Skip to content

clang/AMDGPU: Emit atomicrmw from {global|flat}_atomic_fadd_v2f16 builtins #96873

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Aug 20, 2024
Merged

clang/AMDGPU: Emit atomicrmw from {global|flat}_atomic_fadd_v2f16 builtins #96873

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 6 additions & 14 deletions clang/lib/CodeGen/CGBuiltin.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18929,22 +18929,15 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
Function *F = CGM.getIntrinsic(Intrin, { Src0->getType() });
return Builder.CreateCall(F, { Src0, Builder.getFalse() });
}
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_v2f16:
case AMDGPU::BI__builtin_amdgcn_global_atomic_fmin_f64:
case AMDGPU::BI__builtin_amdgcn_global_atomic_fmax_f64:
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fadd_f64:
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fmin_f64:
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fmax_f64:
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fadd_f32:
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fadd_v2f16: {
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fadd_f32: {
Intrinsic::ID IID;
llvm::Type *ArgTy = llvm::Type::getDoubleTy(getLLVMContext());
switch (BuiltinID) {
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_v2f16:
ArgTy = llvm::FixedVectorType::get(
llvm::Type::getHalfTy(getLLVMContext()), 2);
IID = Intrinsic::amdgcn_global_atomic_fadd;
break;
case AMDGPU::BI__builtin_amdgcn_global_atomic_fmin_f64:
IID = Intrinsic::amdgcn_global_atomic_fmin;
break;
Expand All @@ -18964,11 +18957,6 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
ArgTy = llvm::Type::getFloatTy(getLLVMContext());
IID = Intrinsic::amdgcn_flat_atomic_fadd;
break;
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fadd_v2f16:
ArgTy = llvm::FixedVectorType::get(
llvm::Type::getHalfTy(getLLVMContext()), 2);
IID = Intrinsic::amdgcn_flat_atomic_fadd;
break;
}
llvm::Value *Addr = EmitScalarExpr(E->getArg(0));
llvm::Value *Val = EmitScalarExpr(E->getArg(1));
Expand Down Expand Up @@ -19369,7 +19357,9 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
case AMDGPU::BI__builtin_amdgcn_ds_fminf:
case AMDGPU::BI__builtin_amdgcn_ds_fmaxf:
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_f32:
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_f64: {
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_f64:
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_v2f16:
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fadd_v2f16: {
llvm::AtomicRMWInst::BinOp BinOp;
switch (BuiltinID) {
case AMDGPU::BI__builtin_amdgcn_atomic_inc32:
Expand All @@ -19387,6 +19377,8 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
case AMDGPU::BI__builtin_amdgcn_ds_atomic_fadd_v2bf16:
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_f32:
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_f64:
case AMDGPU::BI__builtin_amdgcn_global_atomic_fadd_v2f16:
case AMDGPU::BI__builtin_amdgcn_flat_atomic_fadd_v2f16:
BinOp = llvm::AtomicRMWInst::FAdd;
break;
case AMDGPU::BI__builtin_amdgcn_ds_fminf:
Expand Down
9 changes: 6 additions & 3 deletions clang/test/CodeGenOpenCL/builtins-fp-atomics-gfx12.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,7 +48,8 @@ void test_local_add_2f16_noret(__local half2 *addr, half2 x) {
}

// CHECK-LABEL: test_flat_add_2f16
// CHECK: call <2 x half> @llvm.amdgcn.flat.atomic.fadd.v2f16.p0.v2f16(ptr %{{.*}}, <2 x half> %{{.*}})
// CHECK: [[RMW:%.+]] = atomicrmw fadd ptr %{{.+}}, <2 x half> %{{.+}} syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory !{{[0-9]+$}}

// GFX12-LABEL: test_flat_add_2f16
// GFX12: flat_atomic_pk_add_f16
half2 test_flat_add_2f16(__generic half2 *addr, half2 x) {
Expand All @@ -64,7 +65,8 @@ short2 test_flat_add_2bf16(__generic short2 *addr, short2 x) {
}

// CHECK-LABEL: test_global_add_half2
// CHECK: call <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) %{{.*}}, <2 x half> %{{.*}})
// CHECK: [[RMW:%.+]] = atomicrmw fadd ptr addrspace(1) %{{.+}}, <2 x half> %{{.+}} syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory !{{[0-9]+$}}

// GFX12-LABEL: test_global_add_half2
// GFX12: global_atomic_pk_add_f16 v2, v[0:1], v2, off th:TH_ATOMIC_RETURN
void test_global_add_half2(__global half2 *addr, half2 x) {
Expand All @@ -73,7 +75,8 @@ void test_global_add_half2(__global half2 *addr, half2 x) {
}

// CHECK-LABEL: test_global_add_half2_noret
// CHECK: call <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) %{{.*}}, <2 x half> %{{.*}})
// CHECK: [[RMW:%.+]] = atomicrmw fadd ptr addrspace(1) %{{.+}}, <2 x half> %{{.+}} syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory !{{[0-9]+$}}

// GFX12-LABEL: test_global_add_half2_noret
// GFX12: global_atomic_pk_add_f16 v[0:1], v2, off
void test_global_add_half2_noret(__global half2 *addr, half2 x) {
Expand Down
2 changes: 1 addition & 1 deletion clang/test/CodeGenOpenCL/builtins-fp-atomics-gfx90a.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,7 @@ void test_global_add_f64(__global double *addr, double x) {
}

// CHECK-LABEL: test_global_add_half2
// CHECK: call <2 x half> @llvm.amdgcn.global.atomic.fadd.v2f16.p1.v2f16(ptr addrspace(1) %{{.*}}, <2 x half> %{{.*}})
// CHECK: = atomicrmw fadd ptr addrspace(1) %{{.+}}, <2 x half> %{{.+}} syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory !{{[0-9]+$}}
// GFX90A-LABEL: test_global_add_half2
// GFX90A: global_atomic_pk_add_f16 v2, v[0:1], v2, off glc
void test_global_add_half2(__global half2 *addr, half2 x) {
Expand Down
3 changes: 2 additions & 1 deletion clang/test/CodeGenOpenCL/builtins-fp-atomics-gfx940.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,8 @@ half2 test_flat_add_f32(__generic float *addr, float x) {
}

// CHECK-LABEL: test_flat_add_2f16
// CHECK: call <2 x half> @llvm.amdgcn.flat.atomic.fadd.v2f16.p0.v2f16(ptr %{{.*}}, <2 x half> %{{.*}})
// CHECK: [[RMW:%.+]] = atomicrmw fadd ptr %{{.+}}, <2 x half> %{{.+}} syncscope("agent") monotonic, align 4, !amdgpu.no.fine.grained.memory !{{[0-9]+$}}

// GFX940-LABEL: test_flat_add_2f16
// GFX940: flat_atomic_pk_add_f16
half2 test_flat_add_2f16(__generic half2 *addr, half2 x) {
Expand Down
Loading