skia2/tests/sksl/shared/ArrayCast.metal

88 lines
3.0 KiB
Metal
Raw Normal View History

Add support for array-cast syntax in SkSL. Compiling a program with "allow narrowing conversions" actually fixes up narrowing casts in the program by inserting casts wherever they would be needed for type-correctness. For instance, compiling the statement `half h = myFloat;` inserts an appropriate narrowing cast: `half h = half(myFloat);`. The Pipeline stage code generator relies on this behavior, as when it re-emits a runtime effect into a complete SkSL program, the narrowing- conversions flag will no longer be set, but that is okay, because the emitted code now contains typecasts anywhere they would be necessary. Logically, this implies that anything which supports narrowing conversions must be castable between high and low precision. In GLSL and SPIR-V, such a cast is trivial, because the types are the same and the precision qualifiers are treated as individual hints on each variable. In Metal, we dodge the issue by only emitting full-precision types. But we also need to emit raw SkSL from an SkSL program (that is what the Pipeline stage generator does). SkSL already supported every typical cast, but GLSL lacked any syntax for casting an array to a different type. This meant SkSL had no array casting syntax as well. SkSL now has array-cast syntax, but it is only allowed for casting low/high-precision arrays to the same base type. (You can't cast an int array to float, or a signed array to unsigned.) Change-Id: Ia20933541c3bd4a946c1ea38209f93008acdb9cb Bug: skia:12248 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/437687 Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
2021-08-10 20:03:44 +00:00
#include <metal_stdlib>
#include <simd/simd.h>
using namespace metal;
struct Uniforms {
float4 colorGreen;
float4 colorRed;
};
struct Inputs {
};
struct Outputs {
float4 sk_FragColor [[color(0)]];
};
template <typename T1, typename T2, size_t N>
bool operator==(thread const array<T1, N>& left, thread const array<T2, N>& right);
template <typename T1, typename T2, size_t N>
bool operator!=(thread const array<T1, N>& left, thread const array<T2, N>& right);
thread bool operator==(const float2x2 left, const float2x2 right);
thread bool operator!=(const float2x2 left, const float2x2 right);
template <size_t N>
array<float, N> array_of_float_from_float(thread const array<float, N>& x) {
array<float, N> result;
for (int i = 0; i < N; ++i) {
result[i] = float(x[i]);
}
return result;
}
template <size_t N>
array<int3, N> array_of_int3_from_int3(thread const array<int3, N>& x) {
array<int3, N> result;
for (int i = 0; i < N; ++i) {
result[i] = int3(x[i]);
}
return result;
}
template <size_t N>
array<float2x2, N> array_of_float2x2_from_float2x2(thread const array<float2x2, N>& x) {
array<float2x2, N> result;
for (int i = 0; i < N; ++i) {
result[i] = float2x2(x[i]);
}
return result;
}
Add support for array-cast syntax in SkSL. Compiling a program with "allow narrowing conversions" actually fixes up narrowing casts in the program by inserting casts wherever they would be needed for type-correctness. For instance, compiling the statement `half h = myFloat;` inserts an appropriate narrowing cast: `half h = half(myFloat);`. The Pipeline stage code generator relies on this behavior, as when it re-emits a runtime effect into a complete SkSL program, the narrowing- conversions flag will no longer be set, but that is okay, because the emitted code now contains typecasts anywhere they would be necessary. Logically, this implies that anything which supports narrowing conversions must be castable between high and low precision. In GLSL and SPIR-V, such a cast is trivial, because the types are the same and the precision qualifiers are treated as individual hints on each variable. In Metal, we dodge the issue by only emitting full-precision types. But we also need to emit raw SkSL from an SkSL program (that is what the Pipeline stage generator does). SkSL already supported every typical cast, but GLSL lacked any syntax for casting an array to a different type. This meant SkSL had no array casting syntax as well. SkSL now has array-cast syntax, but it is only allowed for casting low/high-precision arrays to the same base type. (You can't cast an int array to float, or a signed array to unsigned.) Change-Id: Ia20933541c3bd4a946c1ea38209f93008acdb9cb Bug: skia:12248 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/437687 Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
2021-08-10 20:03:44 +00:00
template <typename T1, typename T2, size_t N>
bool operator==(thread const array<T1, N>& left, thread const array<T2, N>& right) {
for (size_t index = 0; index < N; ++index) {
if (!all(left[index] == right[index])) {
return false;
}
}
return true;
}
template <typename T1, typename T2, size_t N>
bool operator!=(thread const array<T1, N>& left, thread const array<T2, N>& right) {
return !(left == right);
}
thread bool operator==(const float2x2 left, const float2x2 right) {
return all(left[0] == right[0]) &&
all(left[1] == right[1]);
}
thread bool operator!=(const float2x2 left, const float2x2 right) {
return !(left == right);
}
fragment Outputs fragmentMain(Inputs _in [[stage_in]], constant Uniforms& _uniforms [[buffer(0)]], bool _frontFacing [[front_facing]], float4 _fragCoord [[position]]) {
Outputs _out;
(void)_out;
array<float, 4> f = array<float, 4>{1.0, 2.0, 3.0, 4.0};
array<float, 4> h = array_of_float_from_float(f);
f = array_of_float_from_float(h);
h = array_of_float_from_float(f);
Add support for array-cast syntax in SkSL. Compiling a program with "allow narrowing conversions" actually fixes up narrowing casts in the program by inserting casts wherever they would be needed for type-correctness. For instance, compiling the statement `half h = myFloat;` inserts an appropriate narrowing cast: `half h = half(myFloat);`. The Pipeline stage code generator relies on this behavior, as when it re-emits a runtime effect into a complete SkSL program, the narrowing- conversions flag will no longer be set, but that is okay, because the emitted code now contains typecasts anywhere they would be necessary. Logically, this implies that anything which supports narrowing conversions must be castable between high and low precision. In GLSL and SPIR-V, such a cast is trivial, because the types are the same and the precision qualifiers are treated as individual hints on each variable. In Metal, we dodge the issue by only emitting full-precision types. But we also need to emit raw SkSL from an SkSL program (that is what the Pipeline stage generator does). SkSL already supported every typical cast, but GLSL lacked any syntax for casting an array to a different type. This meant SkSL had no array casting syntax as well. SkSL now has array-cast syntax, but it is only allowed for casting low/high-precision arrays to the same base type. (You can't cast an int array to float, or a signed array to unsigned.) Change-Id: Ia20933541c3bd4a946c1ea38209f93008acdb9cb Bug: skia:12248 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/437687 Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
2021-08-10 20:03:44 +00:00
array<int3, 3> i3 = array<int3, 3>{int3(1), int3(2), int3(3)};
array<int3, 3> s3 = array_of_int3_from_int3(i3);
i3 = array_of_int3_from_int3(s3);
s3 = array_of_int3_from_int3(i3);
Add support for array-cast syntax in SkSL. Compiling a program with "allow narrowing conversions" actually fixes up narrowing casts in the program by inserting casts wherever they would be needed for type-correctness. For instance, compiling the statement `half h = myFloat;` inserts an appropriate narrowing cast: `half h = half(myFloat);`. The Pipeline stage code generator relies on this behavior, as when it re-emits a runtime effect into a complete SkSL program, the narrowing- conversions flag will no longer be set, but that is okay, because the emitted code now contains typecasts anywhere they would be necessary. Logically, this implies that anything which supports narrowing conversions must be castable between high and low precision. In GLSL and SPIR-V, such a cast is trivial, because the types are the same and the precision qualifiers are treated as individual hints on each variable. In Metal, we dodge the issue by only emitting full-precision types. But we also need to emit raw SkSL from an SkSL program (that is what the Pipeline stage generator does). SkSL already supported every typical cast, but GLSL lacked any syntax for casting an array to a different type. This meant SkSL had no array casting syntax as well. SkSL now has array-cast syntax, but it is only allowed for casting low/high-precision arrays to the same base type. (You can't cast an int array to float, or a signed array to unsigned.) Change-Id: Ia20933541c3bd4a946c1ea38209f93008acdb9cb Bug: skia:12248 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/437687 Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
2021-08-10 20:03:44 +00:00
array<float2x2, 2> h2x2 = array<float2x2, 2>{float2x2(float2(1.0, 2.0), float2(3.0, 4.0)), float2x2(float2(5.0, 6.0), float2(7.0, 8.0))};
array<float2x2, 2> f2x2 = array_of_float2x2_from_float2x2(h2x2);
f2x2 = array_of_float2x2_from_float2x2(h2x2);
h2x2 = array_of_float2x2_from_float2x2(f2x2);
_out.sk_FragColor = (f == array_of_float_from_float(h) && i3 == array_of_int3_from_int3(s3)) && f2x2 == array_of_float2x2_from_float2x2(h2x2) ? _uniforms.colorGreen : _uniforms.colorRed;
Add support for array-cast syntax in SkSL. Compiling a program with "allow narrowing conversions" actually fixes up narrowing casts in the program by inserting casts wherever they would be needed for type-correctness. For instance, compiling the statement `half h = myFloat;` inserts an appropriate narrowing cast: `half h = half(myFloat);`. The Pipeline stage code generator relies on this behavior, as when it re-emits a runtime effect into a complete SkSL program, the narrowing- conversions flag will no longer be set, but that is okay, because the emitted code now contains typecasts anywhere they would be necessary. Logically, this implies that anything which supports narrowing conversions must be castable between high and low precision. In GLSL and SPIR-V, such a cast is trivial, because the types are the same and the precision qualifiers are treated as individual hints on each variable. In Metal, we dodge the issue by only emitting full-precision types. But we also need to emit raw SkSL from an SkSL program (that is what the Pipeline stage generator does). SkSL already supported every typical cast, but GLSL lacked any syntax for casting an array to a different type. This meant SkSL had no array casting syntax as well. SkSL now has array-cast syntax, but it is only allowed for casting low/high-precision arrays to the same base type. (You can't cast an int array to float, or a signed array to unsigned.) Change-Id: Ia20933541c3bd4a946c1ea38209f93008acdb9cb Bug: skia:12248 Reviewed-on: https://skia-review.googlesource.com/c/skia/+/437687 Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
2021-08-10 20:03:44 +00:00
return _out;
}