#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_packed_vec3_f32_array_element {
  /* 0x0000 */ packed_float3 elements;
  /* 0x000c */ tint_array<int8_t, 4> tint_pad;
};

float4x3 tint_unpack_vec3_in_composite(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;
}

void assign_and_preserve_padding(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]);
}

kernel void f(device tint_array<tint_packed_vec3_f32_array_element, 4>* tint_symbol [[buffer(1)]], const constant tint_array<tint_packed_vec3_f32_array_element, 4>* tint_symbol_1 [[buffer(0)]]) {
  assign_and_preserve_padding(tint_symbol, tint_unpack_vec3_in_composite(*(tint_symbol_1)));
  (*(tint_symbol))[1].elements = (*(tint_symbol_1))[0].elements;
  (*(tint_symbol))[1].elements = packed_float3(float3((*(tint_symbol_1))[0].elements).zxy);
  (*(tint_symbol))[0].elements[1] = (*(tint_symbol_1))[1].elements[0];
  return;
}