test/tint/extensions/subgroup_matrix/override_array.wgsl.expected.dxc.hlsl

#include <dx/linalg.h>
using namespace dx::linalg;
using Matrix_left_f32_8x8 = Matrix<ComponentType::F32, 8, 8, MatrixUse::A, MatrixScope::Wave>;
using Matrix_right_f32_8x8 = Matrix<ComponentType::F32, 8, 8, MatrixUse::B, MatrixScope::Wave>;
struct main_inputs {
  uint tint_local_index : SV_GroupIndex;
};


RWByteAddressBuffer s_var : register(u0);
groupshared float wg_var[1024];
void main_inner(uint tint_local_index) {
  {
    uint idx = tint_local_index;
    while(true) {
      uint v = idx;
      if ((v >= 1024u)) {
        break;
      }
      wg_var[v] = 0.0f;
      {
        idx = (idx + 32u);
      }
    }
  }
  GroupMemoryBarrierWithGroupSync();
  uint v_1 = 0u;
  s_var.GetDimensions(v_1);
  bool v_2 = (64u <= (v_1 / 4u));
  Matrix_left_f32_8x8 m = Matrix_left_f32_8x8::Load(s_var, (0u + (select(v_2, 0u, 0u) * 4u)), (select(v_2, 8u, 8u) * 4u), MatrixLayout::RowMajor);
  m.Store(wg_var, 0u, 8u, MatrixLayout::RowMajor);
  GroupMemoryBarrierWithGroupSync();
  Matrix_right_f32_8x8 m2 = Matrix_right_f32_8x8::Load(wg_var, 0u, 8u, MatrixLayout::ColMajor);
  uint v_3 = 0u;
  s_var.GetDimensions(v_3);
  bool v_4 = (64u <= (v_3 / 4u));
  m2.Store(s_var, (0u + (select(v_4, 0u, 0u) * 4u)), (select(v_4, 8u, 8u) * 4u), MatrixLayout::ColMajor);
}

[numthreads(32, 1, 1)]
void main(main_inputs inputs) {
  main_inner(inputs.tint_local_index);
}