// 
// sub_100AA8 from 0x100aa8 to 0x100dec (836 bytes)
// 
// 1 xrefs:
// > <???> @ 0x118b10
// 
// This function is hookable on this platform!
// 
// Using BromaIDA 8.0.0 @ https://github.com/Stazzical/BromaIDA
// Using bindings at commit ba9f177b at Thu Jul 16 22:52:06 2026 from https://github.com/geode-sdk/bindings
// 

float32x2_t *__fastcall sub_100AA8(
        float32x2_t *result,
        float32x4_t *s,
        unsigned int a3,
        float *a4,
        __int64 a5,
        __int64 a6,
        int a7)
{
  float v7; // s6
  float32x2_t v8; // d0
  float v9; // s7
  float32x2_t v10; // d1
  float v11; // s5
  float32x2_t v12; // d2
  float32x2_t v13; // d30
  float v14; // s17
  float v15; // s31
  float32x2_t v16; // d3
  float v17; // s16
  float32x2_t v18; // d4
  float32x4_t *v20; // x19
  float32x2_t *v21; // x20
  float32x2_t v22; // d8
  unsigned int i; // w8
  unsigned __int64 v24; // d4
  unsigned __int64 v25; // d5
  unsigned __int64 v26; // d6
  unsigned __int64 v27; // d7
  __int64 v28; // x8
  float32x2_t v29; // d0
  float32x4_t v30; // q18
  float32x4_t v31; // q19
  float32x4_t v32; // q20
  float32x4_t v33; // q21
  float32x4_t v34; // q22
  float32x4_t v35; // q23
  unsigned int v36; // w8
  int32x4_t v37; // t1
  int v38; // w8
  float v39; // s18
  float v40; // s19
  int32x4_t v41; // t1
  float32x4_t v42; // q26
  float32x4_t v43; // q24
  int v44; // w8
  float *v45; // x9
  float32x2_t v46; // d5
  float32x2_t v47; // d17
  float32x2_t v48; // d18
  float32x2_t v49; // d5
  float32x2_t v50; // d16
  unsigned __int64 v51; // d7
  float v52; // [xsp+0h] [xbp-50h]
  float v53; // [xsp+10h] [xbp-40h]

  v7 = a4[161];
  v8.n64_f32[0] = a4[64];
  v9 = a4[96];
  v10.n64_f32[0] = a4[128];
  v11 = a4[160];
  v13.n64_f32[0] = *a4;
  v12.n64_f32[0] = a4[1];
  v14 = a4[32];
  v15 = a4[33];
  v16.n64_f32[0] = a4[65];
  v17 = a4[97];
  v18.n64_f32[0] = a4[129];
  v20 = s;
  v21 = result;
  if ( fabsf(v7) >= 0.00002
    || fabsf(v11) >= 0.00002
    || fabsf(v18.n64_f32[0]) >= 0.00002
    || fabsf(v10.n64_f32[0]) >= 0.00002
    || fabsf(v17) >= 0.00002
    || fabsf(v9) >= 0.00002
    || fabsf(v8.n64_f32[0]) >= 0.00002
    || fabsf(v14) >= 0.00002
    || fabsf(v12.n64_f32[0]) >= 0.00002
    || fabsf(v16.n64_f32[0]) >= 0.00002 )
  {
    v30.n128_f32[0] = *a4;
    v31.n128_f32[0] = a4[128];
    v32.n128_f32[0] = a4[64];
    v33.n128_f32[0] = a4[1];
    v34.n128_f32[0] = a4[129];
    v35.n128_f32[0] = a4[65];
    v30.n128_f32[1] = a4[32];
    v31.n128_f32[1] = a4[160];
    v32.n128_f32[1] = a4[96];
    v33.n128_f32[1] = a4[33];
    v34.n128_f32[1] = a4[161];
    v35.n128_f32[1] = a4[97];
    v30.n128_u64[1] = __PAIR64__(LODWORD(v9), v8.n64_u32[0]);
    v31.n128_u64[1] = __PAIR64__(LODWORD(v14), v13.n64_u32[0]);
    v32.n128_u64[1] = __PAIR64__(LODWORD(v11), v10.n64_u32[0]);
    v33.n128_u64[1] = __PAIR64__(LODWORD(v17), v16.n64_u32[0]);
    v34.n128_u64[1] = __PAIR64__(LODWORD(v15), v12.n64_u32[0]);
    v35.n128_u64[1] = __PAIR64__(LODWORD(v7), v18.n64_u32[0]);
    v36 = a3 >> 1;
    if ( a7 )
    {
      for ( ; v36; v20 += 3 )
      {
        v37 = *(int32x4_t *)v21->n64_u64;
        v21 += 2;
        --v36;
        v20[1] = vmlaq_f32(vmulq_f32(vtrn1q_s32(v37, v37), v31), v34, vtrn2q_s32(v37, v37));
        v20[2] = vmlaq_laneq_f32(vmulq_laneq_f32(v32, v37, 2), v35, v37, 3);
        *v20 = vmlaq_lane_f32(vmulq_n_f32(v30, v37.n128_f32[0]), v33, (float32x2_t)v37.n128_u64[0], 1);
      }
      v38 = a3 & 1;
      if ( (a3 & 1) != 0 )
      {
        do
        {
          v39 = v21->n64_f32[0];
          v40 = v21->n64_f32[1];
          ++v21;
          --v38;
          v20->n128_f32[0] = (float)(v39 * v13.n64_f32[0]) + (float)(v40 * v12.n64_f32[0]);
          v20->n128_f32[1] = (float)(v39 * v14) + (float)(v40 * v15);
          v20->n128_f32[2] = (float)(v39 * v8.n64_f32[0]) + (float)(v40 * v16.n64_f32[0]);
          v20->n128_f32[3] = (float)(v39 * v9) + (float)(v40 * v17);
          v20[1].n128_f32[0] = (float)(v39 * v10.n64_f32[0]) + (float)(v40 * v18.n64_f32[0]);
          v20[1].n128_f32[1] = (float)(v39 * v11) + (float)(v40 * v7);
          v20 = (float32x4_t *)((char *)v20 + 24);
        }
        while ( v38 );
      }
    }
    else
    {
      while ( v36 )
      {
        v41 = *(int32x4_t *)v21->n64_u64;
        v21 += 2;
        --v36;
        v42 = vmlaq_f32(vmlaq_f32(v20[1], v31, vtrn1q_s32(v41, v41)), v34, vtrn2q_s32(v41, v41));
        v43 = vaddq_f32(vmlaq_laneq_f32(vmulq_laneq_f32(v32, v41, 2), v35, v41, 3), v20[2]);
        *v20 = vmlaq_lane_f32(vmlaq_n_f32(*v20, v30, v41.n128_f32[0]), v33, (float32x2_t)v41.n128_u64[0], 1);
        v20[1] = v42;
        v20[2] = v43;
        v20 += 3;
      }
      v44 = a3 & 1;
      v12.n64_f32[1] = v15;
      v13.n64_f32[1] = v14;
      v16.n64_f32[1] = v17;
      v8.n64_f32[1] = v9;
      v18.n64_f32[1] = v7;
      v10.n64_f32[1] = v11;
      if ( (a3 & 1) != 0 )
      {
        do
        {
          v46.n64_u64[0] = vld1_dup_f32((const float *)v21).n64_u64[0];
          v45 = (float *)v21 + 1;
          --v44;
          ++v21;
          v47.n64_u64[0] = vmul_f32(v46, v13).n64_u64[0];
          v48.n64_u64[0] = vmul_f32(v46, v8).n64_u64[0];
          v49.n64_u64[0] = vmla_n_f32(vmul_f32(v46, v10), v18, *v45).n64_u64[0];
          v50.n64_u64[0] = v20[1].n128_u64[0];
          v51 = vadd_f32(vmla_n_f32(v48, v16, *v45), *(float32x2_t *)&v20->n128_i8[8]).n64_u64[0];
          v20->n128_u64[0] = vadd_f32(vmla_n_f32(v47, v12, *v45), (float32x2_t)v20->n128_u64[0]).n64_u64[0];
          v20->n128_u64[1] = v51;
          v20[1].n128_u64[0] = vadd_f32(v49, v50).n64_u64[0];
          v20 = (float32x4_t *)((char *)v20 + 24);
        }
        while ( v44 );
      }
    }
  }
  else
  {
    v22.n64_u64[0] = __PAIR64__(LODWORD(v15), v13.n64_u32[0]);
    if ( a7 )
    {
      v52 = a4[33];
      v53 = *a4;
      result = (float32x2_t *)memset(s, 0, 24LL * a3);
      v15 = v52;
      v13.n64_f32[0] = v53;
    }
    for ( i = a3 >> 2; i; --i )
    {
      v24 = vmla_f32((float32x2_t)v20->n128_u64[0], v22, (float32x2_t)v21->n64_u64[0]).n64_u64[0];
      v25 = vmla_f32(*(float32x2_t *)&v20[1].n128_i8[8], v22, v21[1]).n64_u64[0];
      v26 = vmla_f32((float32x2_t)v20[3].n128_u64[0], v22, v21[2]).n64_u64[0];
      v27 = vmla_f32(*(float32x2_t *)&v20[4].n128_i8[8], v22, v21[3]).n64_u64[0];
      v21 += 4;
      v20->n128_u64[0] = v24;
      v20[1].n128_u64[1] = v25;
      v20[3].n128_u64[0] = v26;
      v20[4].n128_u64[1] = v27;
      v20 += 6;
    }
    v28 = 0;
    v13.n64_f32[1] = v15;
    while ( (a3 & 3) != (_DWORD)v28 )
    {
      v29.n64_u64[0] = (unsigned __int64)v21[v28++];
      v20->n128_u64[0] = vmla_f32((float32x2_t)v20->n128_u64[0], v13, v29).n64_u64[0];
      v20 = (float32x4_t *)((char *)v20 + 24);
    }
  }
  return result;
}