cbuffer cbuffer_u : register(b0, space0) {
  uint4 u[8];
};
RWByteAddressBuffer s : register(u1, space0);

void s_store_1(uint offset, matrix<float16_t, 4, 4> value) {
  s.Store<vector<float16_t, 4> >((offset + 0u), value[0u]);
  s.Store<vector<float16_t, 4> >((offset + 8u), value[1u]);
  s.Store<vector<float16_t, 4> >((offset + 16u), value[2u]);
  s.Store<vector<float16_t, 4> >((offset + 24u), value[3u]);
}

void s_store(uint offset, matrix<float16_t, 4, 4> value[4]) {
  matrix<float16_t, 4, 4> array_1[4] = value;
  {
    for(uint i = 0u; (i < 4u); i = (i + 1u)) {
      s_store_1((offset + (i * 32u)), array_1[i]);
    }
  }
}

matrix<float16_t, 4, 4> u_load_1(uint offset) {
  const uint scalar_offset = ((offset + 0u)) / 4;
  uint4 ubo_load_1 = u[scalar_offset / 4];
  uint2 ubo_load = ((scalar_offset & 2) ? ubo_load_1.zw : ubo_load_1.xy);
  vector<float16_t, 2> ubo_load_xz = vector<float16_t, 2>(f16tof32(ubo_load & 0xFFFF));
  vector<float16_t, 2> ubo_load_yw = vector<float16_t, 2>(f16tof32(ubo_load >> 16));
  const uint scalar_offset_1 = ((offset + 8u)) / 4;
  uint4 ubo_load_3 = u[scalar_offset_1 / 4];
  uint2 ubo_load_2 = ((scalar_offset_1 & 2) ? ubo_load_3.zw : ubo_load_3.xy);
  vector<float16_t, 2> ubo_load_2_xz = vector<float16_t, 2>(f16tof32(ubo_load_2 & 0xFFFF));
  vector<float16_t, 2> ubo_load_2_yw = vector<float16_t, 2>(f16tof32(ubo_load_2 >> 16));
  const uint scalar_offset_2 = ((offset + 16u)) / 4;
  uint4 ubo_load_5 = u[scalar_offset_2 / 4];
  uint2 ubo_load_4 = ((scalar_offset_2 & 2) ? ubo_load_5.zw : ubo_load_5.xy);
  vector<float16_t, 2> ubo_load_4_xz = vector<float16_t, 2>(f16tof32(ubo_load_4 & 0xFFFF));
  vector<float16_t, 2> ubo_load_4_yw = vector<float16_t, 2>(f16tof32(ubo_load_4 >> 16));
  const uint scalar_offset_3 = ((offset + 24u)) / 4;
  uint4 ubo_load_7 = u[scalar_offset_3 / 4];
  uint2 ubo_load_6 = ((scalar_offset_3 & 2) ? ubo_load_7.zw : ubo_load_7.xy);
  vector<float16_t, 2> ubo_load_6_xz = vector<float16_t, 2>(f16tof32(ubo_load_6 & 0xFFFF));
  vector<float16_t, 2> ubo_load_6_yw = vector<float16_t, 2>(f16tof32(ubo_load_6 >> 16));
  return matrix<float16_t, 4, 4>(vector<float16_t, 4>(ubo_load_xz[0], ubo_load_yw[0], ubo_load_xz[1], ubo_load_yw[1]), vector<float16_t, 4>(ubo_load_2_xz[0], ubo_load_2_yw[0], ubo_load_2_xz[1], ubo_load_2_yw[1]), vector<float16_t, 4>(ubo_load_4_xz[0], ubo_load_4_yw[0], ubo_load_4_xz[1], ubo_load_4_yw[1]), vector<float16_t, 4>(ubo_load_6_xz[0], ubo_load_6_yw[0], ubo_load_6_xz[1], ubo_load_6_yw[1]));
}

typedef matrix<float16_t, 4, 4> u_load_ret[4];
u_load_ret u_load(uint offset) {
  matrix<float16_t, 4, 4> arr[4] = (matrix<float16_t, 4, 4>[4])0;
  {
    for(uint i_1 = 0u; (i_1 < 4u); i_1 = (i_1 + 1u)) {
      arr[i_1] = u_load_1((offset + (i_1 * 32u)));
    }
  }
  return arr;
}

[numthreads(1, 1, 1)]
void f() {
  s_store(0u, u_load(0u));
  s_store_1(32u, u_load_1(64u));
  uint2 ubo_load_8 = u[0].zw;
  vector<float16_t, 2> ubo_load_8_xz = vector<float16_t, 2>(f16tof32(ubo_load_8 & 0xFFFF));
  vector<float16_t, 2> ubo_load_8_yw = vector<float16_t, 2>(f16tof32(ubo_load_8 >> 16));
  s.Store<vector<float16_t, 4> >(32u, vector<float16_t, 4>(ubo_load_8_xz[0], ubo_load_8_yw[0], ubo_load_8_xz[1], ubo_load_8_yw[1]).ywxz);
  s.Store<float16_t>(32u, float16_t(f16tof32(((u[0].z) & 0xFFFF))));
  return;
}