#pragma clang diagnostic ignored "-Wmissing-prototypes"

#include <metal_stdlib>
#include <simd/simd.h>

using namespace metal;

template<typename T, uint A>
inline void spvArrayCopyFromConstantToStack1(thread T (&dst)[A], constant T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromConstantToThreadGroup1(threadgroup T (&dst)[A], constant T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromStackToStack1(thread T (&dst)[A], thread const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromStackToThreadGroup1(threadgroup T (&dst)[A], thread const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromThreadGroupToStack1(thread T (&dst)[A], threadgroup const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromThreadGroupToThreadGroup1(threadgroup T (&dst)[A], threadgroup const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromDeviceToDevice1(device T (&dst)[A], device const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromConstantToDevice1(device T (&dst)[A], constant T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromStackToDevice1(device T (&dst)[A], thread const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromThreadGroupToDevice1(device T (&dst)[A], threadgroup const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromDeviceToStack1(thread T (&dst)[A], device const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A>
inline void spvArrayCopyFromDeviceToThreadGroup1(threadgroup T (&dst)[A], device const T (&src)[A])
{
    for (uint i = 0; i < A; i++)
    {
        dst[i] = src[i];
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromConstantToStack2(thread T (&dst)[A][B], constant T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromConstantToStack1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromConstantToThreadGroup2(threadgroup T (&dst)[A][B], constant T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromConstantToThreadGroup1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromStackToStack2(thread T (&dst)[A][B], thread const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromStackToStack1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromStackToThreadGroup2(threadgroup T (&dst)[A][B], thread const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromStackToThreadGroup1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromThreadGroupToStack2(thread T (&dst)[A][B], threadgroup const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromThreadGroupToStack1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromThreadGroupToThreadGroup2(threadgroup T (&dst)[A][B], threadgroup const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromThreadGroupToThreadGroup1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromDeviceToDevice2(device T (&dst)[A][B], device const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromDeviceToDevice1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromConstantToDevice2(device T (&dst)[A][B], constant T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromConstantToDevice1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromStackToDevice2(device T (&dst)[A][B], thread const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromStackToDevice1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromThreadGroupToDevice2(device T (&dst)[A][B], threadgroup const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromThreadGroupToDevice1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromDeviceToStack2(thread T (&dst)[A][B], device const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromDeviceToStack1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B>
inline void spvArrayCopyFromDeviceToThreadGroup2(threadgroup T (&dst)[A][B], device const T (&src)[A][B])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromDeviceToThreadGroup1(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromConstantToStack3(thread T (&dst)[A][B][C], constant T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromConstantToStack2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromConstantToThreadGroup3(threadgroup T (&dst)[A][B][C], constant T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromConstantToThreadGroup2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromStackToStack3(thread T (&dst)[A][B][C], thread const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromStackToStack2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromStackToThreadGroup3(threadgroup T (&dst)[A][B][C], thread const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromStackToThreadGroup2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromThreadGroupToStack3(thread T (&dst)[A][B][C], threadgroup const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromThreadGroupToStack2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromThreadGroupToThreadGroup3(threadgroup T (&dst)[A][B][C], threadgroup const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromThreadGroupToThreadGroup2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromDeviceToDevice3(device T (&dst)[A][B][C], device const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromDeviceToDevice2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromConstantToDevice3(device T (&dst)[A][B][C], constant T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromConstantToDevice2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromStackToDevice3(device T (&dst)[A][B][C], thread const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromStackToDevice2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromThreadGroupToDevice3(device T (&dst)[A][B][C], threadgroup const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromThreadGroupToDevice2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromDeviceToStack3(thread T (&dst)[A][B][C], device const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromDeviceToStack2(dst[i], src[i]);
    }
}

template<typename T, uint A, uint B, uint C>
inline void spvArrayCopyFromDeviceToThreadGroup3(threadgroup T (&dst)[A][B][C], device const T (&src)[A][B][C])
{
    for (uint i = 0; i < A; i++)
    {
        spvArrayCopyFromDeviceToThreadGroup2(dst[i], src[i]);
    }
}

struct BUF
{
    int a;
    float b;
    float c;
};

constant uint3 gl_WorkGroupSize [[maybe_unused]] = uint3(1u);

constant float _16[2] = { 1.0, 2.0 };
constant float _19[2] = { 3.0, 4.0 };
constant float _20[2][2] = { { 1.0, 2.0 }, { 3.0, 4.0 } };
constant float _21[2][2][2] = { { { 1.0, 2.0 }, { 3.0, 4.0 } }, { { 1.0, 2.0 }, { 3.0, 4.0 } } };

kernel void main0(device BUF& o [[buffer(0)]])
{
    float c[2][2][2];
    spvArrayCopyFromConstantToStack3(c, _21);
    o.a = int(c[1][1][1]);
    float _43[2] = { o.b, o.c };
    float _48[2] = { o.b, o.b };
    float _49[2][2] = { { _43[0], _43[1] }, { _48[0], _48[1] } };
    float _54[2] = { o.c, o.c };
    float _59[2] = { o.c, o.b };
    float _60[2][2] = { { _54[0], _54[1] }, { _59[0], _59[1] } };
    float _61[2][2][2] = { { { _49[0][0], _49[0][1] }, { _49[1][0], _49[1][1] } }, { { _60[0][0], _60[0][1] }, { _60[1][0], _60[1][1] } } };
    float d[2][2][2];
    spvArrayCopyFromStackToStack3(d, _61);
    float e[2][2][2];
    spvArrayCopyFromStackToStack3(e, d);
    o.b = e[1][0][1];
}