#include <metal_stdlib>

using namespace metal;
struct Params {
  /* 0x0000 */ uint filterDim;
  /* 0x0004 */ uint blockDim;
};
struct Flip {
  /* 0x0000 */ uint value;
};
struct tint_array_wrapper_1 {
  float3 arr[256];
};
struct tint_array_wrapper {
  tint_array_wrapper_1 arr[4];
};

kernel void tint_symbol(texture2d<float, access::sample> tint_symbol_4 [[texture(1)]], sampler tint_symbol_5 [[sampler(0)]], texture2d<float, access::write> tint_symbol_6 [[texture(2)]], uint3 WorkGroupID [[threadgroup_position_in_grid]], uint3 LocalInvocationID [[thread_position_in_threadgroup]], uint local_invocation_index [[thread_index_in_threadgroup]], constant Params& params [[buffer(1)]], constant Flip& flip [[buffer(3)]]) {
  threadgroup tint_array_wrapper tint_symbol_3;
  if ((local_invocation_index == 0u)) {
    tint_array_wrapper const tint_symbol_2 = {.arr={}};
    tint_symbol_3 = tint_symbol_2;
  }
  threadgroup_barrier(mem_flags::mem_threadgroup);
  uint const filterOffset = ((params.filterDim - 1u) / 2u);
  int2 const dims = int2(tint_symbol_4.get_width(0), tint_symbol_4.get_height(0));
  int2 const baseIndex = (int2(((WorkGroupID.xy * uint2(params.blockDim, 4u)) + (LocalInvocationID.xy * uint2(4u, 1u)))) - int2(int(filterOffset), 0));
  {
    uint r = 0u;
    while (true) {
      if (!((r < 4u))) {
        break;
      }
      {
        uint c = 0u;
        while (true) {
          if (!((c < 4u))) {
            break;
          }
          int2 loadIndex = (baseIndex + int2(int(c), int(r)));
          if ((flip.value != 0u)) {
            loadIndex = loadIndex.yx;
          }
          tint_symbol_3.arr[r].arr[((4u * LocalInvocationID.x) + c)] = tint_symbol_4.sample(tint_symbol_5, ((float2(loadIndex) + float2(0.25f, 0.25f)) / float2(dims)), level(0.0f)).rgb;
          {
            c = (c + 1u);
          }
        }
      }
      {
        r = (r + 1u);
      }
    }
  }
  threadgroup_barrier(mem_flags::mem_threadgroup);
  {
    uint r = 0u;
    while (true) {
      if (!((r < 4u))) {
        break;
      }
      {
        uint c = 0u;
        while (true) {
          if (!((c < 4u))) {
            break;
          }
          int2 writeIndex = (baseIndex + int2(int(c), int(r)));
          if ((flip.value != 0u)) {
            writeIndex = writeIndex.yx;
          }
          uint const center = ((4u * LocalInvocationID.x) + c);
          if ((((center >= filterOffset) && (center < (256u - filterOffset))) && all((writeIndex < dims)))) {
            float3 acc = float3(0.0f, 0.0f, 0.0f);
            {
              uint f = 0u;
              while (true) {
                if (!((f < params.filterDim))) {
                  break;
                }
                uint i = ((center + f) - filterOffset);
                acc = (acc + ((1.0f / float(params.filterDim)) * tint_symbol_3.arr[r].arr[i]));
                {
                  f = (f + 1u);
                }
              }
            }
            tint_symbol_6.write(float4(acc, 1.0f), uint2(writeIndex));
          }
          {
            c = (c + 1u);
          }
        }
      }
      {
        r = (r + 1u);
      }
    }
  }
  return;
}