cbuffer cbuffer_m : register(b0, space0) {
  uint4 m[2];
};
static int counter = 0;

int i() {
  counter = (counter + 1);
  return counter;
}

matrix<float16_t, 3, 4> m_load(uint offset) {
  const uint scalar_offset = ((offset + 0u)) / 4;
  uint4 ubo_load_1 = m[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 = m[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 = m[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));
  return matrix<float16_t, 3, 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]));
}

[numthreads(1, 1, 1)]
void f() {
  const matrix<float16_t, 3, 4> l_m = m_load(0u);
  uint2 ubo_load_6 = m[0].zw;
  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));
  const vector<float16_t, 4> l_m_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]);
  return;
}