encounter/dawn-cmake
1
0
Fork 0
mirror of https://github.com/encounter/dawn-cmake.git synced 2025-12-16 16:37:08 +00:00
Code Issues Packages Projects Releases Wiki Activity

tint/msl: Preserve trailing vec3 padding

Browse Source 
In order to preserve padding properly for MSL, we need to use its
packed_vec type for all vec3 types in storage buffers, not just struct
members. This commit includes a complete rewrite of the PackedVec3
transform to achieve this. The key details are:

* An internal `__packed_vec3<>` type was added, which corresponds to a
  `type::Vector` with an additional flag to indicate that it will be
  emitted as packed vector.

* The `PackedVec3` transform replaces all vec3 types used in
  host-shareable address spaces with the internal `__packed_vec3`
  type. This includes vec3 types that appear as the store type of a
  pointer.

* When used as an array element, these `__packed_vec3` types are
  wrapped in a struct that contains a single `__packed_vec3`
  member. This allows us to add an `@align()` attribute that ensures
  that `array<vec3<T>>` still has the correct array element stride.

* When the `vec3<T>` appears as a struct member in the input program,
  we apply the `@align()` to that member to ensure that we do not
  change its offset.

* Matrix types with three rows that are used in memory are replaced
  with an array of columns, where each column uses a `__packed_vec3`
  inside an aligned wrapper structure as above.

* Accesses to host-shareable memory that involve any of these types
  invoke a "pack" or "unpack" helper function to convert them to the
  equivalent type that uses `__packed_vec3` or a regular `vec3` as
  required.

* The `chromium_internal_relaxed_uniform_layout` extension is used to
  avoid issues where modifying a type in the uniform address space
  triggers stricter layout validation rules.

Bug: tint:1571
Fixed: tint:1837
Change-Id: Idaf2da2f5bcb2be00c85ec657edfb614186476bb
Reviewed-on: https://dawn-review.googlesource.com/c/dawn/+/121200
Reviewed-by: Ben Clayton <bclayton@google.com>
Commit-Queue: James Price <jrprice@google.com>
Kokoro: Kokoro <noreply+kokoro@google.com>
This commit is contained in:
James Price
2023-02-27 20:21:03 +00:00
committed by Dawn LUCI CQ
parent 55183e6c3a
commit 4d3af66bbd
389 changed files with 16692 additions and 3527 deletions
Download Patch File Download Diff File Expand all files Collapse all files

102
test/tint/buffer/storage/dynamic_index/read.wgsl.expected.msl
View File

@@ -14,42 +14,108 @@ struct tint_array {
T elements[N];
};
struct Inner {
struct tint_packed_vec3_f32_array_element {
/* 0x0000 */ packed_float3 elements;
/* 0x000c */ tint_array<int8_t, 4> tint_pad;
};
struct Inner_tint_packed_vec3 {
/* 0x0000 */ float scalar_f32;
/* 0x0004 */ int scalar_i32;
/* 0x0008 */ uint scalar_u32;
/* 0x000c */ tint_array<int8_t, 4> tint_pad;
/* 0x000c */ tint_array<int8_t, 4> tint_pad_1;
/* 0x0010 */ float2 vec2_f32;
/* 0x0018 */ int2 vec2_i32;
/* 0x0020 */ uint2 vec2_u32;
/* 0x0028 */ tint_array<int8_t, 8> tint_pad_1;
/* 0x0028 */ tint_array<int8_t, 8> tint_pad_2;
/* 0x0030 */ packed_float3 vec3_f32;
/* 0x003c */ tint_array<int8_t, 4> tint_pad_2;
/* 0x003c */ tint_array<int8_t, 4> tint_pad_3;
/* 0x0040 */ packed_int3 vec3_i32;
/* 0x004c */ tint_array<int8_t, 4> tint_pad_3;
/* 0x004c */ tint_array<int8_t, 4> tint_pad_4;
/* 0x0050 */ packed_uint3 vec3_u32;
/* 0x005c */ tint_array<int8_t, 4> tint_pad_4;
/* 0x005c */ tint_array<int8_t, 4> tint_pad_5;
/* 0x0060 */ float4 vec4_f32;
/* 0x0070 */ int4 vec4_i32;
/* 0x0080 */ uint4 vec4_u32;
/* 0x0090 */ float2x2 mat2x2_f32;
/* 0x00a0 */ float2x3 mat2x3_f32;
/* 0x00a0 */ tint_array<tint_packed_vec3_f32_array_element, 2> mat2x3_f32;
/* 0x00c0 */ float2x4 mat2x4_f32;
/* 0x00e0 */ float3x2 mat3x2_f32;
/* 0x00f8 */ tint_array<int8_t, 8> tint_pad_5;
/* 0x0100 */ float3x3 mat3x3_f32;
/* 0x00f8 */ tint_array<int8_t, 8> tint_pad_6;
/* 0x0100 */ tint_array<tint_packed_vec3_f32_array_element, 3> mat3x3_f32;
/* 0x0130 */ float3x4 mat3x4_f32;
/* 0x0160 */ float4x2 mat4x2_f32;
/* 0x0180 */ float4x3 mat4x3_f32;
/* 0x0180 */ tint_array<tint_packed_vec3_f32_array_element, 4> mat4x3_f32;
/* 0x01c0 */ float4x4 mat4x4_f32;
/* 0x0200 */ tint_array<float3, 2> arr2_vec3_f32;
/* 0x0200 */ tint_array<tint_packed_vec3_f32_array_element, 2> arr2_vec3_f32;
};
struct S_tint_packed_vec3 {
/* 0x0000 */ tint_array<Inner_tint_packed_vec3, 1> arr;
};
float2x3 tint_unpack_vec3_in_composite(tint_array<tint_packed_vec3_f32_array_element, 2> in) {
float2x3 result = float2x3(0.0f);
for(uint i = 0u; (i < 2u); i = (i + 1u)) {
result[i] = float3(in[i].elements);
}
return result;
}
float3x3 tint_unpack_vec3_in_composite_1(tint_array<tint_packed_vec3_f32_array_element, 3> in) {
float3x3 result = float3x3(0.0f);
for(uint i = 0u; (i < 3u); i = (i + 1u)) {
result[i] = float3(in[i].elements);
}
return result;
}
float4x3 tint_unpack_vec3_in_composite_2(tint_array<tint_packed_vec3_f32_array_element, 4> in) {
float4x3 result = float4x3(0.0f);
for(uint i = 0u; (i < 4u); i = (i + 1u)) {
result[i] = float3(in[i].elements);
}
return result;
}
tint_array<float3, 2> tint_unpack_vec3_in_composite_3(tint_array<tint_packed_vec3_f32_array_element, 2> in) {
tint_array<float3, 2> result = {};
for(uint i = 0u; (i < 2u); i = (i + 1u)) {
result[i] = float3(in[i].elements);
}
return result;
}
struct Inner {
float scalar_f32;
int scalar_i32;
uint scalar_u32;
float2 vec2_f32;
int2 vec2_i32;
uint2 vec2_u32;
float3 vec3_f32;
int3 vec3_i32;
uint3 vec3_u32;
float4 vec4_f32;
int4 vec4_i32;
uint4 vec4_u32;
float2x2 mat2x2_f32;
float2x3 mat2x3_f32;
float2x4 mat2x4_f32;
float3x2 mat3x2_f32;
float3x3 mat3x3_f32;
float3x4 mat3x4_f32;
float4x2 mat4x2_f32;
float4x3 mat4x3_f32;
float4x4 mat4x4_f32;
tint_array<float3, 2> arr2_vec3_f32;
};
struct S {
/* 0x0000 */ tint_array<Inner, 1> arr;
tint_array<Inner, 1> arr;
};
void tint_symbol_inner(uint idx, const device S* const tint_symbol_1) {
void tint_symbol_inner(uint idx, const device S_tint_packed_vec3* const tint_symbol_1) {
float const scalar_f32 = (*(tint_symbol_1)).arr[idx].scalar_f32;
int const scalar_i32 = (*(tint_symbol_1)).arr[idx].scalar_i32;
uint const scalar_u32 = (*(tint_symbol_1)).arr[idx].scalar_u32;
@@ -63,18 +129,18 @@ void tint_symbol_inner(uint idx, const device S* const tint_symbol_1) {
int4 const vec4_i32 = (*(tint_symbol_1)).arr[idx].vec4_i32;
uint4 const vec4_u32 = (*(tint_symbol_1)).arr[idx].vec4_u32;
float2x2 const mat2x2_f32 = (*(tint_symbol_1)).arr[idx].mat2x2_f32;
float2x3 const mat2x3_f32 = (*(tint_symbol_1)).arr[idx].mat2x3_f32;
float2x3 const mat2x3_f32 = tint_unpack_vec3_in_composite((*(tint_symbol_1)).arr[idx].mat2x3_f32);
float2x4 const mat2x4_f32 = (*(tint_symbol_1)).arr[idx].mat2x4_f32;
float3x2 const mat3x2_f32 = (*(tint_symbol_1)).arr[idx].mat3x2_f32;
float3x3 const mat3x3_f32 = (*(tint_symbol_1)).arr[idx].mat3x3_f32;
float3x3 const mat3x3_f32 = tint_unpack_vec3_in_composite_1((*(tint_symbol_1)).arr[idx].mat3x3_f32);
float3x4 const mat3x4_f32 = (*(tint_symbol_1)).arr[idx].mat3x4_f32;
float4x2 const mat4x2_f32 = (*(tint_symbol_1)).arr[idx].mat4x2_f32;
float4x3 const mat4x3_f32 = (*(tint_symbol_1)).arr[idx].mat4x3_f32;
float4x3 const mat4x3_f32 = tint_unpack_vec3_in_composite_2((*(tint_symbol_1)).arr[idx].mat4x3_f32);
float4x4 const mat4x4_f32 = (*(tint_symbol_1)).arr[idx].mat4x4_f32;
tint_array<float3, 2> const arr2_vec3_f32 = (*(tint_symbol_1)).arr[idx].arr2_vec3_f32;
tint_array<float3, 2> const arr2_vec3_f32 = tint_unpack_vec3_in_composite_3((*(tint_symbol_1)).arr[idx].arr2_vec3_f32);
}
kernel void tint_symbol(const device S* tint_symbol_2 [[buffer(0)]], uint idx [[thread_index_in_threadgroup]]) {
kernel void tint_symbol(const device S_tint_packed_vec3* tint_symbol_2 [[buffer(0)]], uint idx [[thread_index_in_threadgroup]]) {
tint_symbol_inner(idx, tint_symbol_2);
return;
}

167
test/tint/buffer/storage/dynamic_index/read_f16.wgsl.expected.msl
View File

@@ -14,60 +14,169 @@ struct tint_array {
T elements[N];
};
struct Inner {
struct tint_packed_vec3_f32_array_element {
/* 0x0000 */ packed_float3 elements;
/* 0x000c */ tint_array<int8_t, 4> tint_pad;
};
struct tint_packed_vec3_f16_array_element {
/* 0x0000 */ packed_half3 elements;
/* 0x0006 */ tint_array<int8_t, 2> tint_pad_1;
};
struct Inner_tint_packed_vec3 {
/* 0x0000 */ float scalar_f32;
/* 0x0004 */ int scalar_i32;
/* 0x0008 */ uint scalar_u32;
/* 0x000c */ half scalar_f16;
/* 0x000e */ tint_array<int8_t, 2> tint_pad;
/* 0x000e */ tint_array<int8_t, 2> tint_pad_2;
/* 0x0010 */ float2 vec2_f32;
/* 0x0018 */ int2 vec2_i32;
/* 0x0020 */ uint2 vec2_u32;
/* 0x0028 */ half2 vec2_f16;
/* 0x002c */ tint_array<int8_t, 4> tint_pad_1;
/* 0x002c */ tint_array<int8_t, 4> tint_pad_3;
/* 0x0030 */ packed_float3 vec3_f32;
/* 0x003c */ tint_array<int8_t, 4> tint_pad_2;
/* 0x003c */ tint_array<int8_t, 4> tint_pad_4;
/* 0x0040 */ packed_int3 vec3_i32;
/* 0x004c */ tint_array<int8_t, 4> tint_pad_3;
/* 0x004c */ tint_array<int8_t, 4> tint_pad_5;
/* 0x0050 */ packed_uint3 vec3_u32;
/* 0x005c */ tint_array<int8_t, 4> tint_pad_4;
/* 0x005c */ tint_array<int8_t, 4> tint_pad_6;
/* 0x0060 */ packed_half3 vec3_f16;
/* 0x0066 */ tint_array<int8_t, 10> tint_pad_5;
/* 0x0066 */ tint_array<int8_t, 10> tint_pad_7;
/* 0x0070 */ float4 vec4_f32;
/* 0x0080 */ int4 vec4_i32;
/* 0x0090 */ uint4 vec4_u32;
/* 0x00a0 */ half4 vec4_f16;
/* 0x00a8 */ float2x2 mat2x2_f32;
/* 0x00b8 */ tint_array<int8_t, 8> tint_pad_6;
/* 0x00c0 */ float2x3 mat2x3_f32;
/* 0x00b8 */ tint_array<int8_t, 8> tint_pad_8;
/* 0x00c0 */ tint_array<tint_packed_vec3_f32_array_element, 2> mat2x3_f32;
/* 0x00e0 */ float2x4 mat2x4_f32;
/* 0x0100 */ float3x2 mat3x2_f32;
/* 0x0118 */ tint_array<int8_t, 8> tint_pad_7;
/* 0x0120 */ float3x3 mat3x3_f32;
/* 0x0118 */ tint_array<int8_t, 8> tint_pad_9;
/* 0x0120 */ tint_array<tint_packed_vec3_f32_array_element, 3> mat3x3_f32;
/* 0x0150 */ float3x4 mat3x4_f32;
/* 0x0180 */ float4x2 mat4x2_f32;
/* 0x01a0 */ float4x3 mat4x3_f32;
/* 0x01a0 */ tint_array<tint_packed_vec3_f32_array_element, 4> mat4x3_f32;
/* 0x01e0 */ float4x4 mat4x4_f32;
/* 0x0220 */ half2x2 mat2x2_f16;
/* 0x0228 */ half2x3 mat2x3_f16;
/* 0x0228 */ tint_array<tint_packed_vec3_f16_array_element, 2> mat2x3_f16;
/* 0x0238 */ half2x4 mat2x4_f16;
/* 0x0248 */ half3x2 mat3x2_f16;
/* 0x0254 */ tint_array<int8_t, 4> tint_pad_8;
/* 0x0258 */ half3x3 mat3x3_f16;
/* 0x0254 */ tint_array<int8_t, 4> tint_pad_10;
/* 0x0258 */ tint_array<tint_packed_vec3_f16_array_element, 3> mat3x3_f16;
/* 0x0270 */ half3x4 mat3x4_f16;
/* 0x0288 */ half4x2 mat4x2_f16;
/* 0x0298 */ half4x3 mat4x3_f16;
/* 0x0298 */ tint_array<tint_packed_vec3_f16_array_element, 4> mat4x3_f16;
/* 0x02b8 */ half4x4 mat4x4_f16;
/* 0x02d8 */ tint_array<int8_t, 8> tint_pad_9;
/* 0x02e0 */ tint_array<float3, 2> arr2_vec3_f32;
/* 0x02d8 */ tint_array<int8_t, 8> tint_pad_11;
/* 0x02e0 */ tint_array<tint_packed_vec3_f32_array_element, 2> arr2_vec3_f32;
/* 0x0300 */ tint_array<half4x2, 2> arr2_mat4x2_f16;
};
struct S {
/* 0x0000 */ tint_array<Inner, 1> arr;
struct S_tint_packed_vec3 {
/* 0x0000 */ tint_array<Inner_tint_packed_vec3, 1> arr;
};
void tint_symbol_inner(uint idx, const device S* const tint_symbol_1) {
float2x3 tint_unpack_vec3_in_composite(tint_array<tint_packed_vec3_f32_array_element, 2> in) {
float2x3 result = float2x3(0.0f);
for(uint i = 0u; (i < 2u); i = (i + 1u)) {
result[i] = float3(in[i].elements);
}
return result;
}
float3x3 tint_unpack_vec3_in_composite_1(tint_array<tint_packed_vec3_f32_array_element, 3> in) {
float3x3 result = float3x3(0.0f);
for(uint i = 0u; (i < 3u); i = (i + 1u)) {
result[i] = float3(in[i].elements);
}
return result;
}
float4x3 tint_unpack_vec3_in_composite_2(tint_array<tint_packed_vec3_f32_array_element, 4> in) {
float4x3 result = float4x3(0.0f);
for(uint i = 0u; (i < 4u); i = (i + 1u)) {
result[i] = float3(in[i].elements);
}
return result;
}
half2x3 tint_unpack_vec3_in_composite_3(tint_array<tint_packed_vec3_f16_array_element, 2> in) {
half2x3 result = half2x3(0.0h);
for(uint i = 0u; (i < 2u); i = (i + 1u)) {
result[i] = half3(in[i].elements);
}
return result;
}
half3x3 tint_unpack_vec3_in_composite_4(tint_array<tint_packed_vec3_f16_array_element, 3> in) {
half3x3 result = half3x3(0.0h);
for(uint i = 0u; (i < 3u); i = (i + 1u)) {
result[i] = half3(in[i].elements);
}
return result;
}
half4x3 tint_unpack_vec3_in_composite_5(tint_array<tint_packed_vec3_f16_array_element, 4> in) {
half4x3 result = half4x3(0.0h);
for(uint i = 0u; (i < 4u); i = (i + 1u)) {
result[i] = half3(in[i].elements);
}
return result;
}
tint_array<float3, 2> tint_unpack_vec3_in_composite_6(tint_array<tint_packed_vec3_f32_array_element, 2> in) {
tint_array<float3, 2> result = {};
for(uint i = 0u; (i < 2u); i = (i + 1u)) {
result[i] = float3(in[i].elements);
}
return result;
}
struct Inner {
float scalar_f32;
int scalar_i32;
uint scalar_u32;
half scalar_f16;
float2 vec2_f32;
int2 vec2_i32;
uint2 vec2_u32;
half2 vec2_f16;
float3 vec3_f32;
int3 vec3_i32;
uint3 vec3_u32;
half3 vec3_f16;
float4 vec4_f32;
int4 vec4_i32;
uint4 vec4_u32;
half4 vec4_f16;
float2x2 mat2x2_f32;
float2x3 mat2x3_f32;
float2x4 mat2x4_f32;
float3x2 mat3x2_f32;
float3x3 mat3x3_f32;
float3x4 mat3x4_f32;
float4x2 mat4x2_f32;
float4x3 mat4x3_f32;
float4x4 mat4x4_f32;
half2x2 mat2x2_f16;
half2x3 mat2x3_f16;
half2x4 mat2x4_f16;
half3x2 mat3x2_f16;
half3x3 mat3x3_f16;
half3x4 mat3x4_f16;
half4x2 mat4x2_f16;
half4x3 mat4x3_f16;
half4x4 mat4x4_f16;
tint_array<float3, 2> arr2_vec3_f32;
tint_array<half4x2, 2> arr2_mat4x2_f16;
};
struct S {
tint_array<Inner, 1> arr;
};
void tint_symbol_inner(uint idx, const device S_tint_packed_vec3* const tint_symbol_1) {
float const scalar_f32 = (*(tint_symbol_1)).arr[idx].scalar_f32;
int const scalar_i32 = (*(tint_symbol_1)).arr[idx].scalar_i32;
uint const scalar_u32 = (*(tint_symbol_1)).arr[idx].scalar_u32;
@@ -85,28 +194,28 @@ void tint_symbol_inner(uint idx, const device S* const tint_symbol_1) {
uint4 const vec4_u32 = (*(tint_symbol_1)).arr[idx].vec4_u32;
half4 const vec4_f16 = (*(tint_symbol_1)).arr[idx].vec4_f16;
float2x2 const mat2x2_f32 = (*(tint_symbol_1)).arr[idx].mat2x2_f32;
float2x3 const mat2x3_f32 = (*(tint_symbol_1)).arr[idx].mat2x3_f32;
float2x3 const mat2x3_f32 = tint_unpack_vec3_in_composite((*(tint_symbol_1)).arr[idx].mat2x3_f32);
float2x4 const mat2x4_f32 = (*(tint_symbol_1)).arr[idx].mat2x4_f32;
float3x2 const mat3x2_f32 = (*(tint_symbol_1)).arr[idx].mat3x2_f32;
float3x3 const mat3x3_f32 = (*(tint_symbol_1)).arr[idx].mat3x3_f32;
float3x3 const mat3x3_f32 = tint_unpack_vec3_in_composite_1((*(tint_symbol_1)).arr[idx].mat3x3_f32);
float3x4 const mat3x4_f32 = (*(tint_symbol_1)).arr[idx].mat3x4_f32;
float4x2 const mat4x2_f32 = (*(tint_symbol_1)).arr[idx].mat4x2_f32;
float4x3 const mat4x3_f32 = (*(tint_symbol_1)).arr[idx].mat4x3_f32;
float4x3 const mat4x3_f32 = tint_unpack_vec3_in_composite_2((*(tint_symbol_1)).arr[idx].mat4x3_f32);
float4x4 const mat4x4_f32 = (*(tint_symbol_1)).arr[idx].mat4x4_f32;
half2x2 const mat2x2_f16 = (*(tint_symbol_1)).arr[idx].mat2x2_f16;
half2x3 const mat2x3_f16 = (*(tint_symbol_1)).arr[idx].mat2x3_f16;
half2x3 const mat2x3_f16 = tint_unpack_vec3_in_composite_3((*(tint_symbol_1)).arr[idx].mat2x3_f16);
half2x4 const mat2x4_f16 = (*(tint_symbol_1)).arr[idx].mat2x4_f16;
half3x2 const mat3x2_f16 = (*(tint_symbol_1)).arr[idx].mat3x2_f16;
half3x3 const mat3x3_f16 = (*(tint_symbol_1)).arr[idx].mat3x3_f16;
half3x3 const mat3x3_f16 = tint_unpack_vec3_in_composite_4((*(tint_symbol_1)).arr[idx].mat3x3_f16);
half3x4 const mat3x4_f16 = (*(tint_symbol_1)).arr[idx].mat3x4_f16;
half4x2 const mat4x2_f16 = (*(tint_symbol_1)).arr[idx].mat4x2_f16;
half4x3 const mat4x3_f16 = (*(tint_symbol_1)).arr[idx].mat4x3_f16;
half4x3 const mat4x3_f16 = tint_unpack_vec3_in_composite_5((*(tint_symbol_1)).arr[idx].mat4x3_f16);
half4x4 const mat4x4_f16 = (*(tint_symbol_1)).arr[idx].mat4x4_f16;
tint_array<float3, 2> const arr2_vec3_f32 = (*(tint_symbol_1)).arr[idx].arr2_vec3_f32;
tint_array<float3, 2> const arr2_vec3_f32 = tint_unpack_vec3_in_composite_6((*(tint_symbol_1)).arr[idx].arr2_vec3_f32);
tint_array<half4x2, 2> const arr2_mat4x2_f16 = (*(tint_symbol_1)).arr[idx].arr2_mat4x2_f16;
}
kernel void tint_symbol(const device S* tint_symbol_2 [[buffer(0)]], uint idx [[thread_index_in_threadgroup]]) {
kernel void tint_symbol(const device S_tint_packed_vec3* tint_symbol_2 [[buffer(0)]], uint idx [[thread_index_in_threadgroup]]) {
tint_symbol_inner(idx, tint_symbol_2);
return;
}

96
test/tint/buffer/storage/dynamic_index/write.wgsl.expected.msl
View File

@@ -14,75 +14,109 @@ struct tint_array {
T elements[N];
};
struct Inner {
struct tint_packed_vec3_f32_array_element {
/* 0x0000 */ packed_float3 elements;
/* 0x000c */ tint_array<int8_t, 4> tint_pad;
};
struct Inner_tint_packed_vec3 {
/* 0x0000 */ float scalar_f32;
/* 0x0004 */ int scalar_i32;
/* 0x0008 */ uint scalar_u32;
/* 0x000c */ tint_array<int8_t, 4> tint_pad;
/* 0x000c */ tint_array<int8_t, 4> tint_pad_1;
/* 0x0010 */ float2 vec2_f32;
/* 0x0018 */ int2 vec2_i32;
/* 0x0020 */ uint2 vec2_u32;
/* 0x0028 */ tint_array<int8_t, 8> tint_pad_1;
/* 0x0028 */ tint_array<int8_t, 8> tint_pad_2;
/* 0x0030 */ packed_float3 vec3_f32;
/* 0x003c */ tint_array<int8_t, 4> tint_pad_2;
/* 0x003c */ tint_array<int8_t, 4> tint_pad_3;
/* 0x0040 */ packed_int3 vec3_i32;
/* 0x004c */ tint_array<int8_t, 4> tint_pad_3;
/* 0x004c */ tint_array<int8_t, 4> tint_pad_4;
/* 0x0050 */ packed_uint3 vec3_u32;
/* 0x005c */ tint_array<int8_t, 4> tint_pad_4;
/* 0x005c */ tint_array<int8_t, 4> tint_pad_5;
/* 0x0060 */ float4 vec4_f32;
/* 0x0070 */ int4 vec4_i32;
/* 0x0080 */ uint4 vec4_u32;
/* 0x0090 */ float2x2 mat2x2_f32;
/* 0x00a0 */ float2x3 mat2x3_f32;
/* 0x00a0 */ tint_array<tint_packed_vec3_f32_array_element, 2> mat2x3_f32;
/* 0x00c0 */ float2x4 mat2x4_f32;
/* 0x00e0 */ float3x2 mat3x2_f32;
/* 0x00f8 */ tint_array<int8_t, 8> tint_pad_5;
/* 0x0100 */ float3x3 mat3x3_f32;
/* 0x00f8 */ tint_array<int8_t, 8> tint_pad_6;
/* 0x0100 */ tint_array<tint_packed_vec3_f32_array_element, 3> mat3x3_f32;
/* 0x0130 */ float3x4 mat3x4_f32;
/* 0x0160 */ float4x2 mat4x2_f32;
/* 0x0180 */ float4x3 mat4x3_f32;
/* 0x0180 */ tint_array<tint_packed_vec3_f32_array_element, 4> mat4x3_f32;
/* 0x01c0 */ float4x4 mat4x4_f32;
/* 0x0200 */ tint_array<float3, 2> arr2_vec3_f32;
/* 0x0200 */ tint_array<tint_packed_vec3_f32_array_element, 2> arr2_vec3_f32;
};
struct S_tint_packed_vec3 {
/* 0x0000 */ tint_array<Inner_tint_packed_vec3, 1> arr;
};
struct Inner {
float scalar_f32;
int scalar_i32;
uint scalar_u32;
float2 vec2_f32;
int2 vec2_i32;
uint2 vec2_u32;
float3 vec3_f32;
int3 vec3_i32;
uint3 vec3_u32;
float4 vec4_f32;
int4 vec4_i32;
uint4 vec4_u32;
float2x2 mat2x2_f32;
float2x3 mat2x3_f32;
float2x4 mat2x4_f32;
float3x2 mat3x2_f32;
float3x3 mat3x3_f32;
float3x4 mat3x4_f32;
float4x2 mat4x2_f32;
float4x3 mat4x3_f32;
float4x4 mat4x4_f32;
tint_array<float3, 2> arr2_vec3_f32;
};
struct S {
/* 0x0000 */ tint_array<Inner, 1> arr;
tint_array<Inner, 1> arr;
};
void assign_and_preserve_padding(device float2x3* const dest, float2x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
void assign_and_preserve_padding(device tint_array<tint_packed_vec3_f32_array_element, 2>* const dest, float2x3 value) {
(*(dest))[0u].elements = packed_float3(value[0u]);
(*(dest))[1u].elements = packed_float3(value[1u]);
}
void assign_and_preserve_padding_1(device float3x3* const dest, float3x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
(*(dest))[2u] = value[2u];
void assign_and_preserve_padding_1(device tint_array<tint_packed_vec3_f32_array_element, 3>* const dest, float3x3 value) {
(*(dest))[0u].elements = packed_float3(value[0u]);
(*(dest))[1u].elements = packed_float3(value[1u]);
(*(dest))[2u].elements = packed_float3(value[2u]);
}
void assign_and_preserve_padding_2(device float4x3* const dest, float4x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
(*(dest))[2u] = value[2u];
(*(dest))[3u] = value[3u];
void assign_and_preserve_padding_2(device tint_array<tint_packed_vec3_f32_array_element, 4>* const dest, float4x3 value) {
(*(dest))[0u].elements = packed_float3(value[0u]);
(*(dest))[1u].elements = packed_float3(value[1u]);
(*(dest))[2u].elements = packed_float3(value[2u]);
(*(dest))[3u].elements = packed_float3(value[3u]);
}
void assign_and_preserve_padding_3(device tint_array<float3, 2>* const dest, tint_array<float3, 2> value) {
void assign_and_preserve_padding_3(device tint_array<tint_packed_vec3_f32_array_element, 2>* const dest, tint_array<float3, 2> value) {
for(uint i = 0u; (i < 2u); i = (i + 1u)) {
(*(dest))[i] = value[i];
(*(dest))[i].elements = packed_float3(value[i]);
}
}
void tint_symbol_inner(uint idx, device S* const tint_symbol_2) {
void tint_symbol_inner(uint idx, device S_tint_packed_vec3* const tint_symbol_2) {
(*(tint_symbol_2)).arr[idx].scalar_f32 = 0.0f;
(*(tint_symbol_2)).arr[idx].scalar_i32 = 0;
(*(tint_symbol_2)).arr[idx].scalar_u32 = 0u;
(*(tint_symbol_2)).arr[idx].vec2_f32 = float2(0.0f);
(*(tint_symbol_2)).arr[idx].vec2_i32 = int2(0);
(*(tint_symbol_2)).arr[idx].vec2_u32 = uint2(0u);
(*(tint_symbol_2)).arr[idx].vec3_f32 = float3(0.0f);
(*(tint_symbol_2)).arr[idx].vec3_i32 = int3(0);
(*(tint_symbol_2)).arr[idx].vec3_u32 = uint3(0u);
(*(tint_symbol_2)).arr[idx].vec3_f32 = packed_float3(0.0f);
(*(tint_symbol_2)).arr[idx].vec3_i32 = packed_int3(0);
(*(tint_symbol_2)).arr[idx].vec3_u32 = packed_uint3(0u);
(*(tint_symbol_2)).arr[idx].vec4_f32 = float4(0.0f);
(*(tint_symbol_2)).arr[idx].vec4_i32 = int4(0);
(*(tint_symbol_2)).arr[idx].vec4_u32 = uint4(0u);
@@ -99,7 +133,7 @@ void tint_symbol_inner(uint idx, device S* const tint_symbol_2) {
assign_and_preserve_padding_3(&((*(tint_symbol_2)).arr[idx].arr2_vec3_f32), tint_symbol_1);
}
kernel void tint_symbol(device S* tint_symbol_3 [[buffer(0)]], uint idx [[thread_index_in_threadgroup]]) {
kernel void tint_symbol(device S_tint_packed_vec3* tint_symbol_3 [[buffer(0)]], uint idx [[thread_index_in_threadgroup]]) {
tint_symbol_inner(idx, tint_symbol_3);
return;
}

157
test/tint/buffer/storage/dynamic_index/write_f16.wgsl.expected.msl
View File

@@ -14,102 +14,155 @@ struct tint_array {
T elements[N];
};
struct Inner {
struct tint_packed_vec3_f32_array_element {
/* 0x0000 */ packed_float3 elements;
/* 0x000c */ tint_array<int8_t, 4> tint_pad;
};
struct tint_packed_vec3_f16_array_element {
/* 0x0000 */ packed_half3 elements;
/* 0x0006 */ tint_array<int8_t, 2> tint_pad_1;
};
struct Inner_tint_packed_vec3 {
/* 0x0000 */ float scalar_f32;
/* 0x0004 */ int scalar_i32;
/* 0x0008 */ uint scalar_u32;
/* 0x000c */ half scalar_f16;
/* 0x000e */ tint_array<int8_t, 2> tint_pad;
/* 0x000e */ tint_array<int8_t, 2> tint_pad_2;
/* 0x0010 */ float2 vec2_f32;
/* 0x0018 */ int2 vec2_i32;
/* 0x0020 */ uint2 vec2_u32;
/* 0x0028 */ half2 vec2_f16;
/* 0x002c */ tint_array<int8_t, 4> tint_pad_1;
/* 0x002c */ tint_array<int8_t, 4> tint_pad_3;
/* 0x0030 */ packed_float3 vec3_f32;
/* 0x003c */ tint_array<int8_t, 4> tint_pad_2;
/* 0x003c */ tint_array<int8_t, 4> tint_pad_4;
/* 0x0040 */ packed_int3 vec3_i32;
/* 0x004c */ tint_array<int8_t, 4> tint_pad_3;
/* 0x004c */ tint_array<int8_t, 4> tint_pad_5;
/* 0x0050 */ packed_uint3 vec3_u32;
/* 0x005c */ tint_array<int8_t, 4> tint_pad_4;
/* 0x005c */ tint_array<int8_t, 4> tint_pad_6;
/* 0x0060 */ packed_half3 vec3_f16;
/* 0x0066 */ tint_array<int8_t, 10> tint_pad_5;
/* 0x0066 */ tint_array<int8_t, 10> tint_pad_7;
/* 0x0070 */ float4 vec4_f32;
/* 0x0080 */ int4 vec4_i32;
/* 0x0090 */ uint4 vec4_u32;
/* 0x00a0 */ half4 vec4_f16;
/* 0x00a8 */ float2x2 mat2x2_f32;
/* 0x00b8 */ tint_array<int8_t, 8> tint_pad_6;
/* 0x00c0 */ float2x3 mat2x3_f32;
/* 0x00b8 */ tint_array<int8_t, 8> tint_pad_8;
/* 0x00c0 */ tint_array<tint_packed_vec3_f32_array_element, 2> mat2x3_f32;
/* 0x00e0 */ float2x4 mat2x4_f32;
/* 0x0100 */ float3x2 mat3x2_f32;
/* 0x0118 */ tint_array<int8_t, 8> tint_pad_7;
/* 0x0120 */ float3x3 mat3x3_f32;
/* 0x0118 */ tint_array<int8_t, 8> tint_pad_9;
/* 0x0120 */ tint_array<tint_packed_vec3_f32_array_element, 3> mat3x3_f32;
/* 0x0150 */ float3x4 mat3x4_f32;
/* 0x0180 */ float4x2 mat4x2_f32;
/* 0x01a0 */ float4x3 mat4x3_f32;
/* 0x01a0 */ tint_array<tint_packed_vec3_f32_array_element, 4> mat4x3_f32;
/* 0x01e0 */ float4x4 mat4x4_f32;
/* 0x0220 */ half2x2 mat2x2_f16;
/* 0x0228 */ half2x3 mat2x3_f16;
/* 0x0228 */ tint_array<tint_packed_vec3_f16_array_element, 2> mat2x3_f16;
/* 0x0238 */ half2x4 mat2x4_f16;
/* 0x0248 */ half3x2 mat3x2_f16;
/* 0x0254 */ tint_array<int8_t, 4> tint_pad_8;
/* 0x0258 */ half3x3 mat3x3_f16;
/* 0x0254 */ tint_array<int8_t, 4> tint_pad_10;
/* 0x0258 */ tint_array<tint_packed_vec3_f16_array_element, 3> mat3x3_f16;
/* 0x0270 */ half3x4 mat3x4_f16;
/* 0x0288 */ half4x2 mat4x2_f16;
/* 0x0298 */ half4x3 mat4x3_f16;
/* 0x0298 */ tint_array<tint_packed_vec3_f16_array_element, 4> mat4x3_f16;
/* 0x02b8 */ half4x4 mat4x4_f16;
/* 0x02d8 */ tint_array<int8_t, 8> tint_pad_9;
/* 0x02e0 */ tint_array<float3, 2> arr2_vec3_f32;
/* 0x02d8 */ tint_array<int8_t, 8> tint_pad_11;
/* 0x02e0 */ tint_array<tint_packed_vec3_f32_array_element, 2> arr2_vec3_f32;
/* 0x0300 */ tint_array<half4x2, 2> arr2_mat4x2_f16;
};
struct S {
/* 0x0000 */ tint_array<Inner, 1> arr;
struct S_tint_packed_vec3 {
/* 0x0000 */ tint_array<Inner_tint_packed_vec3, 1> arr;
};
void assign_and_preserve_padding(device float2x3* const dest, float2x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
struct Inner {
float scalar_f32;
int scalar_i32;
uint scalar_u32;
half scalar_f16;
float2 vec2_f32;
int2 vec2_i32;
uint2 vec2_u32;
half2 vec2_f16;
float3 vec3_f32;
int3 vec3_i32;
uint3 vec3_u32;
half3 vec3_f16;
float4 vec4_f32;
int4 vec4_i32;
uint4 vec4_u32;
half4 vec4_f16;
float2x2 mat2x2_f32;
float2x3 mat2x3_f32;
float2x4 mat2x4_f32;
float3x2 mat3x2_f32;
float3x3 mat3x3_f32;
float3x4 mat3x4_f32;
float4x2 mat4x2_f32;
float4x3 mat4x3_f32;
float4x4 mat4x4_f32;
half2x2 mat2x2_f16;
half2x3 mat2x3_f16;
half2x4 mat2x4_f16;
half3x2 mat3x2_f16;
half3x3 mat3x3_f16;
half3x4 mat3x4_f16;
half4x2 mat4x2_f16;
half4x3 mat4x3_f16;
half4x4 mat4x4_f16;
tint_array<float3, 2> arr2_vec3_f32;
tint_array<half4x2, 2> arr2_mat4x2_f16;
};
struct S {
tint_array<Inner, 1> arr;
};
void assign_and_preserve_padding(device tint_array<tint_packed_vec3_f32_array_element, 2>* const dest, float2x3 value) {
(*(dest))[0u].elements = packed_float3(value[0u]);
(*(dest))[1u].elements = packed_float3(value[1u]);
}
void assign_and_preserve_padding_1(device float3x3* const dest, float3x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
(*(dest))[2u] = value[2u];
void assign_and_preserve_padding_1(device tint_array<tint_packed_vec3_f32_array_element, 3>* const dest, float3x3 value) {
(*(dest))[0u].elements = packed_float3(value[0u]);
(*(dest))[1u].elements = packed_float3(value[1u]);
(*(dest))[2u].elements = packed_float3(value[2u]);
}
void assign_and_preserve_padding_2(device float4x3* const dest, float4x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
(*(dest))[2u] = value[2u];
(*(dest))[3u] = value[3u];
void assign_and_preserve_padding_2(device tint_array<tint_packed_vec3_f32_array_element, 4>* const dest, float4x3 value) {
(*(dest))[0u].elements = packed_float3(value[0u]);
(*(dest))[1u].elements = packed_float3(value[1u]);
(*(dest))[2u].elements = packed_float3(value[2u]);
(*(dest))[3u].elements = packed_float3(value[3u]);
}
void assign_and_preserve_padding_3(device half2x3* const dest, half2x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
void assign_and_preserve_padding_3(device tint_array<tint_packed_vec3_f16_array_element, 2>* const dest, half2x3 value) {
(*(dest))[0u].elements = packed_half3(value[0u]);
(*(dest))[1u].elements = packed_half3(value[1u]);
}
void assign_and_preserve_padding_4(device half3x3* const dest, half3x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
(*(dest))[2u] = value[2u];
void assign_and_preserve_padding_4(device tint_array<tint_packed_vec3_f16_array_element, 3>* const dest, half3x3 value) {
(*(dest))[0u].elements = packed_half3(value[0u]);
(*(dest))[1u].elements = packed_half3(value[1u]);
(*(dest))[2u].elements = packed_half3(value[2u]);
}
void assign_and_preserve_padding_5(device half4x3* const dest, half4x3 value) {
(*(dest))[0u] = value[0u];
(*(dest))[1u] = value[1u];
(*(dest))[2u] = value[2u];
(*(dest))[3u] = value[3u];
void assign_and_preserve_padding_5(device tint_array<tint_packed_vec3_f16_array_element, 4>* const dest, half4x3 value) {
(*(dest))[0u].elements = packed_half3(value[0u]);
(*(dest))[1u].elements = packed_half3(value[1u]);
(*(dest))[2u].elements = packed_half3(value[2u]);
(*(dest))[3u].elements = packed_half3(value[3u]);
}
void assign_and_preserve_padding_6(device tint_array<float3, 2>* const dest, tint_array<float3, 2> value) {
void assign_and_preserve_padding_6(device tint_array<tint_packed_vec3_f32_array_element, 2>* const dest, tint_array<float3, 2> value) {
for(uint i = 0u; (i < 2u); i = (i + 1u)) {
(*(dest))[i] = value[i];
(*(dest))[i].elements = packed_float3(value[i]);
}
}
void tint_symbol_inner(uint idx, device S* const tint_symbol_3) {
void tint_symbol_inner(uint idx, device S_tint_packed_vec3* const tint_symbol_3) {
(*(tint_symbol_3)).arr[idx].scalar_f32 = 0.0f;
(*(tint_symbol_3)).arr[idx].scalar_i32 = 0;
(*(tint_symbol_3)).arr[idx].scalar_u32 = 0u;
@@ -118,10 +171,10 @@ void tint_symbol_inner(uint idx, device S* const tint_symbol_3) {
(*(tint_symbol_3)).arr[idx].vec2_i32 = int2(0);
(*(tint_symbol_3)).arr[idx].vec2_u32 = uint2(0u);
(*(tint_symbol_3)).arr[idx].vec2_f16 = half2(0.0h);
(*(tint_symbol_3)).arr[idx].vec3_f32 = float3(0.0f);
(*(tint_symbol_3)).arr[idx].vec3_i32 = int3(0);
(*(tint_symbol_3)).arr[idx].vec3_u32 = uint3(0u);
(*(tint_symbol_3)).arr[idx].vec3_f16 = half3(0.0h);
(*(tint_symbol_3)).arr[idx].vec3_f32 = packed_float3(0.0f);
(*(tint_symbol_3)).arr[idx].vec3_i32 = packed_int3(0);
(*(tint_symbol_3)).arr[idx].vec3_u32 = packed_uint3(0u);
(*(tint_symbol_3)).arr[idx].vec3_f16 = packed_half3(0.0h);
(*(tint_symbol_3)).arr[idx].vec4_f32 = float4(0.0f);
(*(tint_symbol_3)).arr[idx].vec4_i32 = int4(0);
(*(tint_symbol_3)).arr[idx].vec4_u32 = uint4(0u);
@@ -150,7 +203,7 @@ void tint_symbol_inner(uint idx, device S* const tint_symbol_3) {
(*(tint_symbol_3)).arr[idx].arr2_mat4x2_f16 = tint_symbol_2;
}
kernel void tint_symbol(device S* tint_symbol_4 [[buffer(0)]], uint idx [[thread_index_in_threadgroup]]) {
kernel void tint_symbol(device S_tint_packed_vec3* tint_symbol_4 [[buffer(0)]], uint idx [[thread_index_in_threadgroup]]) {
tint_symbol_inner(idx, tint_symbol_4);
return;
}