//
|
|
void test_amdgcn_global_store_b128_00(global_ptr_to_v4u32 ptr, v4u32 data) {
|
|
__builtin_amdgcn_global_store_b128(ptr, data, "wavefront");
|
|
}
|
|
|
|
// GFX-LABEL: @test_amdgcn_global_store_b128_01(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: tail call void @llvm.amdgcn.global.store.b128(ptr addrspace(1) [[PTR:%.*]], <4 x i32> [[DATA:%.*]], metadata [[META9:![0-9]+]])
|
|
// GFX-NEXT: ret void
|
|
//
|
|
void test_amdgcn_global_store_b128_01(global_ptr_to_v4u32 ptr, v4u32 data) {
|
|
__builtin_amdgcn_global_store_b128(ptr, data, "workgroup");
|
|
}
|
|
|
|
// GFX-LABEL: @test_amdgcn_global_store_b128_cluster(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: tail call void @llvm.amdgcn.global.store.b128(ptr addrspace(1) [[PTR:%.*]], <4 x i32> [[DATA:%.*]], metadata [[META10:![0-9]+]])
|
|
// GFX-NEXT: ret void
|
|
//
|
|
void test_amdgcn_global_store_b128_cluster(global_ptr_to_v4u32 ptr, v4u32 data) {
|
|
__builtin_amdgcn_global_store_b128(ptr, data, "cluster");
|
|
}
|
|
|
|
// GFX-LABEL: @test_amdgcn_global_store_b128_10(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: tail call void @llvm.amdgcn.global.store.b128(ptr addrspace(1) [[PTR:%.*]], <4 x i32> [[DATA:%.*]], metadata [[META11:![0-9]+]])
|
|
// GFX-NEXT: ret void
|
|
//
|
|
void test_amdgcn_global_store_b128_10(global_ptr_to_v4u32 ptr, v4u32 data) {
|
|
__builtin_amdgcn_global_store_b128(ptr, data, "agent");
|
|
}
|
|
|
|
// GFX-LABEL: @test_amdgcn_global_store_b128_11(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: tail call void @llvm.amdgcn.global.store.b128(ptr addrspace(1) [[PTR:%.*]], <4 x i32> [[DATA:%.*]], metadata [[META12:![0-9]+]])
|
|
// GFX-NEXT: ret void
|
|
//
|
|
void test_amdgcn_global_store_b128_11(global_ptr_to_v4u32 ptr, v4u32 data) {
|
|
__builtin_amdgcn_global_store_b128(ptr, data, "");
|
|
}
|
|
|
|
//------------------------------------------------------------------------------
|
|
// Load
|
|
//------------------------------------------------------------------------------
|
|
// GFX-LABEL: @test_amdgcn_global_load_b128_00(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.global.load.b128(ptr addrspace(1) [[PTR:%.*]], metadata [[META8]])
|
|
// GFX-NEXT: ret <4 x i32> [[TMP0]]
|
|
//
|
|
v4u32 test_amdgcn_global_load_b128_00(global_ptr_to_v4u32 ptr) {
|
|
return __builtin_amdgcn_global_load_b128(ptr, "wavefront");
|
|
}
|
|
|
|
// GFX-LABEL: @test_amdgcn_global_load_b128_01(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.global.load.b128(ptr addrspace(1) [[PTR:%.*]], metadata [[META9]])
|
|
// GFX-NEXT: ret <4 x i32> [[TMP0]]
|
|
//
|
|
v4u32 test_amdgcn_global_load_b128_01(global_ptr_to_v4u32 ptr) {
|
|
return __builtin_amdgcn_global_load_b128(ptr, "workgroup");
|
|
}
|
|
|
|
// GFX-LABEL: @test_amdgcn_global_load_b128_cluster(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.global.load.b128(ptr addrspace(1) [[PTR:%.*]], metadata [[META10]])
|
|
// GFX-NEXT: ret <4 x i32> [[TMP0]]
|
|
//
|
|
v4u32 test_amdgcn_global_load_b128_cluster(global_ptr_to_v4u32 ptr) {
|
|
return __builtin_amdgcn_global_load_b128(ptr, "cluster");
|
|
}
|
|
|
|
// GFX-LABEL: @test_amdgcn_global_load_b128_10(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.global.load.b128(ptr addrspace(1) [[PTR:%.*]], metadata [[META11]])
|
|
// GFX-NEXT: ret <4 x i32> [[TMP0]]
|
|
//
|
|
v4u32 test_amdgcn_global_load_b128_10(global_ptr_to_v4u32 ptr) {
|
|
return __builtin_amdgcn_global_load_b128(ptr, "agent");
|
|
}
|
|
|
|
// GFX-LABEL: @test_amdgcn_global_load_b128_11(
|
|
// GFX-NEXT: entry:
|
|
// GFX-NEXT: [[TMP0:%.*]] = tail call <4 x i32> @llvm.amdgcn.global.load.b128(ptr addrspace(1) [[PTR:%.*]], metadata [[META12]])
|
|
// GFX-NEXT: ret <4 x i32> [[TMP0]]
|
|
//
|
|
v4u32 test_amdgcn_global_load_b128_11(global_ptr_to_v4u32 ptr) {
|
|
return __builtin_amdgcn_global_load_b128(ptr, "");
|
|
}
|
|
//.
|
|
// GFX950: [[META8]] = !{!"wavefront"}
|
|
// GFX950: [[META9]] = !{!"workgroup"}
|
|
// GFX950: [[META10]] = !{!"cluster"}
|
|
// GFX950: [[META11]] = !{!"agent"}
|
|
// GFX950: [[META12]] = !{!""}
|
|
//.
|
|
// GFX9_4_GENERIC: [[META8]] = !{!"wavefront"}
|
|
// GFX9_4_GENERIC: [[META9]] = !{!"workgroup"}
|
|
// GFX9_4_GENERIC: [[META10]] = !{!"cluster"}
|
|
// GFX9_4_GENERIC: [[META11]] = !{!"agent"}
|
|
// GFX9_4_GENERIC: [[META12]] = !{!""}
|
|
//.
|
|
// GFX1250: [[META8]] = !{!"wavefront"}
|
|
// GFX1250: [[META9]] = !{!"workgroup"}
|
|
// GFX1250: [[META10]] = !{!"cluster"}
|
|
// GFX1250: [[META11]] = !{!"agent"}
|
|
// GFX1250: [[META12]] = !{!""}
|
|
//.
|
|
// GFX12_GENERIC: [[META8]] = !{!"wavefront"}
|
|
// GFX12_GENERIC: [[META9]] = !{!"workgroup"}
|
|
// GFX12_GENERIC: [[META10]] = !{!"cluster"}
|
|
// GFX12_GENERIC: [[META11]] = !{!"agent"}
|
|
// GFX12_GENERIC: [[META12]] = !{!""}
|
|
//.
|
|
//// NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
|
|
// GFX1250: {{.*}}
|
|
// GFX12_GENERIC: {{.*}}
|
|
// GFX950: {{.*}}
|
|
// GFX9_4_GENERIC: {{.*}}
|