#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;
};

struct S_tint_packed_vec3 {
  /* 0x0000 */ int before;
  /* 0x0004 */ tint_array<int8_t, 12> tint_pad_1;
  /* 0x0010 */ tint_array<tint_packed_vec3_f32_array_element, 4> m;
  /* 0x0050 */ tint_array<int8_t, 48> tint_pad_2;
  /* 0x0080 */ int after;
  /* 0x0084 */ tint_array<int8_t, 60> tint_pad_3;
};

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;
}

struct S {
  int before;
  float4x3 m;
  int after;
};

kernel void f(const constant tint_array<S_tint_packed_vec3, 4>* tint_symbol [[buffer(0)]]) {
  float3x4 const t = transpose(tint_unpack_vec3_in_composite((*(tint_symbol))[2].m));
  float const l = length(float3((*(tint_symbol))[0].m[1].elements).zxy);
  float const a = fabs(float3((*(tint_symbol))[0].m[1].elements).zxy[0]);
  return;
}