| #include <metal_stdlib> |
| using namespace metal; |
| |
| template<typename T, size_t N> |
| struct tint_array { |
| const constant T& operator[](size_t i) const constant { return elements[i]; } |
| device T& operator[](size_t i) device { return elements[i]; } |
| const device T& operator[](size_t i) const device { return elements[i]; } |
| thread T& operator[](size_t i) thread { return elements[i]; } |
| const thread T& operator[](size_t i) const thread { return elements[i]; } |
| threadgroup T& operator[](size_t i) threadgroup { return elements[i]; } |
| const threadgroup T& operator[](size_t i) const threadgroup { return elements[i]; } |
| T elements[N]; |
| }; |
| |
| struct Inner { |
| /* 0x0000 */ float scalar_f32; |
| /* 0x0004 */ int scalar_i32; |
| /* 0x0008 */ uint scalar_u32; |
| /* 0x000c */ half scalar_f16; |
| /* 0x000e */ tint_array<int8_t, 2> tint_pad; |
| /* 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; |
| /* 0x0030 */ float3 vec3_f32; |
| /* 0x003c */ tint_array<int8_t, 4> tint_pad_2; |
| /* 0x0040 */ int3 vec3_i32; |
| /* 0x004c */ tint_array<int8_t, 4> tint_pad_3; |
| /* 0x0050 */ uint3 vec3_u32; |
| /* 0x005c */ tint_array<int8_t, 4> tint_pad_4; |
| /* 0x0060 */ half3 vec3_f16; |
| /* 0x0066 */ tint_array<int8_t, 10> tint_pad_5; |
| /* 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; |
| /* 0x00e0 */ float2x4 mat2x4_f32; |
| /* 0x0100 */ float3x2 mat3x2_f32; |
| /* 0x0118 */ tint_array<int8_t, 8> tint_pad_7; |
| /* 0x0120 */ float3x3 mat3x3_f32; |
| /* 0x0150 */ float3x4 mat3x4_f32; |
| /* 0x0180 */ float4x2 mat4x2_f32; |
| /* 0x01a0 */ float4x3 mat4x3_f32; |
| /* 0x01e0 */ float4x4 mat4x4_f32; |
| /* 0x0220 */ half2x2 mat2x2_f16; |
| /* 0x0228 */ half2x3 mat2x3_f16; |
| /* 0x0238 */ half2x4 mat2x4_f16; |
| /* 0x0248 */ half3x2 mat3x2_f16; |
| /* 0x0254 */ tint_array<int8_t, 4> tint_pad_8; |
| /* 0x0258 */ half3x3 mat3x3_f16; |
| /* 0x0270 */ half3x4 mat3x4_f16; |
| /* 0x0288 */ half4x2 mat4x2_f16; |
| /* 0x0298 */ half4x3 mat4x3_f16; |
| /* 0x02b8 */ half4x4 mat4x4_f16; |
| /* 0x02d8 */ tint_array<int8_t, 8> tint_pad_9; |
| /* 0x02e0 */ tint_array<float3, 2> arr2_vec3_f32; |
| /* 0x0300 */ tint_array<half4x2, 2> arr2_mat4x2_f16; |
| }; |
| |
| struct S { |
| /* 0x0000 */ tint_array<Inner, 1> arr; |
| }; |
| |
| struct tint_module_vars_struct { |
| const device S* sb; |
| device int* s; |
| }; |
| |
| int tint_f32_to_i32(float value) { |
| return select(2147483647, select((-2147483647 - 1), int(value), (value >= -2147483648.0f)), (value <= 2147483520.0f)); |
| } |
| |
| int tint_f16_to_i32(half value) { |
| return select(2147483647, select((-2147483647 - 1), int(value), (value >= -65504.0h)), (value <= 65504.0h)); |
| } |
| |
| void tint_symbol_inner(uint idx, tint_module_vars_struct tint_module_vars) { |
| float const scalar_f32 = (*tint_module_vars.sb).arr[idx].scalar_f32; |
| int const scalar_i32 = (*tint_module_vars.sb).arr[idx].scalar_i32; |
| uint const scalar_u32 = (*tint_module_vars.sb).arr[idx].scalar_u32; |
| half const scalar_f16 = (*tint_module_vars.sb).arr[idx].scalar_f16; |
| float2 const vec2_f32 = (*tint_module_vars.sb).arr[idx].vec2_f32; |
| int2 const vec2_i32 = (*tint_module_vars.sb).arr[idx].vec2_i32; |
| uint2 const vec2_u32 = (*tint_module_vars.sb).arr[idx].vec2_u32; |
| half2 const vec2_f16 = (*tint_module_vars.sb).arr[idx].vec2_f16; |
| float3 const vec3_f32 = (*tint_module_vars.sb).arr[idx].vec3_f32; |
| int3 const vec3_i32 = (*tint_module_vars.sb).arr[idx].vec3_i32; |
| uint3 const vec3_u32 = (*tint_module_vars.sb).arr[idx].vec3_u32; |
| half3 const vec3_f16 = (*tint_module_vars.sb).arr[idx].vec3_f16; |
| float4 const vec4_f32 = (*tint_module_vars.sb).arr[idx].vec4_f32; |
| int4 const vec4_i32 = (*tint_module_vars.sb).arr[idx].vec4_i32; |
| uint4 const vec4_u32 = (*tint_module_vars.sb).arr[idx].vec4_u32; |
| half4 const vec4_f16 = (*tint_module_vars.sb).arr[idx].vec4_f16; |
| float2x2 const mat2x2_f32 = (*tint_module_vars.sb).arr[idx].mat2x2_f32; |
| float2x3 const mat2x3_f32 = (*tint_module_vars.sb).arr[idx].mat2x3_f32; |
| float2x4 const mat2x4_f32 = (*tint_module_vars.sb).arr[idx].mat2x4_f32; |
| float3x2 const mat3x2_f32 = (*tint_module_vars.sb).arr[idx].mat3x2_f32; |
| float3x3 const mat3x3_f32 = (*tint_module_vars.sb).arr[idx].mat3x3_f32; |
| float3x4 const mat3x4_f32 = (*tint_module_vars.sb).arr[idx].mat3x4_f32; |
| float4x2 const mat4x2_f32 = (*tint_module_vars.sb).arr[idx].mat4x2_f32; |
| float4x3 const mat4x3_f32 = (*tint_module_vars.sb).arr[idx].mat4x3_f32; |
| float4x4 const mat4x4_f32 = (*tint_module_vars.sb).arr[idx].mat4x4_f32; |
| half2x2 const mat2x2_f16 = (*tint_module_vars.sb).arr[idx].mat2x2_f16; |
| half2x3 const mat2x3_f16 = (*tint_module_vars.sb).arr[idx].mat2x3_f16; |
| half2x4 const mat2x4_f16 = (*tint_module_vars.sb).arr[idx].mat2x4_f16; |
| half3x2 const mat3x2_f16 = (*tint_module_vars.sb).arr[idx].mat3x2_f16; |
| half3x3 const mat3x3_f16 = (*tint_module_vars.sb).arr[idx].mat3x3_f16; |
| half3x4 const mat3x4_f16 = (*tint_module_vars.sb).arr[idx].mat3x4_f16; |
| half4x2 const mat4x2_f16 = (*tint_module_vars.sb).arr[idx].mat4x2_f16; |
| half4x3 const mat4x3_f16 = (*tint_module_vars.sb).arr[idx].mat4x3_f16; |
| half4x4 const mat4x4_f16 = (*tint_module_vars.sb).arr[idx].mat4x4_f16; |
| tint_array<float3, 2> const arr2_vec3_f32 = (*tint_module_vars.sb).arr[idx].arr2_vec3_f32; |
| tint_array<half4x2, 2> const arr2_mat4x2_f16 = (*tint_module_vars.sb).arr[idx].arr2_mat4x2_f16; |
| int const v = (tint_f32_to_i32(scalar_f32) + scalar_i32); |
| int const v_1 = (v + int(scalar_u32)); |
| int const v_2 = (v_1 + tint_f16_to_i32(scalar_f16)); |
| int const v_3 = ((v_2 + tint_f32_to_i32(vec2_f32[0u])) + vec2_i32[0u]); |
| int const v_4 = (v_3 + int(vec2_u32[0u])); |
| int const v_5 = (v_4 + tint_f16_to_i32(vec2_f16[0u])); |
| int const v_6 = ((v_5 + tint_f32_to_i32(vec3_f32[1u])) + vec3_i32[1u]); |
| int const v_7 = (v_6 + int(vec3_u32[1u])); |
| int const v_8 = (v_7 + tint_f16_to_i32(vec3_f16[1u])); |
| int const v_9 = ((v_8 + tint_f32_to_i32(vec4_f32[2u])) + vec4_i32[2u]); |
| int const v_10 = (v_9 + int(vec4_u32[2u])); |
| int const v_11 = (v_10 + tint_f16_to_i32(vec4_f16[2u])); |
| int const v_12 = (v_11 + tint_f32_to_i32(mat2x2_f32[0][0u])); |
| int const v_13 = (v_12 + tint_f32_to_i32(mat2x3_f32[0][0u])); |
| int const v_14 = (v_13 + tint_f32_to_i32(mat2x4_f32[0][0u])); |
| int const v_15 = (v_14 + tint_f32_to_i32(mat3x2_f32[0][0u])); |
| int const v_16 = (v_15 + tint_f32_to_i32(mat3x3_f32[0][0u])); |
| int const v_17 = (v_16 + tint_f32_to_i32(mat3x4_f32[0][0u])); |
| int const v_18 = (v_17 + tint_f32_to_i32(mat4x2_f32[0][0u])); |
| int const v_19 = (v_18 + tint_f32_to_i32(mat4x3_f32[0][0u])); |
| int const v_20 = (v_19 + tint_f32_to_i32(mat4x4_f32[0][0u])); |
| int const v_21 = (v_20 + tint_f16_to_i32(mat2x2_f16[0][0u])); |
| int const v_22 = (v_21 + tint_f16_to_i32(mat2x3_f16[0][0u])); |
| int const v_23 = (v_22 + tint_f16_to_i32(mat2x4_f16[0][0u])); |
| int const v_24 = (v_23 + tint_f16_to_i32(mat3x2_f16[0][0u])); |
| int const v_25 = (v_24 + tint_f16_to_i32(mat3x3_f16[0][0u])); |
| int const v_26 = (v_25 + tint_f16_to_i32(mat3x4_f16[0][0u])); |
| int const v_27 = (v_26 + tint_f16_to_i32(mat4x2_f16[0][0u])); |
| int const v_28 = (v_27 + tint_f16_to_i32(mat4x3_f16[0][0u])); |
| int const v_29 = (v_28 + tint_f16_to_i32(mat4x4_f16[0][0u])); |
| int const v_30 = (v_29 + tint_f16_to_i32(arr2_mat4x2_f16[0][0][0u])); |
| (*tint_module_vars.s) = (v_30 + tint_f32_to_i32(arr2_vec3_f32[0][0u])); |
| } |
| |
| kernel void tint_symbol(uint idx [[thread_index_in_threadgroup]], const device S* sb [[buffer(1)]], device int* s [[buffer(0)]]) { |
| tint_module_vars_struct const tint_module_vars = tint_module_vars_struct{.sb=sb, .s=s}; |
| tint_symbol_inner(idx, tint_module_vars); |
| } |