Skip to content

[CIR][AMDGPU] Add lowering for amdgcn sqrt builtins #2063

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
bcardosolopes merged 1 commit into from
Dec 24, 2025
+86 −1
Merged

[CIR][AMDGPU] Add lowering for amdgcn sqrt builtins #2063

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
3 changes: 2 additions & 1 deletion clang/lib/CIR/CodeGen/CIRGenBuiltinAMDGPU.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -326,7 +326,8 @@ mlir::Value CIRGenFunction::emitAMDGPUBuiltinExpr(unsigned builtinId,
case AMDGPU::BI__builtin_amdgcn_sqrtf:
case AMDGPU::BI__builtin_amdgcn_sqrth:
case AMDGPU::BI__builtin_amdgcn_sqrt_bf16: {
llvm_unreachable("sqrt_* NYI");
return emitBuiltinWithOneOverloadedType<1>(expr, "amdgcn.sqrt")
.getScalarVal();
}
case AMDGPU::BI__builtin_amdgcn_rsq:
case AMDGPU::BI__builtin_amdgcn_rsqf:
Expand Down
11 changes: 11 additions & 0 deletions clang/test/CIR/CodeGen/HIP/builtins-amdgcn-gfx1250.hip
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,3 +27,14 @@ __device__ void test_rcp_bf16(__bf16* out, __bf16 a)
{
*out = __builtin_amdgcn_rcp_bf16(a);
}

// CIR-LABEL: @_Z14test_sqrt_bf16PDF16bDF16b
// CIR: cir.llvm.intrinsic "amdgcn.sqrt" {{.*}} : (!cir.bf16) -> !cir.bf16
// LLVM: define{{.*}} void @_Z14test_sqrt_bf16PDF16bDF16b
// LLVM: call{{.*}} bfloat @llvm.amdgcn.sqrt.bf16(bfloat %{{.*}})
// OGCG: define{{.*}} void @_Z14test_sqrt_bf16PDF16bDF16b
// OGCG: call{{.*}} bfloat @llvm.amdgcn.sqrt.bf16(bfloat %{{.*}})
__device__ void test_sqrt_bf16(__bf16* out, __bf16 a)
{
*out = __builtin_amdgcn_sqrt_bf16(a);
}
11 changes: 11 additions & 0 deletions clang/test/CIR/CodeGen/HIP/builtins-amdgcn-vi.hip
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,3 +76,14 @@ __device__ void test_rcp_f16(_Float16* out, _Float16 a)
{
*out = __builtin_amdgcn_rcph(a);
}

// CIR-LABEL: @_Z13test_sqrt_f16PDF16_DF16_
// CIR: cir.llvm.intrinsic "amdgcn.sqrt" {{.*}} : (!cir.f16) -> !cir.f16
// LLVM: define{{.*}} void @_Z13test_sqrt_f16PDF16_DF16_
// LLVM: call{{.*}} half @llvm.{{((amdgcn.){0,1})}}sqrt.f16(half %{{.*}})
// OGCG: define{{.*}} void @_Z13test_sqrt_f16PDF16_DF16_
// OGCG: call{{.*}} half @llvm.{{((amdgcn.){0,1})}}sqrt.f16(half %{{.*}})
__device__ void test_sqrt_f16(_Float16* out, _Float16 a)
{
*out = __builtin_amdgcn_sqrth(a);
}
20 changes: 20 additions & 0 deletions clang/test/CIR/CodeGen/HIP/builtins-amdgcn.hip
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -365,3 +365,23 @@ __device__ void test_rcp_f32(float* out, float a) {
__device__ void test_rcp_f64(double* out, double a) {
*out = __builtin_amdgcn_rcp(a);
}

// CIR-LABEL: @_Z13test_sqrt_f32Pff
// CIR: cir.llvm.intrinsic "amdgcn.sqrt" {{.*}} : (!cir.float) -> !cir.float
// LLVM: define{{.*}} void @_Z13test_sqrt_f32Pff
// LLVM: call{{.*}} float @llvm.amdgcn.sqrt.f32(float %{{.*}})
// OGCG: define{{.*}} void @_Z13test_sqrt_f32Pff
// OGCG: call{{.*}} float @llvm.amdgcn.sqrt.f32(float %{{.*}})
__device__ void test_sqrt_f32(float* out, float a) {
*out = __builtin_amdgcn_sqrtf(a);
}

// CIR-LABEL: @_Z13test_sqrt_f64Pdd
// CIR: cir.llvm.intrinsic "amdgcn.sqrt" {{.*}} : (!cir.double) -> !cir.double
// LLVM: define{{.*}} void @_Z13test_sqrt_f64Pdd
// LLVM: call{{.*}} double @llvm.amdgcn.sqrt.f64(double %{{.*}})
// OGCG: define{{.*}} void @_Z13test_sqrt_f64Pdd
// OGCG: call{{.*}} double @llvm.amdgcn.sqrt.f64(double %{{.*}})
__device__ void test_sqrt_f64(double* out, double a) {
*out = __builtin_amdgcn_sqrt(a);
}
11 changes: 11 additions & 0 deletions clang/test/CIR/CodeGen/OpenCL/builtins-amdgcn-gfx1250.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,3 +27,14 @@ void test_rcp_bf16(global __bf16* out, __bf16 a)
{
*out = __builtin_amdgcn_rcp_bf16(a);
}

// CIR-LABEL: @test_sqrt_bf16
// CIR: cir.llvm.intrinsic "amdgcn.sqrt" {{.*}} : (!cir.bf16) -> !cir.bf16
// LLVM: define{{.*}} void @test_sqrt_bf16
// LLVM: call{{.*}} bfloat @llvm.amdgcn.sqrt.bf16(bfloat %{{.*}})
// OGCG: define{{.*}} void @test_sqrt_bf16
// OGCG: call{{.*}} bfloat @llvm.amdgcn.sqrt.bf16(bfloat %{{.*}})
void test_sqrt_bf16(global __bf16* out, __bf16 a)
{
*out = __builtin_amdgcn_sqrt_bf16(a);
}
11 changes: 11 additions & 0 deletions clang/test/CIR/CodeGen/OpenCL/builtins-amdgcn-vi.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,3 +76,14 @@ void test_rcp_f16(global half* out, half a)
{
*out = __builtin_amdgcn_rcph(a);
}

// CIR-LABEL: @test_sqrt_f16
// CIR: cir.llvm.intrinsic "amdgcn.sqrt" {{.*}} : (!cir.f16) -> !cir.f16
// LLVM: define{{.*}} void @test_sqrt_f16
// LLVM: call{{.*}} half @llvm.{{((amdgcn.){0,1})}}sqrt.f16(half %{{.*}})
// OGCG: define{{.*}} void @test_sqrt_f16
// OGCG: call{{.*}} half @llvm.{{((amdgcn.){0,1})}}sqrt.f16(half %{{.*}})
void test_sqrt_f16(global half* out, half a)
{
*out = __builtin_amdgcn_sqrth(a);
}
20 changes: 20 additions & 0 deletions clang/test/CIR/CodeGen/OpenCL/builtins_amdgcn.cl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -380,3 +380,23 @@ void test_rcp_f32(global float* out, float a) {
void test_rcp_f64(global double* out, double a) {
*out = __builtin_amdgcn_rcp(a);
}

// CIR-LABEL: @test_sqrt_f32
// CIR: cir.llvm.intrinsic "amdgcn.sqrt" {{.*}} : (!cir.float) -> !cir.float
// LLVM: define{{.*}} void @test_sqrt_f32
// LLVM: call{{.*}} float @llvm.amdgcn.sqrt.f32(float %{{.*}})
// OGCG: define{{.*}} void @test_sqrt_f32
// OGCG: call{{.*}} float @llvm.amdgcn.sqrt.f32(float %{{.*}})
void test_sqrt_f32(global float* out, float a) {
*out = __builtin_amdgcn_sqrtf(a);
}

// CIR-LABEL: @test_sqrt_f64
// CIR: cir.llvm.intrinsic "amdgcn.sqrt" {{.*}} : (!cir.double) -> !cir.double
// LLVM: define{{.*}} void @test_sqrt_f64
// LLVM: call{{.*}} double @llvm.amdgcn.sqrt.f64(double %{{.*}})
// OGCG: define{{.*}} void @test_sqrt_f64
// OGCG: call{{.*}} double @llvm.amdgcn.sqrt.f64(double %{{.*}})
void test_sqrt_f64(global double* out, double a) {
*out = __builtin_amdgcn_sqrt(a);
}