Skip to content

[HLSL] implement clamp intrinsic #85424

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.

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 3 commits into from
Mar 16, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
6 changes: 6 additions & 0 deletions clang/include/clang/Basic/Builtins.td
Original file line number Diff line number Diff line change
Expand Up @@ -4554,6 +4554,12 @@ def HLSLWaveActiveCountBits : LangBuiltin<"HLSL_LANG"> {
let Prototype = "unsigned int(bool)";
}

def HLSLClamp : LangBuiltin<"HLSL_LANG"> {
let Spellings = ["__builtin_hlsl_elementwise_clamp"];
let Attributes = [NoThrow, Const];
let Prototype = "void(...)";
}

def HLSLCreateHandle : LangBuiltin<"HLSL_LANG"> {
let Spellings = ["__builtin_hlsl_create_handle"];
let Attributes = [NoThrow, Const];
Expand Down
15 changes: 15 additions & 0 deletions clang/lib/CodeGen/CGBuiltin.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -17981,6 +17981,21 @@ Value *CodeGenFunction::EmitHLSLBuiltinExpr(unsigned BuiltinID,
/*ReturnType=*/llvm::Type::getInt1Ty(getLLVMContext()),
Intrinsic::dx_any, ArrayRef<Value *>{Op0}, nullptr, "dx.any");
}
case Builtin::BI__builtin_hlsl_elementwise_clamp: {
Value *OpX = EmitScalarExpr(E->getArg(0));
Value *OpMin = EmitScalarExpr(E->getArg(1));
Value *OpMax = EmitScalarExpr(E->getArg(2));

QualType Ty = E->getArg(0)->getType();
bool IsUnsigned = false;
if (auto *VecTy = Ty->getAs<VectorType>())
Ty = VecTy->getElementType();
IsUnsigned = Ty->isUnsignedIntegerType();
return Builder.CreateIntrinsic(
/*ReturnType=*/OpX->getType(),
IsUnsigned ? Intrinsic::dx_uclamp : Intrinsic::dx_clamp,
ArrayRef<Value *>{OpX, OpMin, OpMax}, nullptr, "dx.clamp");
}
case Builtin::BI__builtin_hlsl_dot: {
Value *Op0 = EmitScalarExpr(E->getArg(0));
Value *Op1 = EmitScalarExpr(E->getArg(1));
Expand Down
110 changes: 110 additions & 0 deletions clang/lib/Headers/hlsl/hlsl_intrinsics.h
Original file line number Diff line number Diff line change
Expand Up @@ -252,6 +252,116 @@ double3 ceil(double3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_ceil)
double4 ceil(double4);

//===----------------------------------------------------------------------===//
// clamp builtins
//===----------------------------------------------------------------------===//

/// \fn T clamp(T X, T Min, T Max)
/// \brief Clamps the specified value \a X to the specified
/// minimum ( \a Min) and maximum ( \a Max) range.
/// \param X A value to clamp.
/// \param Min The specified minimum range.
/// \param Max The specified maximum range.
///
/// Returns The clamped value for the \a X parameter.
/// For values of -INF or INF, clamp will behave as expected.
/// However for values of NaN, the results are undefined.

_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
half clamp(half, half, half);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
half2 clamp(half2, half2, half2);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
half3 clamp(half3, half3, half3);
_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
half4 clamp(half4, half4, half4);

#ifdef __HLSL_ENABLE_16_BIT
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int16_t clamp(int16_t, int16_t, int16_t);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int16_t2 clamp(int16_t2, int16_t2, int16_t2);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int16_t3 clamp(int16_t3, int16_t3, int16_t3);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int16_t4 clamp(int16_t4, int16_t4, int16_t4);

_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint16_t clamp(uint16_t, uint16_t, uint16_t);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint16_t2 clamp(uint16_t2, uint16_t2, uint16_t2);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint16_t3 clamp(uint16_t3, uint16_t3, uint16_t3);
_HLSL_AVAILABILITY(shadermodel, 6.2)
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint16_t4 clamp(uint16_t4, uint16_t4, uint16_t4);
#endif

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int clamp(int, int, int);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int2 clamp(int2, int2, int2);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int3 clamp(int3, int3, int3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int4 clamp(int4, int4, int4);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint clamp(uint, uint, uint);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint2 clamp(uint2, uint2, uint2);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint3 clamp(uint3, uint3, uint3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint4 clamp(uint4, uint4, uint4);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int64_t clamp(int64_t, int64_t, int64_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int64_t2 clamp(int64_t2, int64_t2, int64_t2);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int64_t3 clamp(int64_t3, int64_t3, int64_t3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
int64_t4 clamp(int64_t4, int64_t4, int64_t4);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint64_t clamp(uint64_t, uint64_t, uint64_t);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint64_t2 clamp(uint64_t2, uint64_t2, uint64_t2);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint64_t3 clamp(uint64_t3, uint64_t3, uint64_t3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
uint64_t4 clamp(uint64_t4, uint64_t4, uint64_t4);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
float clamp(float, float, float);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
float2 clamp(float2, float2, float2);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
float3 clamp(float3, float3, float3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
float4 clamp(float4, float4, float4);

_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
double clamp(double, double, double);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
double2 clamp(double2, double2, double2);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
double3 clamp(double3, double3, double3);
_HLSL_BUILTIN_ALIAS(__builtin_hlsl_elementwise_clamp)
double4 clamp(double4, double4, double4);

//===----------------------------------------------------------------------===//
// cos builtins
//===----------------------------------------------------------------------===//
Expand Down
15 changes: 14 additions & 1 deletion clang/lib/Sema/SemaChecking.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -5303,6 +5303,17 @@ bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
return true;
break;
}
case Builtin::BI__builtin_hlsl_elementwise_clamp: {
if (checkArgCount(*this, TheCall, 3))
return true;
if (CheckVectorElementCallArgs(this, TheCall))
return true;
if (SemaBuiltinElementwiseTernaryMath(
TheCall, /*CheckForFloatArgs*/
TheCall->getArg(0)->getType()->hasFloatingRepresentation()))
return true;
break;
}
case Builtin::BI__builtin_hlsl_dot: {
if (checkArgCount(*this, TheCall, 2))
return true;
Expand Down Expand Up @@ -5351,7 +5362,9 @@ bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
return true;
if (CheckVectorElementCallArgs(this, TheCall))
return true;
if (SemaBuiltinElementwiseTernaryMath(TheCall, /*CheckForFloatArgs*/ false))
if (SemaBuiltinElementwiseTernaryMath(
TheCall, /*CheckForFloatArgs*/
TheCall->getArg(0)->getType()->hasFloatingRepresentation()))
return true;
}
}
Expand Down
8 changes: 8 additions & 0 deletions clang/test/CodeGenHLSL/builtins/clamp-builtin.hlsl
Original file line number Diff line number Diff line change
@@ -0,0 +1,8 @@
// RUN: %clang_cc1 -finclude-default-header -x hlsl -triple dxil-pc-shadermodel6.3-library %s -fnative-half-type -emit-llvm -disable-llvm-passes -o - | FileCheck %s

// CHECK-LABEL: builtin_test_clamp_int4
// CHECK: %dx.clamp = call <4 x i32> @llvm.dx.clamp.v4i32(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2)
// CHECK: ret <4 x i32> %dx.clamp
int4 builtin_test_clamp_int4(int4 p0, int4 p1, int4 p2) {
return __builtin_hlsl_elementwise_clamp(p0, p1, p2);
}
134 changes: 134 additions & 0 deletions clang/test/CodeGenHLSL/builtins/clamp.hlsl
Original file line number Diff line number Diff line change
@@ -0,0 +1,134 @@
// RUN: %clang_cc1 -std=hlsl2021 -finclude-default-header -x hlsl -triple \
// RUN: dxil-pc-shadermodel6.3-library %s -fnative-half-type \
// RUN: -emit-llvm -disable-llvm-passes -o - | FileCheck %s \
// RUN: --check-prefixes=CHECK,NATIVE_HALF
// RUN: %clang_cc1 -std=hlsl2021 -finclude-default-header -x hlsl -triple \
// RUN: dxil-pc-shadermodel6.3-library %s -emit-llvm -disable-llvm-passes \
// RUN: -o - | FileCheck %s --check-prefixes=CHECK,NO_HALF

#ifdef __HLSL_ENABLE_16_BIT
// NATIVE_HALF: define noundef i16 @
// NATIVE_HALF: call i16 @llvm.dx.clamp.i16(
int16_t test_clamp_short(int16_t p0, int16_t p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <2 x i16> @
// NATIVE_HALF: call <2 x i16> @llvm.dx.clamp.v2i16(
int16_t2 test_clamp_short2(int16_t2 p0, int16_t2 p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <3 x i16> @
// NATIVE_HALF: call <3 x i16> @llvm.dx.clamp.v3i16
int16_t3 test_clamp_short3(int16_t3 p0, int16_t3 p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <4 x i16> @
// NATIVE_HALF: call <4 x i16> @llvm.dx.clamp.v4i16
int16_t4 test_clamp_short4(int16_t4 p0, int16_t4 p1) { return clamp(p0, p1,p1); }

// NATIVE_HALF: define noundef i16 @
// NATIVE_HALF: call i16 @llvm.dx.uclamp.i16(
uint16_t test_clamp_ushort(uint16_t p0, uint16_t p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <2 x i16> @
// NATIVE_HALF: call <2 x i16> @llvm.dx.uclamp.v2i16
uint16_t2 test_clamp_ushort2(uint16_t2 p0, uint16_t2 p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <3 x i16> @
// NATIVE_HALF: call <3 x i16> @llvm.dx.uclamp.v3i16
uint16_t3 test_clamp_ushort3(uint16_t3 p0, uint16_t3 p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <4 x i16> @
// NATIVE_HALF: call <4 x i16> @llvm.dx.uclamp.v4i16
uint16_t4 test_clamp_ushort4(uint16_t4 p0, uint16_t4 p1) { return clamp(p0, p1,p1); }
#endif

// CHECK: define noundef i32 @
// CHECK: call i32 @llvm.dx.clamp.i32(
int test_clamp_int(int p0, int p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <2 x i32> @
// CHECK: call <2 x i32> @llvm.dx.clamp.v2i32
int2 test_clamp_int2(int2 p0, int2 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <3 x i32> @
// CHECK: call <3 x i32> @llvm.dx.clamp.v3i32
int3 test_clamp_int3(int3 p0, int3 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <4 x i32> @
// CHECK: call <4 x i32> @llvm.dx.clamp.v4i32
int4 test_clamp_int4(int4 p0, int4 p1) { return clamp(p0, p1,p1); }

// CHECK: define noundef i32 @
// CHECK: call i32 @llvm.dx.uclamp.i32(
int test_clamp_uint(uint p0, uint p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <2 x i32> @
// CHECK: call <2 x i32> @llvm.dx.uclamp.v2i32
uint2 test_clamp_uint2(uint2 p0, uint2 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <3 x i32> @
// CHECK: call <3 x i32> @llvm.dx.uclamp.v3i32
uint3 test_clamp_uint3(uint3 p0, uint3 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <4 x i32> @
// CHECK: call <4 x i32> @llvm.dx.uclamp.v4i32
uint4 test_clamp_uint4(uint4 p0, uint4 p1) { return clamp(p0, p1,p1); }

// CHECK: define noundef i64 @
// CHECK: call i64 @llvm.dx.clamp.i64(
int64_t test_clamp_long(int64_t p0, int64_t p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <2 x i64> @
// CHECK: call <2 x i64> @llvm.dx.clamp.v2i64
int64_t2 test_clamp_long2(int64_t2 p0, int64_t2 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <3 x i64> @
// CHECK: call <3 x i64> @llvm.dx.clamp.v3i64
int64_t3 test_clamp_long3(int64_t3 p0, int64_t3 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <4 x i64> @
// CHECK: call <4 x i64> @llvm.dx.clamp.v4i64
int64_t4 test_clamp_long4(int64_t4 p0, int64_t4 p1) { return clamp(p0, p1,p1); }

// CHECK: define noundef i64 @
// CHECK: call i64 @llvm.dx.uclamp.i64(
uint64_t test_clamp_long(uint64_t p0, uint64_t p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <2 x i64> @
// CHECK: call <2 x i64> @llvm.dx.uclamp.v2i64
uint64_t2 test_clamp_long2(uint64_t2 p0, uint64_t2 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <3 x i64> @
// CHECK: call <3 x i64> @llvm.dx.uclamp.v3i64
uint64_t3 test_clamp_long3(uint64_t3 p0, uint64_t3 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <4 x i64> @
// CHECK: call <4 x i64> @llvm.dx.uclamp.v4i64
uint64_t4 test_clamp_long4(uint64_t4 p0, uint64_t4 p1) { return clamp(p0, p1,p1); }

// NATIVE_HALF: define noundef half @
// NATIVE_HALF: call half @llvm.dx.clamp.f16(
// NO_HALF: define noundef float @"?test_clamp_half
// NO_HALF: call float @llvm.dx.clamp.f32(
half test_clamp_half(half p0, half p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <2 x half> @
// NATIVE_HALF: call <2 x half> @llvm.dx.clamp.v2f16
// NO_HALF: define noundef <2 x float> @"?test_clamp_half2
// NO_HALF: call <2 x float> @llvm.dx.clamp.v2f32(
half2 test_clamp_half2(half2 p0, half2 p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <3 x half> @
// NATIVE_HALF: call <3 x half> @llvm.dx.clamp.v3f16
// NO_HALF: define noundef <3 x float> @"?test_clamp_half3
// NO_HALF: call <3 x float> @llvm.dx.clamp.v3f32(
half3 test_clamp_half3(half3 p0, half3 p1) { return clamp(p0, p1,p1); }
// NATIVE_HALF: define noundef <4 x half> @
// NATIVE_HALF: call <4 x half> @llvm.dx.clamp.v4f16
// NO_HALF: define noundef <4 x float> @"?test_clamp_half4
// NO_HALF: call <4 x float> @llvm.dx.clamp.v4f32(
half4 test_clamp_half4(half4 p0, half4 p1) { return clamp(p0, p1,p1); }

// CHECK: define noundef float @"?test_clamp_float
// CHECK: call float @llvm.dx.clamp.f32(
float test_clamp_float(float p0, float p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <2 x float> @"?test_clamp_float2
// CHECK: call <2 x float> @llvm.dx.clamp.v2f32
float2 test_clamp_float2(float2 p0, float2 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <3 x float> @"?test_clamp_float3
// CHECK: call <3 x float> @llvm.dx.clamp.v3f32
float3 test_clamp_float3(float3 p0, float3 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <4 x float> @"?test_clamp_float4
// CHECK: call <4 x float> @llvm.dx.clamp.v4f32
float4 test_clamp_float4(float4 p0, float4 p1) { return clamp(p0, p1,p1); }

// CHECK: define noundef double @
// CHECK: call double @llvm.dx.clamp.f64(
double test_clamp_double(double p0, double p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <2 x double> @
// CHECK: call <2 x double> @llvm.dx.clamp.v2f64
double2 test_clamp_double2(double2 p0, double2 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <3 x double> @
// CHECK: call <3 x double> @llvm.dx.clamp.v3f64
double3 test_clamp_double3(double3 p0, double3 p1) { return clamp(p0, p1,p1); }
// CHECK: define noundef <4 x double> @
// CHECK: call <4 x double> @llvm.dx.clamp.v4f64
double4 test_clamp_double4(double4 p0, double4 p1) { return clamp(p0, p1,p1); }
Loading