SPIRV-Cross/reference/shaders-msl/frag/subgroup-globals-extract.msl22.frag
Bill Hollings ec054dad7f MSL: Support synthetic functions in function constants.
Emit synthetic functions before function constants.
Support use of spvQuantizeToF16() in function constants for numerical
behavior consistency with the op code.
Ensure subnormal results from OpQuantizeToF16 are flushed to zero per SPIR-V spec.

Adjust SPIRV-Cross unit test reference shaders to accommodate these changes.
Any MSL reference shader that inclues a synthetic function is affected,
since the location it is emitted has changed.
2021-09-28 19:10:16 -04:00

87 lines
3.3 KiB
GLSL

#pragma clang diagnostic ignored "-Wmissing-prototypes"
#include <metal_stdlib>
#include <simd/simd.h>
using namespace metal;
inline uint spvSubgroupBallotFindLSB(uint4 ballot, uint gl_SubgroupSize)
{
uint4 mask = uint4(extract_bits(0xFFFFFFFF, 0, min(gl_SubgroupSize, 32u)), extract_bits(0xFFFFFFFF, 0, (uint)max((int)gl_SubgroupSize - 32, 0)), uint2(0));
ballot &= mask;
return select(ctz(ballot.x), select(32 + ctz(ballot.y), select(64 + ctz(ballot.z), select(96 + ctz(ballot.w), uint(-1), ballot.w == 0), ballot.z == 0), ballot.y == 0), ballot.x == 0);
}
inline uint spvSubgroupBallotFindMSB(uint4 ballot, uint gl_SubgroupSize)
{
uint4 mask = uint4(extract_bits(0xFFFFFFFF, 0, min(gl_SubgroupSize, 32u)), extract_bits(0xFFFFFFFF, 0, (uint)max((int)gl_SubgroupSize - 32, 0)), uint2(0));
ballot &= mask;
return select(128 - (clz(ballot.w) + 1), select(96 - (clz(ballot.z) + 1), select(64 - (clz(ballot.y) + 1), select(32 - (clz(ballot.x) + 1), uint(-1), ballot.x == 0), ballot.y == 0), ballot.z == 0), ballot.w == 0);
}
inline uint spvPopCount4(uint4 ballot)
{
return popcount(ballot.x) + popcount(ballot.y) + popcount(ballot.z) + popcount(ballot.w);
}
inline uint spvSubgroupBallotBitCount(uint4 ballot, uint gl_SubgroupSize)
{
uint4 mask = uint4(extract_bits(0xFFFFFFFF, 0, min(gl_SubgroupSize, 32u)), extract_bits(0xFFFFFFFF, 0, (uint)max((int)gl_SubgroupSize - 32, 0)), uint2(0));
return spvPopCount4(ballot & mask);
}
inline uint spvSubgroupBallotInclusiveBitCount(uint4 ballot, uint gl_SubgroupInvocationID)
{
uint4 mask = uint4(extract_bits(0xFFFFFFFF, 0, min(gl_SubgroupInvocationID + 1, 32u)), extract_bits(0xFFFFFFFF, 0, (uint)max((int)gl_SubgroupInvocationID + 1 - 32, 0)), uint2(0));
return spvPopCount4(ballot & mask);
}
inline uint spvSubgroupBallotExclusiveBitCount(uint4 ballot, uint gl_SubgroupInvocationID)
{
uint4 mask = uint4(extract_bits(0xFFFFFFFF, 0, min(gl_SubgroupInvocationID, 32u)), extract_bits(0xFFFFFFFF, 0, (uint)max((int)gl_SubgroupInvocationID - 32, 0)), uint2(0));
return spvPopCount4(ballot & mask);
}
struct main0_out
{
uint2 FragColor [[color(0)]];
};
static inline __attribute__((always_inline))
uint sub1(thread uint& gl_SubgroupSize)
{
return spvSubgroupBallotFindLSB(uint4(1u, 2u, 3u, 4u), gl_SubgroupSize);
}
static inline __attribute__((always_inline))
uint sub2(thread uint& gl_SubgroupSize)
{
return spvSubgroupBallotFindMSB(uint4(1u, 2u, 3u, 4u), gl_SubgroupSize);
}
static inline __attribute__((always_inline))
uint sub3(thread uint& gl_SubgroupSize)
{
return spvSubgroupBallotBitCount(uint4(1u, 2u, 3u, 4u), gl_SubgroupSize);
}
static inline __attribute__((always_inline))
uint sub4(thread uint& gl_SubgroupInvocationID)
{
return spvSubgroupBallotInclusiveBitCount(uint4(1u, 2u, 3u, 4u), gl_SubgroupInvocationID);
}
static inline __attribute__((always_inline))
uint sub5(thread uint& gl_SubgroupInvocationID)
{
return spvSubgroupBallotExclusiveBitCount(uint4(1u, 2u, 3u, 4u), gl_SubgroupInvocationID);
}
fragment main0_out main0(uint gl_SubgroupInvocationID [[thread_index_in_simdgroup]], uint gl_SubgroupSize [[threads_per_simdgroup]])
{
main0_out out = {};
out.FragColor.x = (((sub1(gl_SubgroupSize) + sub2(gl_SubgroupSize)) + sub3(gl_SubgroupSize)) + sub4(gl_SubgroupInvocationID)) + sub5(gl_SubgroupInvocationID);
return out;
}