AuroraMiddleware/SPIRV-Cross
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
51be601922
SPIRV-Cross/reference/shaders-msl-no-opt/frag/texture-access.swizzle.frag

266 lines
12 KiB
GLSL
Raw Normal View History

MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
#pragma clang diagnostic ignored "-Wmissing-prototypes"
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
#pragma clang diagnostic ignored "-Wmissing-braces"
#pragma clang diagnostic ignored "-Wunused-variable"
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
#include <metal_stdlib>
#include <simd/simd.h>
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
template <typename T, size_t Num>
struct unsafe_array
{
T __Elements[Num ? Num : 1];
constexpr size_t size() const thread { return Num; }
constexpr size_t max_size() const thread { return Num; }
constexpr bool empty() const thread { return Num == 0; }
constexpr size_t size() const device { return Num; }
constexpr size_t max_size() const device { return Num; }
constexpr bool empty() const device { return Num == 0; }
constexpr size_t size() const constant { return Num; }
constexpr size_t max_size() const constant { return Num; }
constexpr bool empty() const constant { return Num == 0; }
constexpr size_t size() const threadgroup { return Num; }
constexpr size_t max_size() const threadgroup { return Num; }
constexpr bool empty() const threadgroup { return Num == 0; }
thread T &operator[](size_t pos) thread
{
return __Elements[pos];
}
constexpr const thread T &operator[](size_t pos) const thread
{
return __Elements[pos];
}
device T &operator[](size_t pos) device
{
return __Elements[pos];
}
constexpr const device T &operator[](size_t pos) const device
{
return __Elements[pos];
}
constexpr const constant T &operator[](size_t pos) const constant
{
return __Elements[pos];
}
threadgroup T &operator[](size_t pos) threadgroup
{
return __Elements[pos];
}
constexpr const threadgroup T &operator[](size_t pos) const threadgroup
{
return __Elements[pos];
}
};
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
using namespace metal;
// Returns 2D texture coords corresponding to 1D texel buffer coords
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
<<<<<<< HEAD
Update tests to account for all non-entry-point functions being inlined
2019-08-29 21:39:06 +00:00
inline uint2 spvTexelBufferCoord(uint tc)
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
=======
static inline __attribute__((always_inline))
uint2 spvTexelBufferCoord(uint tc)
>>>>>>> 22755a0c... Update the Metal shaders to account for changes in the shader compilation.
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
return uint2(tc % 4096, tc / 4096);
}
template<typename T> struct spvRemoveReference { typedef T type; };
template<typename T> struct spvRemoveReference<thread T&> { typedef T type; };
template<typename T> struct spvRemoveReference<thread T&&> { typedef T type; };
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
template<typename T> static inline __attribute__((always_inline)) constexpr thread T&& spvForward(thread typename spvRemoveReference<T>::type& x)
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
return static_cast<thread T&&>(x);
}
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
template<typename T> static inline __attribute__((always_inline)) constexpr thread T&& spvForward(thread typename spvRemoveReference<T>::type&& x)
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
return static_cast<thread T&&>(x);
}
MSL: Add support for sampler Y'CbCr conversion. This change introduces functions and in one case, a class, to support the `VK_KHR_sampler_ycbcr_conversion` extension. Except in the case of GBGR8 and BGRG8 formats, for which Metal natively supports implicit chroma reconstruction, we're on our own here. We have to do everything ourselves. Much of the complexity comes from the need to support multiple planes, which must now be passed to functions that use the corresponding combined image-samplers. The rest is from the actual Y'CbCr conversion itself, which requires additional post-processing of the sample retrieved from the image. Passing sampled images to a function was a particular problem. To support this, I've added a new class which is emitted to MSL shaders that pass sampled images with Y'CbCr conversions attached around. It can handle sampled images with or without Y'CbCr conversion. This is an awful abomination that should not exist, but I'm worried that there's some shader out there which does this. This support requires Metal 2.0 to work properly, because it uses default-constructed texture objects, which were only added in MSL 2. I'm not even going to get into arrays of combined image-samplers--that's a whole other can of worms. They are deliberately unsupported in this change. I've taken the liberty of refactoring the support for texture swizzling while I'm at it. It's now treated as a post-processing step similar to Y'CbCr conversion. I'd like to think this is cleaner than having everything in `to_function_name()`/`to_function_args()`. It still looks really hairy, though. I did, however, get rid of the explicit type arguments to `spvGatherSwizzle()`/`spvGatherCompareSwizzle()`. Update the C API. In addition to supporting this new functionality, add some compiler options that I added in previous changes, but for which I neglected to update the C API.
2019-08-02 20:11:19 +00:00
enum class spvSwizzle : uint
{
none = 0,
zero,
one,
red,
green,
blue,
alpha
};
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
template<typename T>
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
static inline __attribute__((always_inline))
T spvGetSwizzle(vec<T, 4> x, T c, spvSwizzle s)
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
switch (s)
{
MSL: Fix mapping of identity-swizzled components. Before, if any component was not identity-mapped, those components that were still identity-mapped were set to 0. Now we properly leave them alone.
2019-01-07 17:17:12 +00:00
case spvSwizzle::none:
return c;
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
case spvSwizzle::zero:
return 0;
case spvSwizzle::one:
return 1;
case spvSwizzle::red:
return x.r;
case spvSwizzle::green:
return x.g;
case spvSwizzle::blue:
return x.b;
case spvSwizzle::alpha:
return x.a;
}
}
// Wrapper function that swizzles texture samples and fetches.
template<typename T>
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
static inline __attribute__((always_inline))
vec<T, 4> spvTextureSwizzle(vec<T, 4> x, uint s)
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
if (!s)
return x;
MSL: Fix mapping of identity-swizzled components. Before, if any component was not identity-mapped, those components that were still identity-mapped were set to 0. Now we properly leave them alone.
2019-01-07 17:17:12 +00:00
return vec<T, 4>(spvGetSwizzle(x, x.r, spvSwizzle((s >> 0) & 0xFF)), spvGetSwizzle(x, x.g, spvSwizzle((s >> 8) & 0xFF)), spvGetSwizzle(x, x.b, spvSwizzle((s >> 16) & 0xFF)), spvGetSwizzle(x, x.a, spvSwizzle((s >> 24) & 0xFF)));
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
}
template<typename T>
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
static inline __attribute__((always_inline))
T spvTextureSwizzle(T x, uint s)
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
return spvTextureSwizzle(vec<T, 4>(x, 0, 0, 1), s).x;
}
// Wrapper function that swizzles texture gathers.
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
<<<<<<< HEAD
MSL: Add support for sampler Y'CbCr conversion. This change introduces functions and in one case, a class, to support the `VK_KHR_sampler_ycbcr_conversion` extension. Except in the case of GBGR8 and BGRG8 formats, for which Metal natively supports implicit chroma reconstruction, we're on our own here. We have to do everything ourselves. Much of the complexity comes from the need to support multiple planes, which must now be passed to functions that use the corresponding combined image-samplers. The rest is from the actual Y'CbCr conversion itself, which requires additional post-processing of the sample retrieved from the image. Passing sampled images to a function was a particular problem. To support this, I've added a new class which is emitted to MSL shaders that pass sampled images with Y'CbCr conversions attached around. It can handle sampled images with or without Y'CbCr conversion. This is an awful abomination that should not exist, but I'm worried that there's some shader out there which does this. This support requires Metal 2.0 to work properly, because it uses default-constructed texture objects, which were only added in MSL 2. I'm not even going to get into arrays of combined image-samplers--that's a whole other can of worms. They are deliberately unsupported in this change. I've taken the liberty of refactoring the support for texture swizzling while I'm at it. It's now treated as a post-processing step similar to Y'CbCr conversion. I'd like to think this is cleaner than having everything in `to_function_name()`/`to_function_args()`. It still looks really hairy, though. I did, however, get rid of the explicit type arguments to `spvGatherSwizzle()`/`spvGatherCompareSwizzle()`. Update the C API. In addition to supporting this new functionality, add some compiler options that I added in previous changes, but for which I neglected to update the C API.
2019-08-02 20:11:19 +00:00
template<typename T, template<typename, access = access::sample, typename = void> class Tex, typename... Ts>
inline vec<T, 4> spvGatherSwizzle(const thread Tex<T>& t, sampler s, uint sw, component c, Ts... params) METAL_CONST_ARG(c)
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
=======
template<typename T, typename Tex, typename... Ts>
static inline __attribute__((always_inline))
vec<T, 4> spvGatherSwizzle(sampler s, thread Tex const& t, Ts... params, component c, uint sw) METAL_CONST_ARG(c)
>>>>>>> 22755a0c... Update the Metal shaders to account for changes in the shader compilation.
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
if (sw)
{
Pack texture component swizzles by bytes.
2018-09-22 03:50:29 +00:00
switch (spvSwizzle((sw >> (uint(c) * 8)) & 0xFF))
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
case spvSwizzle::none:
break;
case spvSwizzle::zero:
return vec<T, 4>(0, 0, 0, 0);
case spvSwizzle::one:
return vec<T, 4>(1, 1, 1, 1);
case spvSwizzle::red:
return t.gather(s, spvForward<Ts>(params)..., component::x);
case spvSwizzle::green:
return t.gather(s, spvForward<Ts>(params)..., component::y);
case spvSwizzle::blue:
return t.gather(s, spvForward<Ts>(params)..., component::z);
case spvSwizzle::alpha:
return t.gather(s, spvForward<Ts>(params)..., component::w);
}
}
switch (c)
{
case component::x:
return t.gather(s, spvForward<Ts>(params)..., component::x);
case component::y:
return t.gather(s, spvForward<Ts>(params)..., component::y);
case component::z:
return t.gather(s, spvForward<Ts>(params)..., component::z);
case component::w:
return t.gather(s, spvForward<Ts>(params)..., component::w);
}
}
MSL: Swizzle gathers on depth textures as well. Might as well.
2018-09-29 14:52:26 +00:00
// Wrapper function that swizzles depth texture gathers.
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
<<<<<<< HEAD
MSL: Add support for sampler Y'CbCr conversion. This change introduces functions and in one case, a class, to support the `VK_KHR_sampler_ycbcr_conversion` extension. Except in the case of GBGR8 and BGRG8 formats, for which Metal natively supports implicit chroma reconstruction, we're on our own here. We have to do everything ourselves. Much of the complexity comes from the need to support multiple planes, which must now be passed to functions that use the corresponding combined image-samplers. The rest is from the actual Y'CbCr conversion itself, which requires additional post-processing of the sample retrieved from the image. Passing sampled images to a function was a particular problem. To support this, I've added a new class which is emitted to MSL shaders that pass sampled images with Y'CbCr conversions attached around. It can handle sampled images with or without Y'CbCr conversion. This is an awful abomination that should not exist, but I'm worried that there's some shader out there which does this. This support requires Metal 2.0 to work properly, because it uses default-constructed texture objects, which were only added in MSL 2. I'm not even going to get into arrays of combined image-samplers--that's a whole other can of worms. They are deliberately unsupported in this change. I've taken the liberty of refactoring the support for texture swizzling while I'm at it. It's now treated as a post-processing step similar to Y'CbCr conversion. I'd like to think this is cleaner than having everything in `to_function_name()`/`to_function_args()`. It still looks really hairy, though. I did, however, get rid of the explicit type arguments to `spvGatherSwizzle()`/`spvGatherCompareSwizzle()`. Update the C API. In addition to supporting this new functionality, add some compiler options that I added in previous changes, but for which I neglected to update the C API.
2019-08-02 20:11:19 +00:00
template<typename T, template<typename, access = access::sample, typename = void> class Tex, typename... Ts>
inline vec<T, 4> spvGatherCompareSwizzle(const thread Tex<T>& t, sampler s, uint sw, Ts... params)
Update the Metal shaders to account for changes in the shader compilation.
2019-08-14 15:09:39 +00:00
=======
template<typename T, typename Tex, typename... Ts>
static inline __attribute__((always_inline))
vec<T, 4> spvGatherCompareSwizzle(sampler s, thread Tex const& t, Ts... params, uint sw)
>>>>>>> 22755a0c... Update the Metal shaders to account for changes in the shader compilation.
MSL: Swizzle gathers on depth textures as well. Might as well.
2018-09-29 14:52:26 +00:00
{
if (sw)
{
switch (spvSwizzle(sw & 0xFF))
{
case spvSwizzle::none:
case spvSwizzle::red:
break;
case spvSwizzle::zero:
case spvSwizzle::green:
case spvSwizzle::blue:
case spvSwizzle::alpha:
return vec<T, 4>(0, 0, 0, 0);
case spvSwizzle::one:
return vec<T, 4>(1, 1, 1, 1);
}
}
return t.gather_compare(s, spvForward<Ts>(params)...);
}
MSL: Rewrite how resource indices are fallback-assigned. We used to use the Binding decoration for this, but this method is hopelessly broken. If no explicit MSL resource remapping exists, we remap automatically in a manner which should always "just work".
2019-06-21 10:44:33 +00:00
fragment void main0(constant uint* spvSwizzleConstants [[buffer(30)]], texture1d<float> tex1d [[texture(0)]], texture2d<float> tex2d [[texture(1)]], texture3d<float> tex3d [[texture(2)]], texturecube<float> texCube [[texture(3)]], texture2d_array<float> tex2dArray [[texture(4)]], texturecube_array<float> texCubeArray [[texture(5)]], depth2d<float> depth2d [[texture(6)]], depthcube<float> depthCube [[texture(7)]], depth2d_array<float> depth2dArray [[texture(8)]], depthcube_array<float> depthCubeArray [[texture(9)]], texture2d<float> texBuffer [[texture(10)]], sampler tex1dSmplr [[sampler(0)]], sampler tex2dSmplr [[sampler(1)]], sampler tex3dSmplr [[sampler(2)]], sampler texCubeSmplr [[sampler(3)]], sampler tex2dArraySmplr [[sampler(4)]], sampler texCubeArraySmplr [[sampler(5)]], sampler depth2dSmplr [[sampler(6)]], sampler depthCubeSmplr [[sampler(7)]], sampler depth2dArraySmplr [[sampler(8)]], sampler depthCubeArraySmplr [[sampler(9)]])
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
{
MSL: Implement OpArrayLength. This gets rather complicated because MSL does not support OpArrayLength natively. We need to pass down a buffer which contains buffer sizes, and we compute the array length on-demand. Support both discrete descriptors as well as argument buffers.
2019-05-27 09:59:29 +00:00
constant uint& tex1dSwzl = spvSwizzleConstants[0];
constant uint& tex2dSwzl = spvSwizzleConstants[1];
constant uint& tex3dSwzl = spvSwizzleConstants[2];
constant uint& texCubeSwzl = spvSwizzleConstants[3];
constant uint& tex2dArraySwzl = spvSwizzleConstants[4];
constant uint& texCubeArraySwzl = spvSwizzleConstants[5];
MSL: Rewrite how resource indices are fallback-assigned. We used to use the Binding decoration for this, but this method is hopelessly broken. If no explicit MSL resource remapping exists, we remap automatically in a manner which should always "just work".
2019-06-21 10:44:33 +00:00
constant uint& depth2dSwzl = spvSwizzleConstants[6];
constant uint& depthCubeSwzl = spvSwizzleConstants[7];
constant uint& depth2dArraySwzl = spvSwizzleConstants[8];
constant uint& depthCubeArraySwzl = spvSwizzleConstants[9];
MSL: Fix passing a sampled image to a function. In the past, SPIRV-Cross threw an error in this case because it couldn't work out which swizzle from the auxiliary buffer needs to be passed. Now, we pass the swizzle around with the texture object, like a combined image-sampler and its associated sampler.
2019-01-13 23:31:50 +00:00
float4 c = spvTextureSwizzle(tex1d.sample(tex1dSmplr, 0.0), tex1dSwzl);
c = spvTextureSwizzle(tex2d.sample(tex2dSmplr, float2(0.0)), tex2dSwzl);
c = spvTextureSwizzle(tex3d.sample(tex3dSmplr, float3(0.0)), tex3dSwzl);
c = spvTextureSwizzle(texCube.sample(texCubeSmplr, float3(0.0)), texCubeSwzl);
c = spvTextureSwizzle(tex2dArray.sample(tex2dArraySmplr, float3(0.0).xy, uint(round(float3(0.0).z))), tex2dArraySwzl);
c = spvTextureSwizzle(texCubeArray.sample(texCubeArraySmplr, float4(0.0).xyz, uint(round(float4(0.0).w))), texCubeArraySwzl);
c.x = spvTextureSwizzle(depth2d.sample_compare(depth2dSmplr, float3(0.0, 0.0, 1.0).xy, float3(0.0, 0.0, 1.0).z), depth2dSwzl);
c.x = spvTextureSwizzle(depthCube.sample_compare(depthCubeSmplr, float4(0.0, 0.0, 0.0, 1.0).xyz, float4(0.0, 0.0, 0.0, 1.0).w), depthCubeSwzl);
c.x = spvTextureSwizzle(depth2dArray.sample_compare(depth2dArraySmplr, float4(0.0, 0.0, 0.0, 1.0).xy, uint(round(float4(0.0, 0.0, 0.0, 1.0).z)), float4(0.0, 0.0, 0.0, 1.0).w), depth2dArraySwzl);
c.x = spvTextureSwizzle(depthCubeArray.sample_compare(depthCubeArraySmplr, float4(0.0).xyz, uint(round(float4(0.0).w)), 1.0), depthCubeArraySwzl);
c = spvTextureSwizzle(tex1d.sample(tex1dSmplr, float2(0.0, 1.0).x / float2(0.0, 1.0).y), tex1dSwzl);
c = spvTextureSwizzle(tex2d.sample(tex2dSmplr, float3(0.0, 0.0, 1.0).xy / float3(0.0, 0.0, 1.0).z), tex2dSwzl);
c = spvTextureSwizzle(tex3d.sample(tex3dSmplr, float4(0.0, 0.0, 0.0, 1.0).xyz / float4(0.0, 0.0, 0.0, 1.0).w), tex3dSwzl);
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
float4 _100 = float4(0.0, 0.0, 1.0, 1.0);
_100.z = float4(0.0, 0.0, 1.0, 1.0).w;
HLSL/MSL: Fix texture projection with Dref. We need to divide the Dref by q.
2019-01-28 08:39:45 +00:00
c.x = spvTextureSwizzle(depth2d.sample_compare(depth2dSmplr, _100.xy / _100.z, float4(0.0, 0.0, 1.0, 1.0).z / _100.z), depth2dSwzl);
MSL: Fix passing a sampled image to a function. In the past, SPIRV-Cross threw an error in this case because it couldn't work out which swizzle from the auxiliary buffer needs to be passed. Now, we pass the swizzle around with the texture object, like a combined image-sampler and its associated sampler.
2019-01-13 23:31:50 +00:00
c = spvTextureSwizzle(tex1d.sample(tex1dSmplr, 0.0), tex1dSwzl);
c = spvTextureSwizzle(tex2d.sample(tex2dSmplr, float2(0.0), level(0.0)), tex2dSwzl);
c = spvTextureSwizzle(tex3d.sample(tex3dSmplr, float3(0.0), level(0.0)), tex3dSwzl);
c = spvTextureSwizzle(texCube.sample(texCubeSmplr, float3(0.0), level(0.0)), texCubeSwzl);
c = spvTextureSwizzle(tex2dArray.sample(tex2dArraySmplr, float3(0.0).xy, uint(round(float3(0.0).z)), level(0.0)), tex2dArraySwzl);
c = spvTextureSwizzle(texCubeArray.sample(texCubeArraySmplr, float4(0.0).xyz, uint(round(float4(0.0).w)), level(0.0)), texCubeArraySwzl);
c.x = spvTextureSwizzle(depth2d.sample_compare(depth2dSmplr, float3(0.0, 0.0, 1.0).xy, float3(0.0, 0.0, 1.0).z, level(0.0)), depth2dSwzl);
c = spvTextureSwizzle(tex1d.sample(tex1dSmplr, float2(0.0, 1.0).x / float2(0.0, 1.0).y), tex1dSwzl);
c = spvTextureSwizzle(tex2d.sample(tex2dSmplr, float3(0.0, 0.0, 1.0).xy / float3(0.0, 0.0, 1.0).z, level(0.0)), tex2dSwzl);
c = spvTextureSwizzle(tex3d.sample(tex3dSmplr, float4(0.0, 0.0, 0.0, 1.0).xyz / float4(0.0, 0.0, 0.0, 1.0).w, level(0.0)), tex3dSwzl);
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
float4 _128 = float4(0.0, 0.0, 1.0, 1.0);
_128.z = float4(0.0, 0.0, 1.0, 1.0).w;
HLSL/MSL: Fix texture projection with Dref. We need to divide the Dref by q.
2019-01-28 08:39:45 +00:00
c.x = spvTextureSwizzle(depth2d.sample_compare(depth2dSmplr, _128.xy / _128.z, float4(0.0, 0.0, 1.0, 1.0).z / _128.z, level(0.0)), depth2dSwzl);
MSL: Fix passing a sampled image to a function. In the past, SPIRV-Cross threw an error in this case because it couldn't work out which swizzle from the auxiliary buffer needs to be passed. Now, we pass the swizzle around with the texture object, like a combined image-sampler and its associated sampler.
2019-01-13 23:31:50 +00:00
c = spvTextureSwizzle(tex1d.read(uint(0)), tex1dSwzl);
c = spvTextureSwizzle(tex2d.read(uint2(int2(0)), 0), tex2dSwzl);
c = spvTextureSwizzle(tex3d.read(uint3(int3(0)), 0), tex3dSwzl);
c = spvTextureSwizzle(tex2dArray.read(uint2(int3(0).xy), uint(int3(0).z), 0), tex2dArraySwzl);
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
c = texBuffer.read(spvTexelBufferCoord(0));
MSL: Add support for sampler Y'CbCr conversion. This change introduces functions and in one case, a class, to support the `VK_KHR_sampler_ycbcr_conversion` extension. Except in the case of GBGR8 and BGRG8 formats, for which Metal natively supports implicit chroma reconstruction, we're on our own here. We have to do everything ourselves. Much of the complexity comes from the need to support multiple planes, which must now be passed to functions that use the corresponding combined image-samplers. The rest is from the actual Y'CbCr conversion itself, which requires additional post-processing of the sample retrieved from the image. Passing sampled images to a function was a particular problem. To support this, I've added a new class which is emitted to MSL shaders that pass sampled images with Y'CbCr conversions attached around. It can handle sampled images with or without Y'CbCr conversion. This is an awful abomination that should not exist, but I'm worried that there's some shader out there which does this. This support requires Metal 2.0 to work properly, because it uses default-constructed texture objects, which were only added in MSL 2. I'm not even going to get into arrays of combined image-samplers--that's a whole other can of worms. They are deliberately unsupported in this change. I've taken the liberty of refactoring the support for texture swizzling while I'm at it. It's now treated as a post-processing step similar to Y'CbCr conversion. I'd like to think this is cleaner than having everything in `to_function_name()`/`to_function_args()`. It still looks really hairy, though. I did, however, get rid of the explicit type arguments to `spvGatherSwizzle()`/`spvGatherCompareSwizzle()`. Update the C API. In addition to supporting this new functionality, add some compiler options that I added in previous changes, but for which I neglected to update the C API.
2019-08-02 20:11:19 +00:00
c = spvGatherSwizzle(tex2d, tex2dSmplr, tex2dSwzl, component::x, float2(0.0), int2(0));
c = spvGatherSwizzle(texCube, texCubeSmplr, texCubeSwzl, component::y, float3(0.0));
c = spvGatherSwizzle(tex2dArray, tex2dArraySmplr, tex2dArraySwzl, component::z, float3(0.0).xy, uint(round(float3(0.0).z)), int2(0));
c = spvGatherSwizzle(texCubeArray, texCubeArraySmplr, texCubeArraySwzl, component::w, float4(0.0).xyz, uint(round(float4(0.0).w)));
c = spvGatherCompareSwizzle(depth2d, depth2dSmplr, depth2dSwzl, float2(0.0), 1.0);
c = spvGatherCompareSwizzle(depthCube, depthCubeSmplr, depthCubeSwzl, float3(0.0), 1.0);
c = spvGatherCompareSwizzle(depth2dArray, depth2dArraySmplr, depth2dArraySwzl, float3(0.0).xy, uint(round(float3(0.0).z)), 1.0);
c = spvGatherCompareSwizzle(depthCubeArray, depthCubeArraySmplr, depthCubeArraySwzl, float4(0.0).xyz, uint(round(float4(0.0).w)), 1.0);
MSL: Add an option to insert texture swizzles into generated shaders. It's intended to be used with MoltenVK to support arbitrary `VkComponentMapping` settings. The idea is that MoltenVK will pass a buffer (which it set to some buffer index that isn't being used) containing packed versions of the `VkComponentMapping` struct, one for each sampled image. Yes, this is horribly ugly. It is unfortunately necessary. Much of the ugliness is to support swizzling gather operations, where we need to alter the component that the gather operates on--something complicated by the `gather()` method requiring the passed-in component to be a constant expression. It doesn't even support swizzling gathers on depth textures, though I could add that if it turns out we need it.
2018-09-20 01:36:33 +00:00
}
Reference in New Issue Copy Permalink