[HLSL] implementation of lerp intrinsic #83077
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
python3kgae
reviewed
Feb 26, 2024
farzonl
force-pushed
the
hlsl-lerp-intrinsic
branch
from
c6f533e
to
d4384c5
Compare
llvmbot
added
clang
|
@llvm/pr-subscribers-backend-x86 @llvm/pr-subscribers-hlsl Author: Farzon Lotfi (farzonl) ChangesThis is the start of implementing the lerp intrinsic https://learn.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl-lerp Builtins.td - defines the builtin this change implements the first half of #70102 Patch is 25.56 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/83077.diff 10 Files Affected:
diff --git a/clang/include/clang/Basic/Builtins.td b/clang/include/clang/Basic/Builtins.td
index e3432f7925ba14..755ef3631b2856 100644
--- a/clang/include/clang/Basic/Builtins.td
+++ b/clang/include/clang/Basic/Builtins.td
@@ -4530,6 +4530,12 @@ def HLSLDotProduct : LangBuiltin<"HLSL_LANG"> {
let Prototype = "void(...)";
}
+def HLSLLerp : LangBuiltin<"HLSL_LANG"> {
+ let Spellings = ["__builtin_hlsl_lerp"];
+ let Attributes = [NoThrow, Const];
+ let Prototype = "void(...)";
+}
+
// Builtins for XRay.
def XRayCustomEvent : Builtin {
let Spellings = ["__xray_customevent"];
diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index 57784a4ba2e388..63115d4bd49091 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -10266,6 +10266,11 @@ def err_block_on_vm : Error<
def err_sizeless_nonlocal : Error<
"non-local variable with sizeless type %0">;
+def err_vec_builtin_non_vector_all : Error<
+ "all arguments to %0 must be vectors">;
+def err_vec_builtin_incompatible_vector_all : Error<
+ "all arguments to %0 must have vectors of the same type">;
+
def err_vec_builtin_non_vector : Error<
"first two arguments to %0 must be vectors">;
def err_vec_builtin_incompatible_vector : Error<
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 54d7451a9d6221..17b7af924e5ff0 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -18006,6 +18006,43 @@ Value *CodeGenFunction::EmitHLSLBuiltinExpr(unsigned BuiltinID,
/*ReturnType*/ T0->getScalarType(), Intrinsic::dx_dot,
ArrayRef<Value *>{Op0, Op1}, nullptr, "dx.dot");
} break;
+ case Builtin::BI__builtin_hlsl_lerp: {
+ Value *X = EmitScalarExpr(E->getArg(0));
+ Value *Y = EmitScalarExpr(E->getArg(1));
+ Value *S = EmitScalarExpr(E->getArg(2));
+ llvm::Type *Xty = X->getType();
+ llvm::Type *Yty = Y->getType();
+ llvm::Type *Sty = S->getType();
+ if (!Xty->isVectorTy() && !Yty->isVectorTy() && !Sty->isVectorTy()) {
+ if (Xty->isFloatingPointTy()) {
+ auto V = Builder.CreateFSub(Y, X);
+ V = Builder.CreateFMul(S, V);
+ return Builder.CreateFAdd(X, V, "dx.lerp");
+ }
+ llvm_unreachable("Scalar Lerp is only supported on floats.");
+ }
+ // A VectorSplat should have happened
+ assert(Xty->isVectorTy() && Yty->isVectorTy() && Sty->isVectorTy() &&
+ "Lerp of vector and scalar is not supported.");
+
+ [[maybe_unused]] auto *XVecTy =
+ E->getArg(0)->getType()->getAs<VectorType>();
+ [[maybe_unused]] auto *YVecTy =
+ E->getArg(1)->getType()->getAs<VectorType>();
+ [[maybe_unused]] auto *SVecTy =
+ E->getArg(2)->getType()->getAs<VectorType>();
+ // A HLSLVectorTruncation should have happend
+ assert(XVecTy->getNumElements() == YVecTy->getNumElements() &&
+ XVecTy->getNumElements() == SVecTy->getNumElements() &&
+ "Lerp requires vectors to be of the same size.");
+ assert(XVecTy->getElementType()->isRealFloatingType() &&
+ XVecTy->getElementType() == YVecTy->getElementType() &&
+ XVecTy->getElementType() == SVecTy->getElementType() &&
+ "Lerp requires float vectors to be of the same type.");
+ return Builder.CreateIntrinsic(
+ /*ReturnType*/ Xty, Intrinsic::dx_lerp, ArrayRef<Value *>{X, Y, S},
+ nullptr, "dx.lerp");
+ }
}
return nullptr;
}
diff --git a/clang/lib/Headers/hlsl/hlsl_intrinsics.h b/clang/lib/Headers/hlsl/hlsl_intrinsics.h
index 08e5d981a4a4ca..1314bdefa37e7b 100644
--- a/clang/lib/Headers/hlsl/hlsl_intrinsics.h
+++ b/clang/lib/Headers/hlsl/hlsl_intrinsics.h
@@ -317,6 +317,42 @@ double3 floor(double3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_floor)
double4 floor(double4);
+//===----------------------------------------------------------------------===//
+// lerp builtins
+//===----------------------------------------------------------------------===//
+
+/// \fn T lerp(T x, T y, T s)
+/// \brief Returns the linear interpolation of x to y by s.
+/// \param x [in] The first-floating point value.
+/// \param y [in] The second-floating point value.
+/// \param s [in] A value that linearly interpolates between the x parameter and
+/// the y parameter.
+///
+/// Linear interpolation is based on the following formula: x*(1-s) + y*s which
+/// can equivalently be written as x + s(y-x).
+
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+half lerp(half, half, half);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+half2 lerp(half2, half2, half2);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+half3 lerp(half3, half3, half3);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+half4 lerp(half4, half4, half4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+float lerp(float, float, float);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+float2 lerp(float2, float2, float2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+float3 lerp(float3, float3, float3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+float4 lerp(float4, float4, float4);
+
//===----------------------------------------------------------------------===//
// log builtins
//===----------------------------------------------------------------------===//
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index 984088e345c806..0c7776b9d00e84 100644
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -5169,49 +5169,63 @@ bool Sema::CheckPPCMMAType(QualType Type, SourceLocation TypeLoc) {
bool CheckVectorElementCallArgs(Sema *S, CallExpr *TheCall) {
assert(TheCall->getNumArgs() > 1);
ExprResult A = TheCall->getArg(0);
- ExprResult B = TheCall->getArg(1);
+
QualType ArgTyA = A.get()->getType();
- QualType ArgTyB = B.get()->getType();
+
auto *VecTyA = ArgTyA->getAs<VectorType>();
- auto *VecTyB = ArgTyB->getAs<VectorType>();
SourceLocation BuiltinLoc = TheCall->getBeginLoc();
- if (VecTyA == nullptr && VecTyB == nullptr)
- return false;
- if (VecTyA && VecTyB) {
- bool retValue = false;
- if (VecTyA->getElementType() != VecTyB->getElementType()) {
- // Note: type promotion is intended to be handeled via the intrinsics
- // and not the builtin itself.
- S->Diag(TheCall->getBeginLoc(), diag::err_vec_builtin_incompatible_vector)
- << TheCall->getDirectCallee()
- << SourceRange(A.get()->getBeginLoc(), B.get()->getEndLoc());
- retValue = true;
- }
- if (VecTyA->getNumElements() != VecTyB->getNumElements()) {
- // if we get here a HLSLVectorTruncation is needed.
- S->Diag(BuiltinLoc, diag::err_vec_builtin_incompatible_vector)
- << TheCall->getDirectCallee()
- << SourceRange(TheCall->getArg(0)->getBeginLoc(),
- TheCall->getArg(1)->getEndLoc());
- retValue = true;
- }
+ for (unsigned i = 1; i < TheCall->getNumArgs(); ++i) {
+ ExprResult B = TheCall->getArg(i);
+ QualType ArgTyB = B.get()->getType();
+ auto *VecTyB = ArgTyB->getAs<VectorType>();
+ if (VecTyA == nullptr && VecTyB == nullptr)
+ return false;
- if (retValue)
- TheCall->setType(VecTyA->getElementType());
+ if (VecTyA && VecTyB) {
+ bool retValue = false;
+ if (VecTyA->getElementType() != VecTyB->getElementType()) {
+ // Note: type promotion is intended to be handeled via the intrinsics
+ // and not the builtin itself.
+ S->Diag(TheCall->getBeginLoc(),
+ diag::err_vec_builtin_incompatible_vector_all)
+ << TheCall->getDirectCallee()
+ << SourceRange(A.get()->getBeginLoc(), B.get()->getEndLoc());
+ retValue = true;
+ }
+ if (VecTyA->getNumElements() != VecTyB->getNumElements()) {
+ // if we get here a HLSLVectorTruncation is needed.
+ S->Diag(BuiltinLoc, diag::err_vec_builtin_incompatible_vector_all)
+ << TheCall->getDirectCallee()
+ << SourceRange(TheCall->getArg(0)->getBeginLoc(),
+ TheCall->getArg(1)->getEndLoc());
+ retValue = true;
+ }
+
+ if (!retValue)
+ TheCall->setType(VecTyA->getElementType());
- return retValue;
+ return retValue;
+ }
}
// Note: if we get here one of the args is a scalar which
// requires a VectorSplat on Arg0 or Arg1
- S->Diag(BuiltinLoc, diag::err_vec_builtin_non_vector)
+ S->Diag(BuiltinLoc, diag::err_vec_builtin_non_vector_all)
<< TheCall->getDirectCallee()
<< SourceRange(TheCall->getArg(0)->getBeginLoc(),
TheCall->getArg(1)->getEndLoc());
return true;
}
+bool checkAllArgsAreFloatRepresnations(CallExpr *TheCall) {
+ for (unsigned i = 0; i < TheCall->getNumArgs(); ++i) {
+ if (!TheCall->getArg(0)->getType()->hasFloatingRepresentation())
+ return true;
+ }
+ return false;
+}
+
// Note: returning true in this case results in CheckBuiltinFunctionCall
// returning an ExprError
bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
@@ -5225,6 +5239,17 @@ bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
return true;
break;
}
+ case Builtin::BI__builtin_hlsl_lerp: {
+ if (checkArgCount(*this, TheCall, 3))
+ return true;
+ if (CheckVectorElementCallArgs(this, TheCall))
+ return true;
+ if (SemaBuiltinElementwiseTernaryMath(TheCall))
+ return true;
+ if (checkAllArgsAreFloatRepresnations(TheCall))
+ return true;
+ break;
+ }
}
return false;
}
diff --git a/clang/test/CodeGenHLSL/builtins/lerp-builtin.hlsl b/clang/test/CodeGenHLSL/builtins/lerp-builtin.hlsl
new file mode 100644
index 00000000000000..1f16dec68212e4
--- /dev/null
+++ b/clang/test/CodeGenHLSL/builtins/lerp-builtin.hlsl
@@ -0,0 +1,37 @@
+// 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_lerp_half_scalar
+// CHECK: %3 = fsub double %conv1, %conv
+// CHECK: %4 = fmul double %conv2, %3
+// CHECK: %dx.lerp = fadd double %conv, %4
+// CHECK: %conv3 = fptrunc double %dx.lerp to half
+// CHECK: ret half %conv3
+half builtin_lerp_half_scalar (half p0) {
+ return __builtin_hlsl_lerp ( p0, p0, p0 );
+}
+
+// CHECK-LABEL: builtin_lerp_float_scalar
+// CHECK: %3 = fsub double %conv1, %conv
+// CHECK: %4 = fmul double %conv2, %3
+// CHECK: %dx.lerp = fadd double %conv, %4
+// CHECK: %conv3 = fptrunc double %dx.lerp to float
+// CHECK: ret float %conv3
+float builtin_lerp_float_scalar ( float p0) {
+ return __builtin_hlsl_lerp ( p0, p0, p0 );
+}
+
+// CHECK-LABEL: builtin_lerp_half_vector
+// CHECK: %dx.lerp = call <3 x half> @llvm.dx.lerp.v3f16(<3 x half> %0, <3 x half> %1, <3 x half> %2)
+// CHECK: ret <3 x half> %dx.lerp
+half3 builtin_lerp_half_vector (half3 p0) {
+ return __builtin_hlsl_lerp ( p0, p0, p0 );
+}
+
+// CHECK-LABEL: builtin_lerp_floar_vector
+// CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %2)
+// CHECK: ret <2 x float> %dx.lerp
+float2 builtin_lerp_floar_vector ( float2 p0) {
+ return __builtin_hlsl_lerp ( p0, p0, p0 );
+}
diff --git a/clang/test/CodeGenHLSL/builtins/lerp.hlsl b/clang/test/CodeGenHLSL/builtins/lerp.hlsl
new file mode 100644
index 00000000000000..1297f6b85bbd48
--- /dev/null
+++ b/clang/test/CodeGenHLSL/builtins/lerp.hlsl
@@ -0,0 +1,105 @@
+// RUN: %clang_cc1 -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 -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
+
+// NATIVE_HALF: %3 = fsub half %1, %0
+// NATIVE_HALF: %4 = fmul half %2, %3
+// NATIVE_HALF: %dx.lerp = fadd half %0, %4
+// NATIVE_HALF: ret half %dx.lerp
+// NO_HALF: %3 = fsub float %1, %0
+// NO_HALF: %4 = fmul float %2, %3
+// NO_HALF: %dx.lerp = fadd float %0, %4
+// NO_HALF: ret float %dx.lerp
+half test_lerp_half ( half p0) {
+ return lerp ( p0, p0, p0 );
+}
+
+// NATIVE_HALF: %dx.lerp = call <2 x half> @llvm.dx.lerp.v2f16(<2 x half> %0, <2 x half> %1, <2 x half> %2)
+// NATIVE_HALF: ret <2 x half> %dx.lerp
+// NO_HALF: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %2)
+// NO_HALF: ret <2 x float> %dx.lerp
+half2 test_lerp_half2 ( half2 p0, half2 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// NATIVE_HALF: %dx.lerp = call <3 x half> @llvm.dx.lerp.v3f16(<3 x half> %0, <3 x half> %1, <3 x half> %2)
+// NATIVE_HALF: ret <3 x half> %dx.lerp
+// NO_HALF: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %0, <3 x float> %1, <3 x float> %2)
+// NO_HALF: ret <3 x float> %dx.lerp
+half3 test_lerp_half3 ( half3 p0, half3 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// NATIVE_HALF: %dx.lerp = call <4 x half> @llvm.dx.lerp.v4f16(<4 x half> %0, <4 x half> %1, <4 x half> %2)
+// NATIVE_HALF: ret <4 x half> %dx.lerp
+// NO_HALF: %dx.lerp = call <4 x float> @llvm.dx.lerp.v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2)
+// NO_HALF: ret <4 x float> %dx.lerp
+half4 test_lerp_half4 ( half4 p0, half4 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %3 = fsub float %1, %0
+// CHECK: %4 = fmul float %2, %3
+// CHECK: %dx.lerp = fadd float %0, %4
+// CHECK: ret float %dx.lerp
+float test_lerp_float ( float p0, float p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %2)
+// CHECK: ret <2 x float> %dx.lerp
+float2 test_lerp_float2 ( float2 p0, float2 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %0, <3 x float> %1, <3 x float> %2)
+// CHECK: ret <3 x float> %dx.lerp
+float3 test_lerp_float3 ( float3 p0, float3 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %dx.lerp = call <4 x float> @llvm.dx.lerp.v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2)
+// CHECK: ret <4 x float> %dx.lerp
+float4 test_lerp_float4 ( float4 p0, float4 p1) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %splat.splat, <2 x float> %1, <2 x float> %2)
+// CHECK: ret <2 x float> %dx.lerp
+float2 test_lerp_float2_splat ( float p0, float2 p1 ) {
+ return lerp( p0, p1, p1 );
+}
+
+// CHECK: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %splat.splat, <3 x float> %1, <3 x float> %2)
+// CHECK: ret <3 x float> %dx.lerp
+float3 test_lerp_float3_splat ( float p0, float3 p1 ) {
+ return lerp( p0, p1, p1 );
+}
+
+// CHECK: %dx.lerp = call <4 x float> @llvm.dx.lerp.v4f32(<4 x float> %splat.splat, <4 x float> %1, <4 x float> %2)
+// CHECK: ret <4 x float> %dx.lerp
+float4 test_lerp_float4_splat ( float p0, float4 p1 ) {
+ return lerp( p0, p1, p1 );
+}
+
+// CHECK: %conv = sitofp i32 %2 to float
+// CHECK: %splat.splatinsert = insertelement <2 x float> poison, float %conv, i64 0
+// CHECK: %splat.splat = shufflevector <2 x float> %splat.splatinsert, <2 x float> poison, <2 x i32> zeroinitializer
+// CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %splat.splat)
+// CHECK: ret <2 x float> %dx.lerp
+float2 test_lerp_float2_int_splat ( float2 p0, int p1 ) {
+ return lerp ( p0, p0, p1 );
+}
+
+// CHECK: %conv = sitofp i32 %2 to float
+// CHECK: %splat.splatinsert = insertelement <3 x float> poison, float %conv, i64 0
+// CHECK: %splat.splat = shufflevector <3 x float> %splat.splatinsert, <3 x float> poison, <3 x i32> zeroinitializer
+// CHECK: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %0, <3 x float> %1, <3 x float> %splat.splat)
+// CHECK: ret <3 x float> %dx.lerp
+float3 test_lerp_float3_int_splat ( float3 p0, int p1 ) {
+ return lerp ( p0, p0, p1 );
+}
diff --git a/clang/test/SemaHLSL/BuiltIns/dot-errors.hlsl b/clang/test/SemaHLSL/BuiltIns/dot-errors.hlsl
index 54d093aa7ce3a4..5dbb52a80c6bd0 100644
--- a/clang/test/SemaHLSL/BuiltIns/dot-errors.hlsl
+++ b/clang/test/SemaHLSL/BuiltIns/dot-errors.hlsl
@@ -22,7 +22,7 @@ float test_dot_vector_size_mismatch ( float3 p0, float2 p1 ) {
float test_dot_builtin_vector_size_mismatch ( float3 p0, float2 p1 ) {
return __builtin_hlsl_dot ( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
float test_dot_scalar_mismatch ( float p0, int p1 ) {
@@ -38,69 +38,69 @@ float test_dot_element_type_mismatch ( int2 p0, float2 p1 ) {
//NOTE: for all the *_promotion we are intentionally not handling type promotion in builtins
float test_builtin_dot_vec_int_to_float_promotion ( int2 p0, float2 p1 ) {
return __builtin_hlsl_dot ( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
int64_t test_builtin_dot_vec_int_to_int64_promotion( int64_t2 p0, int2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
float test_builtin_dot_vec_half_to_float_promotion( float2 p0, half2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
#ifdef __HLSL_ENABLE_16_BIT
float test_builtin_dot_vec_int16_to_float_promotion( float2 p0, int16_t2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
half test_builtin_dot_vec_int16_to_half_promotion( half2 p0, int16_t2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
int test_builtin_dot_vec_int16_to_int_promotion( int2 p0, int16_t2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
int64_t test_builtin_dot_vec_int16_to_int64_promotion( int64_t2 p0, int16_t2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
#endif
float test_builtin_dot_float2_splat ( float p0, float2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must be vectors}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must be vectors}}
}
float test_builtin_dot_float3_splat ( float p0, float3 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must be vectors}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must be vectors}}
}
float test_builtin_dot_float4_splat ( float p0, float4 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must be vectors}}
+ // expected-error...
[truncated]
|
|
@llvm/pr-subscribers-backend-x86 Author: Farzon Lotfi (farzonl) ChangesThis is the start of implementing the lerp intrinsic https://learn.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl-lerp Builtins.td - defines the builtin this change implements the first half of #70102 Patch is 25.56 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/83077.diff 10 Files Affected:
diff --git a/clang/include/clang/Basic/Builtins.td b/clang/include/clang/Basic/Builtins.td
index e3432f7925ba14..755ef3631b2856 100644
--- a/clang/include/clang/Basic/Builtins.td
+++ b/clang/include/clang/Basic/Builtins.td
@@ -4530,6 +4530,12 @@ def HLSLDotProduct : LangBuiltin<"HLSL_LANG"> {
let Prototype = "void(...)";
}
+def HLSLLerp : LangBuiltin<"HLSL_LANG"> {
+ let Spellings = ["__builtin_hlsl_lerp"];
+ let Attributes = [NoThrow, Const];
+ let Prototype = "void(...)";
+}
+
// Builtins for XRay.
def XRayCustomEvent : Builtin {
let Spellings = ["__xray_customevent"];
diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index 57784a4ba2e388..63115d4bd49091 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -10266,6 +10266,11 @@ def err_block_on_vm : Error<
def err_sizeless_nonlocal : Error<
"non-local variable with sizeless type %0">;
+def err_vec_builtin_non_vector_all : Error<
+ "all arguments to %0 must be vectors">;
+def err_vec_builtin_incompatible_vector_all : Error<
+ "all arguments to %0 must have vectors of the same type">;
+
def err_vec_builtin_non_vector : Error<
"first two arguments to %0 must be vectors">;
def err_vec_builtin_incompatible_vector : Error<
diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 54d7451a9d6221..17b7af924e5ff0 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -18006,6 +18006,43 @@ Value *CodeGenFunction::EmitHLSLBuiltinExpr(unsigned BuiltinID,
/*ReturnType*/ T0->getScalarType(), Intrinsic::dx_dot,
ArrayRef<Value *>{Op0, Op1}, nullptr, "dx.dot");
} break;
+ case Builtin::BI__builtin_hlsl_lerp: {
+ Value *X = EmitScalarExpr(E->getArg(0));
+ Value *Y = EmitScalarExpr(E->getArg(1));
+ Value *S = EmitScalarExpr(E->getArg(2));
+ llvm::Type *Xty = X->getType();
+ llvm::Type *Yty = Y->getType();
+ llvm::Type *Sty = S->getType();
+ if (!Xty->isVectorTy() && !Yty->isVectorTy() && !Sty->isVectorTy()) {
+ if (Xty->isFloatingPointTy()) {
+ auto V = Builder.CreateFSub(Y, X);
+ V = Builder.CreateFMul(S, V);
+ return Builder.CreateFAdd(X, V, "dx.lerp");
+ }
+ llvm_unreachable("Scalar Lerp is only supported on floats.");
+ }
+ // A VectorSplat should have happened
+ assert(Xty->isVectorTy() && Yty->isVectorTy() && Sty->isVectorTy() &&
+ "Lerp of vector and scalar is not supported.");
+
+ [[maybe_unused]] auto *XVecTy =
+ E->getArg(0)->getType()->getAs<VectorType>();
+ [[maybe_unused]] auto *YVecTy =
+ E->getArg(1)->getType()->getAs<VectorType>();
+ [[maybe_unused]] auto *SVecTy =
+ E->getArg(2)->getType()->getAs<VectorType>();
+ // A HLSLVectorTruncation should have happend
+ assert(XVecTy->getNumElements() == YVecTy->getNumElements() &&
+ XVecTy->getNumElements() == SVecTy->getNumElements() &&
+ "Lerp requires vectors to be of the same size.");
+ assert(XVecTy->getElementType()->isRealFloatingType() &&
+ XVecTy->getElementType() == YVecTy->getElementType() &&
+ XVecTy->getElementType() == SVecTy->getElementType() &&
+ "Lerp requires float vectors to be of the same type.");
+ return Builder.CreateIntrinsic(
+ /*ReturnType*/ Xty, Intrinsic::dx_lerp, ArrayRef<Value *>{X, Y, S},
+ nullptr, "dx.lerp");
+ }
}
return nullptr;
}
diff --git a/clang/lib/Headers/hlsl/hlsl_intrinsics.h b/clang/lib/Headers/hlsl/hlsl_intrinsics.h
index 08e5d981a4a4ca..1314bdefa37e7b 100644
--- a/clang/lib/Headers/hlsl/hlsl_intrinsics.h
+++ b/clang/lib/Headers/hlsl/hlsl_intrinsics.h
@@ -317,6 +317,42 @@ double3 floor(double3);
_HLSL_BUILTIN_ALIAS(__builtin_elementwise_floor)
double4 floor(double4);
+//===----------------------------------------------------------------------===//
+// lerp builtins
+//===----------------------------------------------------------------------===//
+
+/// \fn T lerp(T x, T y, T s)
+/// \brief Returns the linear interpolation of x to y by s.
+/// \param x [in] The first-floating point value.
+/// \param y [in] The second-floating point value.
+/// \param s [in] A value that linearly interpolates between the x parameter and
+/// the y parameter.
+///
+/// Linear interpolation is based on the following formula: x*(1-s) + y*s which
+/// can equivalently be written as x + s(y-x).
+
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+half lerp(half, half, half);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+half2 lerp(half2, half2, half2);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+half3 lerp(half3, half3, half3);
+_HLSL_16BIT_AVAILABILITY(shadermodel, 6.2)
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+half4 lerp(half4, half4, half4);
+
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+float lerp(float, float, float);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+float2 lerp(float2, float2, float2);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+float3 lerp(float3, float3, float3);
+_HLSL_BUILTIN_ALIAS(__builtin_hlsl_lerp)
+float4 lerp(float4, float4, float4);
+
//===----------------------------------------------------------------------===//
// log builtins
//===----------------------------------------------------------------------===//
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index 984088e345c806..0c7776b9d00e84 100644
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -5169,49 +5169,63 @@ bool Sema::CheckPPCMMAType(QualType Type, SourceLocation TypeLoc) {
bool CheckVectorElementCallArgs(Sema *S, CallExpr *TheCall) {
assert(TheCall->getNumArgs() > 1);
ExprResult A = TheCall->getArg(0);
- ExprResult B = TheCall->getArg(1);
+
QualType ArgTyA = A.get()->getType();
- QualType ArgTyB = B.get()->getType();
+
auto *VecTyA = ArgTyA->getAs<VectorType>();
- auto *VecTyB = ArgTyB->getAs<VectorType>();
SourceLocation BuiltinLoc = TheCall->getBeginLoc();
- if (VecTyA == nullptr && VecTyB == nullptr)
- return false;
- if (VecTyA && VecTyB) {
- bool retValue = false;
- if (VecTyA->getElementType() != VecTyB->getElementType()) {
- // Note: type promotion is intended to be handeled via the intrinsics
- // and not the builtin itself.
- S->Diag(TheCall->getBeginLoc(), diag::err_vec_builtin_incompatible_vector)
- << TheCall->getDirectCallee()
- << SourceRange(A.get()->getBeginLoc(), B.get()->getEndLoc());
- retValue = true;
- }
- if (VecTyA->getNumElements() != VecTyB->getNumElements()) {
- // if we get here a HLSLVectorTruncation is needed.
- S->Diag(BuiltinLoc, diag::err_vec_builtin_incompatible_vector)
- << TheCall->getDirectCallee()
- << SourceRange(TheCall->getArg(0)->getBeginLoc(),
- TheCall->getArg(1)->getEndLoc());
- retValue = true;
- }
+ for (unsigned i = 1; i < TheCall->getNumArgs(); ++i) {
+ ExprResult B = TheCall->getArg(i);
+ QualType ArgTyB = B.get()->getType();
+ auto *VecTyB = ArgTyB->getAs<VectorType>();
+ if (VecTyA == nullptr && VecTyB == nullptr)
+ return false;
- if (retValue)
- TheCall->setType(VecTyA->getElementType());
+ if (VecTyA && VecTyB) {
+ bool retValue = false;
+ if (VecTyA->getElementType() != VecTyB->getElementType()) {
+ // Note: type promotion is intended to be handeled via the intrinsics
+ // and not the builtin itself.
+ S->Diag(TheCall->getBeginLoc(),
+ diag::err_vec_builtin_incompatible_vector_all)
+ << TheCall->getDirectCallee()
+ << SourceRange(A.get()->getBeginLoc(), B.get()->getEndLoc());
+ retValue = true;
+ }
+ if (VecTyA->getNumElements() != VecTyB->getNumElements()) {
+ // if we get here a HLSLVectorTruncation is needed.
+ S->Diag(BuiltinLoc, diag::err_vec_builtin_incompatible_vector_all)
+ << TheCall->getDirectCallee()
+ << SourceRange(TheCall->getArg(0)->getBeginLoc(),
+ TheCall->getArg(1)->getEndLoc());
+ retValue = true;
+ }
+
+ if (!retValue)
+ TheCall->setType(VecTyA->getElementType());
- return retValue;
+ return retValue;
+ }
}
// Note: if we get here one of the args is a scalar which
// requires a VectorSplat on Arg0 or Arg1
- S->Diag(BuiltinLoc, diag::err_vec_builtin_non_vector)
+ S->Diag(BuiltinLoc, diag::err_vec_builtin_non_vector_all)
<< TheCall->getDirectCallee()
<< SourceRange(TheCall->getArg(0)->getBeginLoc(),
TheCall->getArg(1)->getEndLoc());
return true;
}
+bool checkAllArgsAreFloatRepresnations(CallExpr *TheCall) {
+ for (unsigned i = 0; i < TheCall->getNumArgs(); ++i) {
+ if (!TheCall->getArg(0)->getType()->hasFloatingRepresentation())
+ return true;
+ }
+ return false;
+}
+
// Note: returning true in this case results in CheckBuiltinFunctionCall
// returning an ExprError
bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
@@ -5225,6 +5239,17 @@ bool Sema::CheckHLSLBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
return true;
break;
}
+ case Builtin::BI__builtin_hlsl_lerp: {
+ if (checkArgCount(*this, TheCall, 3))
+ return true;
+ if (CheckVectorElementCallArgs(this, TheCall))
+ return true;
+ if (SemaBuiltinElementwiseTernaryMath(TheCall))
+ return true;
+ if (checkAllArgsAreFloatRepresnations(TheCall))
+ return true;
+ break;
+ }
}
return false;
}
diff --git a/clang/test/CodeGenHLSL/builtins/lerp-builtin.hlsl b/clang/test/CodeGenHLSL/builtins/lerp-builtin.hlsl
new file mode 100644
index 00000000000000..1f16dec68212e4
--- /dev/null
+++ b/clang/test/CodeGenHLSL/builtins/lerp-builtin.hlsl
@@ -0,0 +1,37 @@
+// 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_lerp_half_scalar
+// CHECK: %3 = fsub double %conv1, %conv
+// CHECK: %4 = fmul double %conv2, %3
+// CHECK: %dx.lerp = fadd double %conv, %4
+// CHECK: %conv3 = fptrunc double %dx.lerp to half
+// CHECK: ret half %conv3
+half builtin_lerp_half_scalar (half p0) {
+ return __builtin_hlsl_lerp ( p0, p0, p0 );
+}
+
+// CHECK-LABEL: builtin_lerp_float_scalar
+// CHECK: %3 = fsub double %conv1, %conv
+// CHECK: %4 = fmul double %conv2, %3
+// CHECK: %dx.lerp = fadd double %conv, %4
+// CHECK: %conv3 = fptrunc double %dx.lerp to float
+// CHECK: ret float %conv3
+float builtin_lerp_float_scalar ( float p0) {
+ return __builtin_hlsl_lerp ( p0, p0, p0 );
+}
+
+// CHECK-LABEL: builtin_lerp_half_vector
+// CHECK: %dx.lerp = call <3 x half> @llvm.dx.lerp.v3f16(<3 x half> %0, <3 x half> %1, <3 x half> %2)
+// CHECK: ret <3 x half> %dx.lerp
+half3 builtin_lerp_half_vector (half3 p0) {
+ return __builtin_hlsl_lerp ( p0, p0, p0 );
+}
+
+// CHECK-LABEL: builtin_lerp_floar_vector
+// CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %2)
+// CHECK: ret <2 x float> %dx.lerp
+float2 builtin_lerp_floar_vector ( float2 p0) {
+ return __builtin_hlsl_lerp ( p0, p0, p0 );
+}
diff --git a/clang/test/CodeGenHLSL/builtins/lerp.hlsl b/clang/test/CodeGenHLSL/builtins/lerp.hlsl
new file mode 100644
index 00000000000000..1297f6b85bbd48
--- /dev/null
+++ b/clang/test/CodeGenHLSL/builtins/lerp.hlsl
@@ -0,0 +1,105 @@
+// RUN: %clang_cc1 -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 -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
+
+// NATIVE_HALF: %3 = fsub half %1, %0
+// NATIVE_HALF: %4 = fmul half %2, %3
+// NATIVE_HALF: %dx.lerp = fadd half %0, %4
+// NATIVE_HALF: ret half %dx.lerp
+// NO_HALF: %3 = fsub float %1, %0
+// NO_HALF: %4 = fmul float %2, %3
+// NO_HALF: %dx.lerp = fadd float %0, %4
+// NO_HALF: ret float %dx.lerp
+half test_lerp_half ( half p0) {
+ return lerp ( p0, p0, p0 );
+}
+
+// NATIVE_HALF: %dx.lerp = call <2 x half> @llvm.dx.lerp.v2f16(<2 x half> %0, <2 x half> %1, <2 x half> %2)
+// NATIVE_HALF: ret <2 x half> %dx.lerp
+// NO_HALF: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %2)
+// NO_HALF: ret <2 x float> %dx.lerp
+half2 test_lerp_half2 ( half2 p0, half2 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// NATIVE_HALF: %dx.lerp = call <3 x half> @llvm.dx.lerp.v3f16(<3 x half> %0, <3 x half> %1, <3 x half> %2)
+// NATIVE_HALF: ret <3 x half> %dx.lerp
+// NO_HALF: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %0, <3 x float> %1, <3 x float> %2)
+// NO_HALF: ret <3 x float> %dx.lerp
+half3 test_lerp_half3 ( half3 p0, half3 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// NATIVE_HALF: %dx.lerp = call <4 x half> @llvm.dx.lerp.v4f16(<4 x half> %0, <4 x half> %1, <4 x half> %2)
+// NATIVE_HALF: ret <4 x half> %dx.lerp
+// NO_HALF: %dx.lerp = call <4 x float> @llvm.dx.lerp.v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2)
+// NO_HALF: ret <4 x float> %dx.lerp
+half4 test_lerp_half4 ( half4 p0, half4 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %3 = fsub float %1, %0
+// CHECK: %4 = fmul float %2, %3
+// CHECK: %dx.lerp = fadd float %0, %4
+// CHECK: ret float %dx.lerp
+float test_lerp_float ( float p0, float p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %2)
+// CHECK: ret <2 x float> %dx.lerp
+float2 test_lerp_float2 ( float2 p0, float2 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %0, <3 x float> %1, <3 x float> %2)
+// CHECK: ret <3 x float> %dx.lerp
+float3 test_lerp_float3 ( float3 p0, float3 p1 ) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %dx.lerp = call <4 x float> @llvm.dx.lerp.v4f32(<4 x float> %0, <4 x float> %1, <4 x float> %2)
+// CHECK: ret <4 x float> %dx.lerp
+float4 test_lerp_float4 ( float4 p0, float4 p1) {
+ return lerp ( p0, p0, p0 );
+}
+
+// CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %splat.splat, <2 x float> %1, <2 x float> %2)
+// CHECK: ret <2 x float> %dx.lerp
+float2 test_lerp_float2_splat ( float p0, float2 p1 ) {
+ return lerp( p0, p1, p1 );
+}
+
+// CHECK: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %splat.splat, <3 x float> %1, <3 x float> %2)
+// CHECK: ret <3 x float> %dx.lerp
+float3 test_lerp_float3_splat ( float p0, float3 p1 ) {
+ return lerp( p0, p1, p1 );
+}
+
+// CHECK: %dx.lerp = call <4 x float> @llvm.dx.lerp.v4f32(<4 x float> %splat.splat, <4 x float> %1, <4 x float> %2)
+// CHECK: ret <4 x float> %dx.lerp
+float4 test_lerp_float4_splat ( float p0, float4 p1 ) {
+ return lerp( p0, p1, p1 );
+}
+
+// CHECK: %conv = sitofp i32 %2 to float
+// CHECK: %splat.splatinsert = insertelement <2 x float> poison, float %conv, i64 0
+// CHECK: %splat.splat = shufflevector <2 x float> %splat.splatinsert, <2 x float> poison, <2 x i32> zeroinitializer
+// CHECK: %dx.lerp = call <2 x float> @llvm.dx.lerp.v2f32(<2 x float> %0, <2 x float> %1, <2 x float> %splat.splat)
+// CHECK: ret <2 x float> %dx.lerp
+float2 test_lerp_float2_int_splat ( float2 p0, int p1 ) {
+ return lerp ( p0, p0, p1 );
+}
+
+// CHECK: %conv = sitofp i32 %2 to float
+// CHECK: %splat.splatinsert = insertelement <3 x float> poison, float %conv, i64 0
+// CHECK: %splat.splat = shufflevector <3 x float> %splat.splatinsert, <3 x float> poison, <3 x i32> zeroinitializer
+// CHECK: %dx.lerp = call <3 x float> @llvm.dx.lerp.v3f32(<3 x float> %0, <3 x float> %1, <3 x float> %splat.splat)
+// CHECK: ret <3 x float> %dx.lerp
+float3 test_lerp_float3_int_splat ( float3 p0, int p1 ) {
+ return lerp ( p0, p0, p1 );
+}
diff --git a/clang/test/SemaHLSL/BuiltIns/dot-errors.hlsl b/clang/test/SemaHLSL/BuiltIns/dot-errors.hlsl
index 54d093aa7ce3a4..5dbb52a80c6bd0 100644
--- a/clang/test/SemaHLSL/BuiltIns/dot-errors.hlsl
+++ b/clang/test/SemaHLSL/BuiltIns/dot-errors.hlsl
@@ -22,7 +22,7 @@ float test_dot_vector_size_mismatch ( float3 p0, float2 p1 ) {
float test_dot_builtin_vector_size_mismatch ( float3 p0, float2 p1 ) {
return __builtin_hlsl_dot ( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
float test_dot_scalar_mismatch ( float p0, int p1 ) {
@@ -38,69 +38,69 @@ float test_dot_element_type_mismatch ( int2 p0, float2 p1 ) {
//NOTE: for all the *_promotion we are intentionally not handling type promotion in builtins
float test_builtin_dot_vec_int_to_float_promotion ( int2 p0, float2 p1 ) {
return __builtin_hlsl_dot ( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
int64_t test_builtin_dot_vec_int_to_int64_promotion( int64_t2 p0, int2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
float test_builtin_dot_vec_half_to_float_promotion( float2 p0, half2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
#ifdef __HLSL_ENABLE_16_BIT
float test_builtin_dot_vec_int16_to_float_promotion( float2 p0, int16_t2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
half test_builtin_dot_vec_int16_to_half_promotion( half2 p0, int16_t2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
int test_builtin_dot_vec_int16_to_int_promotion( int2 p0, int16_t2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
int64_t test_builtin_dot_vec_int16_to_int64_promotion( int64_t2 p0, int16_t2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must have the same type}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must have vectors of the same type}}
}
#endif
float test_builtin_dot_float2_splat ( float p0, float2 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must be vectors}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must be vectors}}
}
float test_builtin_dot_float3_splat ( float p0, float3 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must be vectors}}
+ // expected-error@-1 {{all arguments to '__builtin_hlsl_dot' must be vectors}}
}
float test_builtin_dot_float4_splat ( float p0, float4 p1 ) {
return __builtin_hlsl_dot( p0, p1 );
- // expected-error@-1 {{first two arguments to '__builtin_hlsl_dot' must be vectors}}
+ // expected-error...
[truncated]
|
bogner
approved these changes
Feb 28, 2024
| def err_vec_builtin_non_vector_all : Error< | ||
| "all arguments to %0 must be vectors">; | ||
| def err_vec_builtin_incompatible_vector_all : Error< | ||
| "all arguments to %0 must have vectors of the same type">; |
There was a problem hiding this comment.
It wouldn't make sense to change it as part of this commit, but with this in place "err_vec_builtin_non_vector" has a bit of an ambiguous name now. Might be worth following up at some point
There was a problem hiding this comment.
I did it this way because the amount of test cases I would need to update to not be tied to two args was very large.
There was a problem hiding this comment.
I disagree here. I think we should collapse these down. You can use a %select{...} block in the error format string to support the old message and the new message from the same diagnostic entry.
These diagnostic strings end up becoming some pretty large constant data tables in the final clang executable so we've generally tried to reuse errors where it makes sense.
farzonl
force-pushed
the
hlsl-lerp-intrinsic
branch
from
d4384c5
to
56240f0
Compare
python3kgae
approved these changes
Feb 29, 2024
|
@python3kgae I wanted to hold off on merging this one because I was investigating why 56240f0 wiped out my changes on d4384c5. In any case I put fixes up on the frac intrinsic pr. There should be a fix landing for any issues caused by this change in: |
farzonl
added a commit
that referenced
this pull request
Mar 2, 2024
removing the additions of `err_vec_builtin_non_vector_all` & `err_vec_builtin_incompatible_vector_all` caused by #83077. Instead adding a select option to `err_vec_builtin_non_vector` & `err_vec_builtin_incompatible_vector` to account for uses where there are more than two arguments.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is the start of implementing the lerp intrinsic https://learn.microsoft.com/en-us/windows/win32/direct3dhlsl/dx-graphics-hlsl-lerp Builtins.td - defines the builtin
hlsl_intrinsics.h - defines the lerp api
DiagnosticSemaKinds.td - needed a new error to be inclusive for more than two operands.
CGBuiltin.cpp - add the lerp intrinsic lowering
SemaChecking.cpp - type checks for lerp builtin
IntrinsicsDirectX.td - define the lerp intrinsic
this change implements the first half of #70102