AuroraMiddleware/SPIRV-Cross
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge branch 'master' of https://github.com/gpx1000/SPIRV-Cross

Browse Source
This commit is contained in:
Hans-Kristian Arntzen 2022-08-09 11:54:53 +02:00
commit 1bc05ef735
19 changed files with 11126 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1353
reference/opt/shaders-msl/comp/shared-matrix-array-of-array.comp Normal file
View File

File diff suppressed because it is too large Load Diff

1017
reference/opt/shaders-msl/comp/shared-matrix-cast.comp Normal file
View File

File diff suppressed because it is too large Load Diff

1369
reference/opt/shaders-msl/comp/shared-matrix-nested-struct-array.comp Normal file
View File

File diff suppressed because it is too large Load Diff

1443
reference/opt/shaders-msl/comp/shared-matrix-nested-struct.comp Normal file
View File

File diff suppressed because it is too large Load Diff

63
reference/opt/shaders-msl/comp/shared-struct-bool-cast.comp Normal file
View File

@ -0,0 +1,63 @@
#include <metal_stdlib>
#include <simd/simd.h>
using namespace metal;
struct S1
{
int3 a;
uint2 b;
short4 c;
uint d;
};
struct block
{
uint passed;
};
constant uint3 gl_WorkGroupSize [[maybe_unused]] = uint3(1u);
kernel void main0(device block& _132 [[buffer(0)]])
{
threadgroup S1 s1;
s1.a = int3(6, 8, 8);
s1.b = uint2(4u);
s1.c = short4(bool4(false, false, false, true));
s1.d = 6u;
threadgroup_barrier(mem_flags::mem_threadgroup);
threadgroup_barrier(mem_flags::mem_device | mem_flags::mem_threadgroup | mem_flags::mem_texture);
bool _144 = all(int3(6, 8, 8) == s1.a);
bool _108;
if (_144)
{
_108 = all(uint2(4u) == s1.b);
}
else
{
_108 = _144;
}
bool _117;
if (_108)
{
_117 = all(bool4(false, false, false, true) == bool4(s1.c));
}
else
{
_117 = _108;
}
bool _126;
if (_117)
{
_126 = 6u == s1.d;
}
else
{
_126 = _117;
}
if (_126)
{
_132.passed++;
}
}

1286
reference/shaders-msl/comp/shared-matrix-array-of-array.comp Normal file
View File

File diff suppressed because it is too large Load Diff

1065
reference/shaders-msl/comp/shared-matrix-cast.comp Normal file
View File

File diff suppressed because it is too large Load Diff

1316
reference/shaders-msl/comp/shared-matrix-nested-struct-array.comp Normal file
View File

File diff suppressed because it is too large Load Diff

1473
reference/shaders-msl/comp/shared-matrix-nested-struct.comp Normal file
View File

File diff suppressed because it is too large Load Diff

110
reference/shaders-msl/comp/shared-struct-bool-cast.comp Normal file
View File

@ -0,0 +1,110 @@
#pragma clang diagnostic ignored "-Wmissing-prototypes"
#include <metal_stdlib>
#include <simd/simd.h>
using namespace metal;
struct S1
{
int3 a;
uint2 b;
short4 c;
uint d;
};
struct block
{
uint passed;
};
constant uint3 gl_WorkGroupSize [[maybe_unused]] = uint3(1u);
static inline __attribute__((always_inline))
bool compare_ivec3(thread const int3& a, thread const int3& b)
{
return all(a == b);
}
static inline __attribute__((always_inline))
bool compare_uvec2(thread const uint2& a, thread const uint2& b)
{
return all(a == b);
}
static inline __attribute__((always_inline))
bool compare_bvec4(thread const bool4& a, thread const bool4& b)
{
return all(a == b);
}
static inline __attribute__((always_inline))
bool compare_uint(thread const uint& a, thread const uint& b)
{
return a == b;
}
kernel void main0(device block& _132 [[buffer(0)]])
{
threadgroup S1 s1;
s1.a = int3(6, 8, 8);
s1.b = uint2(4u);
s1.c = short4(bool4(false, false, false, true));
s1.d = 6u;
threadgroup_barrier(mem_flags::mem_threadgroup);
threadgroup_barrier(mem_flags::mem_device | mem_flags::mem_threadgroup | mem_flags::mem_texture);
bool allOk = true;
bool _99;
if (allOk)
{
int3 param = int3(6, 8, 8);
int3 param_1 = s1.a;
_99 = compare_ivec3(param, param_1);
}
else
{
_99 = allOk;
}
allOk = _99;
bool _108;
if (allOk)
{
uint2 param_2 = uint2(4u);
uint2 param_3 = s1.b;
_108 = compare_uvec2(param_2, param_3);
}
else
{
_108 = allOk;
}
allOk = _108;
bool _117;
if (allOk)
{
bool4 param_4 = bool4(false, false, false, true);
bool4 param_5 = bool4(s1.c);
_117 = compare_bvec4(param_4, param_5);
}
else
{
_117 = allOk;
}
allOk = _117;
bool _126;
if (allOk)
{
uint param_6 = 6u;
uint param_7 = s1.d;
_126 = compare_uint(param_6, param_7);
}
else
{
_126 = allOk;
}
allOk = _126;
if (allOk)
{
_132.passed++;
}
}

65
shaders-msl/comp/shared-matrix-array-of-array.comp Normal file
View File

@ -0,0 +1,65 @@
#version 450
layout(local_size_x = 1) in;
layout(std140, binding = 0) buffer block { highp uint passed; };
struct S1 {
mediump mat4x3 a[2];
lowp float b;
lowp vec2 c[3];
};
struct S2 {
highp ivec4 a;
bool b[3][1][3];
};
bool compare_float (highp float a, highp float b) { return abs(a - b) < 0.05; }
bool compare_vec2 (highp vec2 a, highp vec2 b) { return compare_float(a.x, b.x)&&compare_float(a.y, b.y); }
bool compare_vec3 (highp vec3 a, highp vec3 b) { return compare_float(a.x, b.x)&&compare_float(a.y, b.y)&&compare_float(a.z, b.z); }
bool compare_mat4x3 (highp mat4x3 a, highp mat4x3 b){ return compare_vec3(a[0], b[0])&&compare_vec3(a[1], b[1])&&compare_vec3(a[2], b[2])&&compare_vec3(a[3], b[3]); }
bool compare_ivec4 (highp ivec4 a, highp ivec4 b) { return a == b; }
bool compare_bool (bool a, bool b) { return a == b; }
shared S1 s1;
shared S2 s2;
void main (void) {
s1.a[0] = mat4x3(0.0, 2.0, -8.0, 6.0, 7.0, 5.0, -6.0, 1.0, 9.0, -4.0, -3.0, 4.0);
s1.a[1] = mat4x3(4.0, 9.0, -9.0, -8.0, -9.0, 8.0, 0.0, 4.0, -4.0, 7.0, 2.0, -1.0);
s1.b = 7.0;
s1.c[0] = vec2(-5.0, -4.0);
s1.c[1] = vec2(3.0, -5.0);
s1.c[2] = vec2(-3.0, -1.0);
s2.a = ivec4(1, 0, -3, 1);
s2.b[0][0][0] = true;
s2.b[0][0][1] = false;
s2.b[0][0][2] = false;
s2.b[1][0][0] = true;
s2.b[1][0][1] = false;
s2.b[1][0][2] = true;
s2.b[2][0][0] = false;
s2.b[2][0][1] = true;
s2.b[2][0][2] = true;
barrier();
memoryBarrier();
bool allOk = true;
allOk = allOk && compare_mat4x3(mat4x3(0.0, 2.0, -8.0, 6.0, 7.0, 5.0, -6.0, 1.0, 9.0, -4.0, -3.0, 4.0), s1.a[0]);
allOk = allOk && compare_mat4x3(mat4x3(4.0, 9.0, -9.0, -8.0, -9.0, 8.0, 0.0, 4.0, -4.0, 7.0, 2.0, -1.0), s1.a[1]);
allOk = allOk && compare_float(7.0, s1.b);
allOk = allOk && compare_vec2(vec2(-5.0, -4.0), s1.c[0]);
allOk = allOk && compare_vec2(vec2(3.0, -5.0), s1.c[1]);
allOk = allOk && compare_vec2(vec2(-3.0, -1.0), s1.c[2]);
allOk = allOk && compare_ivec4(ivec4(1, 0, -3, 1), s2.a);
allOk = allOk && compare_bool(true, s2.b[0][0][0]);
allOk = allOk && compare_bool(false, s2.b[0][0][1]);
allOk = allOk && compare_bool(false, s2.b[0][0][2]);
allOk = allOk && compare_bool(true, s2.b[1][0][0]);
allOk = allOk && compare_bool(false, s2.b[1][0][1]);
allOk = allOk && compare_bool(true, s2.b[1][0][2]);
allOk = allOk && compare_bool(false, s2.b[2][0][0]);
allOk = allOk && compare_bool(true, s2.b[2][0][1]);
allOk = allOk && compare_bool(true, s2.b[2][0][2]);
if (allOk)
passed++;
}

33
shaders-msl/comp/shared-matrix-cast.comp Normal file
View File

@ -0,0 +1,33 @@
#version 450
layout(local_size_x = 1) in;
layout(std140, binding = 0) buffer block { highp uint passed; };
struct S1 {
mediump vec4 a;
highp mat3x2 b;
bvec4 c;
};
bool compare_float (highp float a, highp float b) { return abs(a - b) < 0.05; }
bool compare_vec2 (highp vec2 a, highp vec2 b) { return compare_float(a.x, b.x)&&compare_float(a.y, b.y); }
bool compare_vec4 (highp vec4 a, highp vec4 b) { return compare_float(a.x, b.x)&&compare_float(a.y, b.y)&&compare_float(a.z, b.z)&&compare_float(a.w, b.w); }
bool compare_mat3x2 (highp mat3x2 a, highp mat3x2 b){ return compare_vec2(a[0], b[0])&&compare_vec2(a[1], b[1])&&compare_vec2(a[2], b[2]); }
bool compare_bvec4 (bvec4 a, bvec4 b) { return a == b; }
shared S1 s1;
void main (void) {
s1.a = vec4(1.0, -5.0, -9.0, -5.0);
s1.b = mat3x2(1.0, -7.0, 1.0, 2.0, 8.0, 7.0);
s1.c = bvec4(false, true, false, false);
barrier();
memoryBarrier();
bool allOk = true;
allOk = allOk && compare_vec4(vec4(1.0, -5.0, -9.0, -5.0), s1.a);
allOk = allOk && compare_mat3x2(mat3x2(1.0, -7.0, 1.0, 2.0, 8.0, 7.0), s1.b);
allOk = allOk && compare_bvec4(bvec4(false, true, false, false), s1.c);
if (allOk)
passed++;
}

87
shaders-msl/comp/shared-matrix-nested-struct-array.comp Normal file
View File

@ -0,0 +1,87 @@
#version 450
layout(local_size_x = 1) in;
layout(std140, binding = 0) buffer block { highp uint passed; };
struct sA
{
mediump mat2x3 mA;
};
struct sB
{
mediump mat2 mA;
mediump mat3x2 mB;
highp uvec3 mC;
};
struct sC
{
sA mA;
sB mB;
};
struct sD
{
sC mA;
};
struct sE
{
lowp mat3x2 mA;
lowp mat4x3 mB;
};
struct sF
{
sE mA;
};
struct sG
{
sF mA;
};
struct sH
{
bvec3 mA[2];
};
struct S1 {
sD a;
sG b;
sH c[2];
};
bool compare_float (highp float a, highp float b) { return abs(a - b) < 0.05; }
bool compare_vec2 (highp vec2 a, highp vec2 b) { return compare_float(a.x, b.x)&&compare_float(a.y, b.y); }
bool compare_vec3 (highp vec3 a, highp vec3 b) { return compare_float(a.x, b.x)&&compare_float(a.y, b.y)&&compare_float(a.z, b.z); }
bool compare_mat2 (highp mat2 a, highp mat2 b) { return compare_vec2(a[0], b[0])&&compare_vec2(a[1], b[1]); }
bool compare_mat2x3 (highp mat2x3 a, highp mat2x3 b){ return compare_vec3(a[0], b[0])&&compare_vec3(a[1], b[1]); }
bool compare_mat3x2 (highp mat3x2 a, highp mat3x2 b){ return compare_vec2(a[0], b[0])&&compare_vec2(a[1], b[1])&&compare_vec2(a[2], b[2]); }
bool compare_mat4x3 (highp mat4x3 a, highp mat4x3 b){ return compare_vec3(a[0], b[0])&&compare_vec3(a[1], b[1])&&compare_vec3(a[2], b[2])&&compare_vec3(a[3], b[3]); }
bool compare_uvec3 (highp uvec3 a, highp uvec3 b) { return a == b; }
bool compare_bvec3 (bvec3 a, bvec3 b) { return a == b; }
shared S1 s1;
void main (void) {
s1.a.mA.mA.mA = mat2x3(6.0, 8.0, 8.0, 0.0, -4.0, -5.0);
s1.a.mA.mB.mA = mat2(9.0, -4.0, -6.0, -1.0);
s1.a.mA.mB.mB = mat3x2(-1.0, -2.0, 1.0, 6.0, 5.0, 7.0);
s1.a.mA.mB.mC = uvec3(3u, 1u, 5u);
s1.b.mA.mA.mA = mat3x2(8.0, 3.0, 0.0, 2.0, 1.0, 8.0);
s1.b.mA.mA.mB = mat4x3(0.0, 9.0, -1.0, -1.0, -7.0, 7.0, -4.0, -3.0, 1.0, -4.0, -9.0, 1.0);
s1.c[0].mA[0] = bvec3(true, false, false);
s1.c[0].mA[1] = bvec3(true, false, false);
s1.c[1].mA[0] = bvec3(false, false, false);
s1.c[1].mA[1] = bvec3(false, false, false);
barrier();
memoryBarrier();
bool allOk = true;
allOk = allOk && compare_mat2x3(mat2x3(6.0, 8.0, 8.0, 0.0, -4.0, -5.0), s1.a.mA.mA.mA);
allOk = allOk && compare_mat2(mat2(9.0, -4.0, -6.0, -1.0), s1.a.mA.mB.mA);
allOk = allOk && compare_mat3x2(mat3x2(-1.0, -2.0, 1.0, 6.0, 5.0, 7.0), s1.a.mA.mB.mB);
allOk = allOk && compare_uvec3(uvec3(3u, 1u, 5u), s1.a.mA.mB.mC);
allOk = allOk && compare_mat3x2(mat3x2(8.0, 3.0, 0.0, 2.0, 1.0, 8.0), s1.b.mA.mA.mA);
allOk = allOk && compare_mat4x3(mat4x3(0.0, 9.0, -1.0, -1.0, -7.0, 7.0, -4.0, -3.0, 1.0, -4.0, -9.0, 1.0), s1.b.mA.mA.mB);
allOk = allOk && compare_bvec3(bvec3(true, false, false), s1.c[0].mA[0]);
allOk = allOk && compare_bvec3(bvec3(true, false, false), s1.c[0].mA[1]);
allOk = allOk && compare_bvec3(bvec3(false, false, false), s1.c[1].mA[0]);
allOk = allOk && compare_bvec3(bvec3(false, false, false), s1.c[1].mA[1]);
if (allOk)
passed++;
}

141
shaders-msl/comp/shared-matrix-nested-struct.comp Normal file
View File

@ -0,0 +1,141 @@
#version 450
layout(local_size_x = 1) in;
layout(std140, binding = 0) buffer block { highp uint passed; };
struct sA
{
highp mat4 mA;
bvec3 mB;
bvec4 mC;
};
struct sB
{
bvec2 mA;
};
struct sC
{
highp float mA;
mediump uvec4 mB;
mediump float mC;
};
struct sD
{
sA mA;
sB mB;
sC mC;
};
struct sE
{
sD mA;
};
struct sF
{
lowp uvec3 mA;
bool mB;
};
struct sG
{
sF mA;
highp mat3x2 mB;
};
struct sH
{
sG mA;
mediump vec2 mB;
};
struct sI
{
mediump mat2 mA;
bvec3 mB;
bvec4 mC;
};
struct sJ
{
sI mA;
bvec3 mB;
};
struct sK
{
bvec2 mA;
sJ mB;
mediump ivec2 mC;
};
struct S1 {
lowp uint a;
mediump vec4 b;
};
struct S2 {
sE a;
highp ivec3 b;
sH c;
sK d;
};
bool compare_float (highp float a, highp float b) { return abs(a - b) < 0.05; }
bool compare_vec2 (highp vec2 a, highp vec2 b) { return compare_float(a.x, b.x)&&compare_float(a.y, b.y); }
bool compare_vec4 (highp vec4 a, highp vec4 b) { return compare_float(a.x, b.x)&&compare_float(a.y, b.y)&&compare_float(a.z, b.z)&&compare_float(a.w, b.w); }
bool compare_mat2 (highp mat2 a, highp mat2 b) { return compare_vec2(a[0], b[0])&&compare_vec2(a[1], b[1]); }
bool compare_mat3x2 (highp mat3x2 a, highp mat3x2 b){ return compare_vec2(a[0], b[0])&&compare_vec2(a[1], b[1])&&compare_vec2(a[2], b[2]); }
bool compare_mat4 (highp mat4 a, highp mat4 b) { return compare_vec4(a[0], b[0])&&compare_vec4(a[1], b[1])&&compare_vec4(a[2], b[2])&&compare_vec4(a[3], b[3]); }
bool compare_ivec2 (highp ivec2 a, highp ivec2 b) { return a == b; }
bool compare_ivec3 (highp ivec3 a, highp ivec3 b) { return a == b; }
bool compare_uint (highp uint a, highp uint b) { return a == b; }
bool compare_uvec3 (highp uvec3 a, highp uvec3 b) { return a == b; }
bool compare_uvec4 (highp uvec4 a, highp uvec4 b) { return a == b; }
bool compare_bool (bool a, bool b) { return a == b; }
bool compare_bvec2 (bvec2 a, bvec2 b) { return a == b; }
bool compare_bvec3 (bvec3 a, bvec3 b) { return a == b; }
bool compare_bvec4 (bvec4 a, bvec4 b) { return a == b; }
shared S1 s1;
shared S2 s2;
void main (void) {
s1.a = 0u;
s1.b = vec4(8.0, 8.0, 0.0, -4.0);
s2.a.mA.mA.mA = mat4(-5.0, 9.0, -4.0, -6.0, -1.0, -1.0, -2.0, 1.0, 6.0, 5.0, 7.0, -2.0, -4.0, -9.0, 8.0, 3.0);
s2.a.mA.mA.mB = bvec3(true, false, false);
s2.a.mA.mA.mC = bvec4(true, true, true, false);
s2.a.mA.mB.mA = bvec2(true, true);
s2.a.mA.mC.mA = 7.0;
s2.a.mA.mC.mB = uvec4(8u, 6u, 2u, 0u);
s2.a.mA.mC.mC = -9.0;
s2.b = ivec3(1, -4, 0);
s2.c.mA.mA.mA = uvec3(4u, 9u, 1u);
s2.c.mA.mA.mB = false;
s2.c.mA.mB = mat3x2(3.0, -5.0, -1.0, -5.0, -1.0, -9.0);
s2.c.mB = vec2(-6.0, -9.0);
s2.d.mA = bvec2(true, false);
s2.d.mB.mA.mA = mat2(-2.0, 3.0, 7.0, 2.0);
s2.d.mB.mA.mB = bvec3(false, false, false);
s2.d.mB.mA.mC = bvec4(false, false, false, true);
s2.d.mB.mB = bvec3(true, false, false);
s2.d.mC = ivec2(-9, 0);
barrier();
memoryBarrier();
bool allOk = true;
allOk = allOk && compare_uint(0u, s1.a);
allOk = allOk && compare_vec4(vec4(8.0, 8.0, 0.0, -4.0), s1.b);
allOk = allOk && compare_mat4(mat4(-5.0, 9.0, -4.0, -6.0, -1.0, -1.0, -2.0, 1.0, 6.0, 5.0, 7.0, -2.0, -4.0, -9.0, 8.0, 3.0), s2.a.mA.mA.mA);
allOk = allOk && compare_bvec3(bvec3(true, false, false), s2.a.mA.mA.mB);
allOk = allOk && compare_bvec4(bvec4(true, true, true, false), s2.a.mA.mA.mC);
allOk = allOk && compare_bvec2(bvec2(true, true), s2.a.mA.mB.mA);
allOk = allOk && compare_float(7.0, s2.a.mA.mC.mA);
allOk = allOk && compare_uvec4(uvec4(8u, 6u, 2u, 0u), s2.a.mA.mC.mB);
allOk = allOk && compare_float(-9.0, s2.a.mA.mC.mC);
allOk = allOk && compare_ivec3(ivec3(1, -4, 0), s2.b);
allOk = allOk && compare_uvec3(uvec3(4u, 9u, 1u), s2.c.mA.mA.mA);
allOk = allOk && compare_bool(false, s2.c.mA.mA.mB);
allOk = allOk && compare_mat3x2(mat3x2(3.0, -5.0, -1.0, -5.0, -1.0, -9.0), s2.c.mA.mB);
allOk = allOk && compare_vec2(vec2(-6.0, -9.0), s2.c.mB);
allOk = allOk && compare_bvec2(bvec2(true, false), s2.d.mA);
allOk = allOk && compare_mat2(mat2(-2.0, 3.0, 7.0, 2.0), s2.d.mB.mA.mA);
allOk = allOk && compare_bvec3(bvec3(false, false, false), s2.d.mB.mA.mB);
allOk = allOk && compare_bvec4(bvec4(false, false, false, true), s2.d.mB.mA.mC);
allOk = allOk && compare_bvec3(bvec3(true, false, false), s2.d.mB.mB);
allOk = allOk && compare_ivec2(ivec2(-9, 0), s2.d.mC);
if (allOk)
passed++;
}

35
shaders-msl/comp/shared-struct-bool-cast.comp Normal file
View File

@ -0,0 +1,35 @@
#version 450
layout(local_size_x = 1) in;
layout(std140, binding = 0) buffer block { highp uint passed; };
struct S1 {
mediump ivec3 a;
highp uvec2 b;
bvec4 c;
mediump uint d;
};
bool compare_ivec3 (highp ivec3 a, highp ivec3 b) { return a == b; }
bool compare_uint (highp uint a, highp uint b) { return a == b; }
bool compare_uvec2 (highp uvec2 a, highp uvec2 b) { return a == b; }
bool compare_bvec4 (bvec4 a, bvec4 b) { return a == b; }
shared S1 s1;
void main (void) {
s1.a = ivec3(6, 8, 8);
s1.b = uvec2(4u, 4u);
s1.c = bvec4(false, false, false, true);
s1.d = 6u;
barrier();
memoryBarrier();
bool allOk = true;
allOk = allOk && compare_ivec3(ivec3(6, 8, 8), s1.a);
allOk = allOk && compare_uvec2(uvec2(4u, 4u), s1.b);
allOk = allOk && compare_bvec4(bvec4(false, false, false, true), s1.c);
allOk = allOk && compare_uint(6u, s1.d);
if (allOk)
passed++;
}

6
spirv_common.hpp
View File

@ -1636,6 +1636,12 @@ enum ExtendedDecorations
// results of interpolation can.
SPIRVCrossDecorationInterpolantComponentExpr,
// Apply to any struct type that is used in the Workgroup storage class.
// This causes matrices in MSL prior to Metal 3.0 to be emitted using a special
// class that is convertible to the standard matrix type, to work around the
// lack of constructors in the 'threadgroup' address space.
SPIRVCrossDecorationWorkgroupStruct,
SPIRVCrossDecorationCount
};

6
spirv_glsl.cpp
View File

@ -10741,9 +10741,15 @@ void CompilerGLSL::emit_instruction(const Instruction &instruction)
if (expr_type.vecsize > type.vecsize)
expr = enclose_expression(expr + vector_swizzle(type.vecsize, 0));
if (forward && ptr_expression)
ptr_expression->need_transpose = old_need_transpose;
// We might need to cast in order to load from a builtin.
cast_from_variable_load(ptr, expr, type);
if (forward && ptr_expression)
ptr_expression->need_transpose = false;
// We might be trying to load a gl_Position[N], where we should be
// doing float4[](gl_in[i].gl_Position, ...) instead.
// Similar workarounds are required for input arrays in tessellation.

268
spirv_msl.cpp
View File

@ -1966,6 +1966,13 @@ void CompilerMSL::mark_packable_structs()
(has_decoration(type.self, DecorationBlock) || has_decoration(type.self, DecorationBufferBlock)))
mark_as_packable(type);
}
if (var.storage == StorageClassWorkgroup)
{
auto *type = &this->get<SPIRType>(var.basetype);
if (type->basetype == SPIRType::Struct)
mark_as_workgroup_struct(*type);
}
});
// Physical storage buffer pointers can appear outside of the context of a variable, if the address
@ -2008,6 +2015,38 @@ void CompilerMSL::mark_as_packable(SPIRType &type)
}
}
// If the specified type is a struct, it and any nested structs
// are marked as used with workgroup storage using the SPIRVCrossDecorationWorkgroupStruct decoration.
void CompilerMSL::mark_as_workgroup_struct(SPIRType &type)
{
// If this is not the base type (eg. it's a pointer or array), tunnel down
if (type.parent_type)
{
mark_as_workgroup_struct(get<SPIRType>(type.parent_type));
return;
}
// Handle possible recursion when a struct contains a pointer to its own type nested somewhere.
if (type.basetype == SPIRType::Struct && !has_extended_decoration(type.self, SPIRVCrossDecorationWorkgroupStruct))
{
set_extended_decoration(type.self, SPIRVCrossDecorationWorkgroupStruct);
// Recurse
uint32_t mbr_cnt = uint32_t(type.member_types.size());
for (uint32_t mbr_idx = 0; mbr_idx < mbr_cnt; mbr_idx++)
{
uint32_t mbr_type_id = type.member_types[mbr_idx];
auto &mbr_type = get<SPIRType>(mbr_type_id);
mark_as_workgroup_struct(mbr_type);
if (mbr_type.type_alias)
{
auto &mbr_type_alias = get<SPIRType>(mbr_type.type_alias);
mark_as_workgroup_struct(mbr_type_alias);
}
}
}
}
// If a shader input exists at the location, it is marked as being used by this shader
void CompilerMSL::mark_location_as_used_by_shader(uint32_t location, const SPIRType &type,
StorageClass storage, bool fallback)
@ -4828,6 +4867,10 @@ void CompilerMSL::add_typedef_line(const string &line)
// Template struct like spvUnsafeArray<> need to be declared *before* any resources are declared
void CompilerMSL::emit_custom_templates()
{
static const string address_spaces[] = {
"thread", "constant", "device", "threadgroup", "threadgroup_imageblock", "ray_data", "object_data"
};
for (const auto &spv_func : spv_function_implementations)
{
switch (spv_func)
@ -4873,6 +4916,122 @@ void CompilerMSL::emit_custom_templates()
statement("");
break;
case SPVFuncImplStorageMatrix:
statement("template<typename T, int Cols, int Rows=Cols>");
statement("struct spvStorageMatrix");
begin_scope();
statement("vec<T, Rows> columns[Cols];");
statement("");
for (size_t method_idx = 0; method_idx < sizeof(address_spaces) / sizeof(address_spaces[0]); ++method_idx)
{
// Some address spaces require particular features.
if (method_idx == 4) // threadgroup_imageblock
statement("#ifdef __HAVE_IMAGEBLOCKS__");
else if (method_idx == 5) // ray_data
statement("#ifdef __HAVE_RAYTRACING__");
else if (method_idx == 6) // object_data
statement("#ifdef __HAVE_MESH__");
const string &method_as = address_spaces[method_idx];
statement("spvStorageMatrix() ", method_as, " = default;");
if (method_idx != 1) // constant
{
statement(method_as, " spvStorageMatrix& operator=(initializer_list<vec<T, Rows>> cols) ",
method_as);
begin_scope();
statement("size_t i;");
statement("thread vec<T, Rows>* col;");
statement("for (i = 0, col = cols.begin(); i < Cols; ++i, ++col)");
statement(" columns[i] = *col;");
statement("return *this;");
end_scope();
}
statement("");
for (size_t param_idx = 0; param_idx < sizeof(address_spaces) / sizeof(address_spaces[0]); ++param_idx)
{
if (param_idx != method_idx)
{
if (param_idx == 4) // threadgroup_imageblock
statement("#ifdef __HAVE_IMAGEBLOCKS__");
else if (param_idx == 5) // ray_data
statement("#ifdef __HAVE_RAYTRACING__");
else if (param_idx == 6) // object_data
statement("#ifdef __HAVE_MESH__");
}
const string &param_as = address_spaces[param_idx];
statement("spvStorageMatrix(const ", param_as, " matrix<T, Cols, Rows>& m) ", method_as);
begin_scope();
statement("for (size_t i = 0; i < Cols; ++i)");
statement(" columns[i] = m.columns[i];");
end_scope();
statement("spvStorageMatrix(const ", param_as, " spvStorageMatrix& m) ", method_as, " = default;");
if (method_idx != 1) // constant
{
statement(method_as, " spvStorageMatrix& operator=(const ", param_as,
" matrix<T, Cols, Rows>& m) ", method_as);
begin_scope();
statement("for (size_t i = 0; i < Cols; ++i)");
statement(" columns[i] = m.columns[i];");
statement("return *this;");
end_scope();
statement(method_as, " spvStorageMatrix& operator=(const ", param_as, " spvStorageMatrix& m) ",
method_as, " = default;");
}
if (param_idx != method_idx && param_idx >= 4)
statement("#endif");
statement("");
}
statement("operator matrix<T, Cols, Rows>() const ", method_as);
begin_scope();
statement("matrix<T, Cols, Rows> m;");
statement("for (int i = 0; i < Cols; ++i)");
statement(" m.columns[i] = columns[i];");
statement("return m;");
end_scope();
statement("");
statement("vec<T, Rows> operator[](size_t idx) const ", method_as);
begin_scope();
statement("return columns[idx];");
end_scope();
if (method_idx != 1) // constant
{
statement(method_as, " vec<T, Rows>& operator[](size_t idx) ", method_as);
begin_scope();
statement("return columns[idx];");
end_scope();
}
if (method_idx >= 4)
statement("#endif");
statement("");
}
end_scope_decl();
statement("");
statement("template<typename T, int Cols, int Rows>");
statement("matrix<T, Rows, Cols> transpose(spvStorageMatrix<T, Cols, Rows> m)");
begin_scope();
statement("return transpose(matrix<T, Cols, Rows>(m));");
end_scope();
statement("");
statement("typedef spvStorageMatrix<half, 2, 2> spvStorage_half2x2;");
statement("typedef spvStorageMatrix<half, 2, 3> spvStorage_half2x3;");
statement("typedef spvStorageMatrix<half, 2, 4> spvStorage_half2x4;");
statement("typedef spvStorageMatrix<half, 3, 2> spvStorage_half3x2;");
statement("typedef spvStorageMatrix<half, 3, 3> spvStorage_half3x3;");
statement("typedef spvStorageMatrix<half, 3, 4> spvStorage_half3x4;");
statement("typedef spvStorageMatrix<half, 4, 2> spvStorage_half4x2;");
statement("typedef spvStorageMatrix<half, 4, 3> spvStorage_half4x3;");
statement("typedef spvStorageMatrix<half, 4, 4> spvStorage_half4x4;");
statement("typedef spvStorageMatrix<float, 2, 2> spvStorage_float2x2;");
statement("typedef spvStorageMatrix<float, 2, 3> spvStorage_float2x3;");
statement("typedef spvStorageMatrix<float, 2, 4> spvStorage_float2x4;");
statement("typedef spvStorageMatrix<float, 3, 2> spvStorage_float3x2;");
statement("typedef spvStorageMatrix<float, 3, 3> spvStorage_float3x3;");
statement("typedef spvStorageMatrix<float, 3, 4> spvStorage_float3x4;");
statement("typedef spvStorageMatrix<float, 4, 2> spvStorage_float4x2;");
statement("typedef spvStorageMatrix<float, 4, 3> spvStorage_float4x3;");
statement("typedef spvStorageMatrix<float, 4, 4> spvStorage_float4x4;");
statement("");
break;
default:
break;
}
@ -10872,12 +11031,23 @@ string CompilerMSL::to_struct_member(const SPIRType &type, uint32_t member_type_
else if (!is_scalar(physical_type)) // scalar type is already packed.
pack_pfx = "packed_";
}
else if (row_major)
else if (is_matrix(physical_type))
{
// Need to declare type with flipped vecsize/columns.
row_major_physical_type = physical_type;
swap(row_major_physical_type.vecsize, row_major_physical_type.columns);
declared_type = &row_major_physical_type;
if (!msl_options.supports_msl_version(3, 0) &&
has_extended_decoration(type.self, SPIRVCrossDecorationWorkgroupStruct))
{
pack_pfx = "spvStorage_";
add_spv_func_and_recompile(SPVFuncImplStorageMatrix);
// The pack prefix causes problems with array<T> wrappers.
is_using_builtin_array = true;
}
if (row_major)
{
// Need to declare type with flipped vecsize/columns.
row_major_physical_type = physical_type;
swap(row_major_physical_type.vecsize, row_major_physical_type.columns);
declared_type = &row_major_physical_type;
}
}
// Very specifically, image load-store in argument buffers are disallowed on MSL on iOS.
@ -10907,8 +11077,8 @@ string CompilerMSL::to_struct_member(const SPIRType &type, uint32_t member_type_
array_type = type_to_array_glsl(physical_type);
}
auto result = join(pack_pfx, type_to_glsl(*declared_type, orig_id), " ", qualifier, to_member_name(type, index),
member_attribute_qualifier(type, index), array_type, ";");
auto result = join(pack_pfx, type_to_glsl(*declared_type, orig_id, true), " ", qualifier,
to_member_name(type, index), member_attribute_qualifier(type, index), array_type, ";");
is_using_builtin_array = false;
return result;
@ -13581,7 +13751,7 @@ string CompilerMSL::to_qualifiers_glsl(uint32_t id)
// The optional id parameter indicates the object whose type we are trying
// to find the description for. It is optional. Most type descriptions do not
// depend on a specific object's use of that type.
string CompilerMSL::type_to_glsl(const SPIRType &type, uint32_t id)
string CompilerMSL::type_to_glsl(const SPIRType &type, uint32_t id, bool member)
{
string type_name;
@ -13671,9 +13841,7 @@ string CompilerMSL::type_to_glsl(const SPIRType &type, uint32_t id)
// Need to special-case threadgroup booleans. They are supposed to be logical
// storage, but MSL compilers will sometimes crash if you use threadgroup bool.
// Workaround this by using 16-bit types instead and fixup on load-store to this data.
// FIXME: We have no sane way of working around this problem if a struct member is boolean
// and that struct is used as a threadgroup variable, but ... sigh.
if ((var && var->storage == StorageClassWorkgroup) || type.storage == StorageClassWorkgroup)
if ((var && var->storage == StorageClassWorkgroup) || type.storage == StorageClassWorkgroup || member)
type_name = "short";
else
type_name = "bool";
@ -13735,7 +13903,24 @@ string CompilerMSL::type_to_glsl(const SPIRType &type, uint32_t id)
// Matrix?
if (type.columns > 1)
{
auto *var = maybe_get_backing_variable(id);
if (var && var->basevariable)
var = &get<SPIRVariable>(var->basevariable);
// Need to special-case threadgroup matrices. Due to an oversight, Metal's
// matrix struct prior to Metal 3 lacks constructors in the threadgroup AS,
// preventing us from default-constructing or initializing matrices in threadgroup storage.
// Work around this by using our own type as storage.
if (((var && var->storage == StorageClassWorkgroup) || type.storage == StorageClassWorkgroup) &&
!msl_options.supports_msl_version(3, 0))
{
add_spv_func_and_recompile(SPVFuncImplStorageMatrix);
type_name = "spvStorage_" + type_name;
}
type_name += to_string(type.columns) + "x";
}
// Vector or Matrix?
if (type.vecsize > 1)
@ -13765,6 +13950,11 @@ string CompilerMSL::type_to_glsl(const SPIRType &type, uint32_t id)
}
}
string CompilerMSL::type_to_glsl(const SPIRType &type, uint32_t id)
{
return type_to_glsl(type, id, false);
}
string CompilerMSL::type_to_array_glsl(const SPIRType &type)
{
// Allow Metal to use the array<T> template to make arrays a value type
@ -15786,13 +15976,40 @@ void CompilerMSL::remap_constexpr_sampler_by_binding(uint32_t desc_set, uint32_t
void CompilerMSL::cast_from_variable_load(uint32_t source_id, std::string &expr, const SPIRType &expr_type)
{
bool is_packed = has_extended_decoration(source_id, SPIRVCrossDecorationPhysicalTypePacked);
auto *source_expr = maybe_get<SPIRExpression>(source_id);
auto *var = maybe_get_backing_variable(source_id);
const SPIRType *var_type, *phys_type;
if (uint32_t phys_id = get_extended_decoration(source_id, SPIRVCrossDecorationPhysicalTypeID))
phys_type = &get<SPIRType>(phys_id);
else
phys_type = &expr_type;
if (var)
{
source_id = var->self;
var_type = &get_variable_data_type(*var);
}
// Type fixups for workgroup variables if they are booleans.
if (var && var->storage == StorageClassWorkgroup && expr_type.basetype == SPIRType::Boolean)
if (var && (var->storage == StorageClassWorkgroup || var_type->basetype == SPIRType::Struct) &&
expr_type.basetype == SPIRType::Boolean)
expr = join(type_to_glsl(expr_type), "(", expr, ")");
// Type fixups for workgroup variables if they are matrices.
// Don't do fixup for packed types; those are handled specially.
// FIXME: Maybe use a type like spvStorageMatrix for packed matrices?
if (!msl_options.supports_msl_version(3, 0) && var &&
(var->storage == StorageClassWorkgroup ||
(var_type->basetype == SPIRType::Struct &&
has_extended_decoration(var_type->self, SPIRVCrossDecorationWorkgroupStruct) && !is_packed)) &&
expr_type.columns > 1)
{
SPIRType matrix_type = *phys_type;
if (source_expr && source_expr->need_transpose)
swap(matrix_type.vecsize, matrix_type.columns);
matrix_type.array.clear();
matrix_type.array_size_literal.clear();
expr = join(type_to_glsl(matrix_type), "(", expr, ")");
}
// Only interested in standalone builtin variables in the switch below.
if (!has_decoration(source_id, DecorationBuiltIn))
@ -15885,17 +16102,42 @@ void CompilerMSL::cast_from_variable_load(uint32_t source_id, std::string &expr,
void CompilerMSL::cast_to_variable_store(uint32_t target_id, std::string &expr, const SPIRType &expr_type)
{
bool is_packed = has_extended_decoration(target_id, SPIRVCrossDecorationPhysicalTypePacked);
auto *target_expr = maybe_get<SPIRExpression>(target_id);
auto *var = maybe_get_backing_variable(target_id);
const SPIRType *var_type, *phys_type;
if (uint32_t phys_id = get_extended_decoration(target_id, SPIRVCrossDecorationPhysicalTypeID))
phys_type = &get<SPIRType>(phys_id);
else
phys_type = &expr_type;
if (var)
{
target_id = var->self;
var_type = &get_variable_data_type(*var);
}
// Type fixups for workgroup variables if they are booleans.
if (var && var->storage == StorageClassWorkgroup && expr_type.basetype == SPIRType::Boolean)
if (var && (var->storage == StorageClassWorkgroup || var_type->basetype == SPIRType::Struct) &&
expr_type.basetype == SPIRType::Boolean)
{
auto short_type = expr_type;
short_type.basetype = SPIRType::Short;
expr = join(type_to_glsl(short_type), "(", expr, ")");
}
// Type fixups for workgroup variables if they are matrices.
// Don't do fixup for packed types; those are handled specially.
// FIXME: Maybe use a type like spvStorageMatrix for packed matrices?
if (!msl_options.supports_msl_version(3, 0) && var &&
(var->storage == StorageClassWorkgroup ||
(var_type->basetype == SPIRType::Struct &&
has_extended_decoration(var_type->self, SPIRVCrossDecorationWorkgroupStruct) && !is_packed)) &&
expr_type.columns > 1)
{
SPIRType matrix_type = *phys_type;
if (target_expr && target_expr->need_transpose)
swap(matrix_type.vecsize, matrix_type.columns);
expr = join("spvStorage_", type_to_glsl(matrix_type), "(", expr, ")");
}
// Only interested in standalone builtin variables.
if (!has_decoration(target_id, DecorationBuiltIn))

3
spirv_msl.hpp
View File

@ -665,6 +665,7 @@ protected:
SPVFuncImplQuantizeToF16,
SPVFuncImplCubemapTo2DArrayFace,
SPVFuncImplUnsafeArray, // Allow Metal to use the array<T> template to make arrays a value type
SPVFuncImplStorageMatrix, // Allow threadgroup construction of matrices
SPVFuncImplInverse4x4,
SPVFuncImplInverse3x3,
SPVFuncImplInverse2x2,
@ -736,6 +737,7 @@ protected:
void emit_struct_member(const SPIRType &type, uint32_t member_type_id, uint32_t index,
const std::string &qualifier = "", uint32_t base_offset = 0) override;
void emit_struct_padding_target(const SPIRType &type) override;
std::string type_to_glsl(const SPIRType &type, uint32_t id, bool member);
std::string type_to_glsl(const SPIRType &type, uint32_t id = 0) override;
void emit_block_hints(const SPIRBlock &block) override;
@ -796,6 +798,7 @@ protected:
void extract_global_variables_from_functions();
void mark_packable_structs();
void mark_as_packable(SPIRType &type);
void mark_as_workgroup_struct(SPIRType &type);
std::unordered_map<uint32_t, std::set<uint32_t>> function_global_vars;
void extract_global_variables_from_function(uint32_t func_id, std::set<uint32_t> &added_arg_ids,