reference/shaders-msl/comp/shared.comp

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

using namespace metal;

struct SSBO
{
    float in_data[1];
};

struct SSBO2
{
    float out_data[1];
};

constant uint3 gl_WorkGroupSize [[maybe_unused]] = uint3(4u, 1u, 1u);

kernel void main0(const device SSBO& _22 [[buffer(0)]], device SSBO2& _44 [[buffer(1)]], uint3 gl_GlobalInvocationID [[thread_position_in_grid]], uint gl_LocalInvocationIndex [[thread_index_in_threadgroup]])
{
    threadgroup float sShared[4];
    uint ident = gl_GlobalInvocationID.x;
    float idata = _22.in_data[ident];
    sShared[gl_LocalInvocationIndex] = idata;
    threadgroup_barrier(mem_flags::mem_threadgroup);
    _44.out_data[ident] = sShared[(4u - gl_LocalInvocationIndex) - 1u];
}