Implement the InterlockedCompareStoreFloatBitwise HLSL Function

[farzonl]

• Implement InterlockedCompareStoreFloatBitwise clang builtin,
• Link InterlockedCompareStoreFloatBitwise clang builtin with hlsl_intrinsics.h
• Add sema checks for InterlockedCompareStoreFloatBitwise to CheckHLSLBuiltinFunctionCall in SemaChecking.cpp
• Add codegen for InterlockedCompareStoreFloatBitwise to EmitHLSLBuiltinExpr in CGBuiltin.cpp
• Add codegen tests to clang/test/CodeGenHLSL/builtins/InterlockedCompareStoreFloatBitwise.hlsl
• Add sema tests to clang/test/SemaHLSL/BuiltIns/InterlockedCompareStoreFloatBitwise-errors.hlsl
• Create the int_dx_InterlockedCompareStoreFloatBitwise intrinsic in IntrinsicsDirectX.td
• Create the DXILOpMapping of int_dx_InterlockedCompareStoreFloatBitwise to 79 in DXIL.td
• Create the InterlockedCompareStoreFloatBitwise.ll and InterlockedCompareStoreFloatBitwise_errors.ll tests in llvm/test/CodeGen/DirectX/

DirectX

DXIL Opcode	DXIL OpName	Shader Model	Shader Stages
79	AtomicCompareExchange	6.0	()

SPIR-V

There is no support for InterlockedCompareStoreFloatBitwise when targeting SPIR-V.

Test Case(s)

Example 1

//dxc InterlockedCompareStoreFloatBitwise_test.hlsl -T lib_6_8 -enable-16bit-types -O0

RWStructuredBuffer<float> buffer : register(u0);
[numthreads(1, 1, 1)]
export void fn(uint3 dispatchThreadID : SV_DispatchThreadID, float p1, float p2) {
int index = dispatchThreadID.x;
    return InterlockedCompareStoreFloatBitwise(buffer[index], p1, p2);
}

HLSL:

Syntax

void InterlockedCompareStoreFloatBitwise(float32_only result, float compare, float value);

void InterlockedCompareStoreFloatBitwise(uint byteOffest, float compare, float value);

Type Description

Name	Template Type	Component Type	Size
ret	void	void	0
result	scalar	float	1
compare	scalar	float	1
value	scalar	float	1

Type Description

Name	Template Type	Component Type	Size
ret	void	void	0
byteOffest	scalar	uint	1
compare	scalar	float	1
value	scalar	float	1

Minimum Shader Model

This function is supported in the following shader models.

Shader Model	Supported
Shader Model 6.3 and higher shader models	yes

Shader Stages

See also

• Intrinsic Functions (DirectX HLSL)
• See InterlockedCompareStoreFloatBitwise