| #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 tint_symbol_5 { |
| tint_array<float2x3, 4> w; |
| }; |
| |
| struct tint_packed_vec3_f32_array_element { |
| /* 0x0000 */ packed_float3 elements; |
| /* 0x000c */ tint_array<int8_t, 4> tint_pad; |
| }; |
| |
| 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; |
| } |
| |
| tint_array<float2x3, 4> tint_unpack_vec3_in_composite_1(tint_array<tint_array<tint_packed_vec3_f32_array_element, 2>, 4> in) { |
| tint_array<float2x3, 4> result = {}; |
| for(uint i = 0u; (i < 4u); i = (i + 1u)) { |
| result[i] = tint_unpack_vec3_in_composite(in[i]); |
| } |
| return result; |
| } |
| |
| void f_inner(uint local_invocation_index, threadgroup tint_array<float2x3, 4>* const tint_symbol, const constant tint_array<tint_array<tint_packed_vec3_f32_array_element, 2>, 4>* const tint_symbol_1) { |
| for(uint idx = local_invocation_index; (idx < 4u); idx = (idx + 1u)) { |
| uint const i = idx; |
| (*(tint_symbol))[i] = float2x3(float3(0.0f), float3(0.0f)); |
| } |
| threadgroup_barrier(mem_flags::mem_threadgroup); |
| *(tint_symbol) = tint_unpack_vec3_in_composite_1(*(tint_symbol_1)); |
| (*(tint_symbol))[1] = tint_unpack_vec3_in_composite((*(tint_symbol_1))[2]); |
| (*(tint_symbol))[1][0] = float3((*(tint_symbol_1))[0][1].elements).zxy; |
| (*(tint_symbol))[1][0][0] = (*(tint_symbol_1))[0][1].elements[0]; |
| } |
| |
| kernel void f(const constant tint_array<tint_array<tint_packed_vec3_f32_array_element, 2>, 4>* tint_symbol_4 [[buffer(0)]], threadgroup tint_symbol_5* tint_symbol_3 [[threadgroup(0)]], uint local_invocation_index [[thread_index_in_threadgroup]]) { |
| threadgroup tint_array<float2x3, 4>* const tint_symbol_2 = &((*(tint_symbol_3)).w); |
| f_inner(local_invocation_index, tint_symbol_2, tint_symbol_4); |
| return; |
| } |
| |