#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_private_vars_struct {
  float4x3 m;
};

struct tint_packed_vec3_f32_array_element {
  /* 0x0000 */ packed_float3 elements;
  /* 0x000c */ tint_array<int8_t, 4> tint_pad;
};

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(0)]]) {
  thread tint_private_vars_struct tint_private_vars = {};
  tint_private_vars.m = float4x3(float3(0.0f, 1.0f, 2.0f), float3(3.0f, 4.0f, 5.0f), float3(6.0f, 7.0f, 8.0f), float3(9.0f, 10.0f, 11.0f));
  assign_and_preserve_padding(tint_symbol, float4x3(tint_private_vars.m));
  return;
}