// 
// sub_100834 from 0x100834 to 0x100928 (244 bytes)
// 
// 1 xrefs:
// > <???> @ 0x118af8
// 
// 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
// 

int32x4_t *__fastcall sub_100834(
        int32x4_t *result,
        float32x4_t *a2,
        unsigned int a3,
        __int64 a4,
        __int64 a5,
        __int64 a6,
        int a7)
{
  float32x2_t v7; // d0
  float v8; // s1
  unsigned int v9; // w8
  float32x4_t v10; // q2
  int32x4_t v11; // q3
  int32x4_t v12; // q4
  __int64 v13; // x8
  float v14; // s2
  int32x4_t v15; // q5
  int32x4_t v16; // q16
  float32x4_t v17; // q4
  float32x4_t v18; // q6
  float32x4_t v19; // q7
  __int64 v20; // x8

  v7.n64_u32[0] = *(_DWORD *)a4;
  v8 = *(float *)(a4 + 128);
  v9 = a3 >> 3;
  v10.n128_u32[0] = *(_DWORD *)a4;
  v10.n128_f32[1] = v8;
  v10.n128_u32[2] = *(_DWORD *)a4;
  v10.n128_f32[3] = v8;
  if ( a7 )
  {
    for ( ; v9; a2 += 4 )
    {
      v11 = *result;
      v12 = result[1];
      result += 2;
      --v9;
      *a2 = vmulq_f32(vzip1q_s32(v11, v11), v10);
      a2[1] = vmulq_f32(vzip2q_s32(v11, v11), v10);
      a2[2] = vmulq_f32(vzip1q_s32(v12, v12), v10);
      a2[3] = vmulq_f32(vzip2q_s32(v12, v12), v10);
    }
    v13 = 0;
    while ( (a3 & 7) != (_DWORD)v13 )
    {
      v14 = result->n128_f32[v13++];
      a2->n128_f32[0] = v14 * v7.n64_f32[0];
      a2->n128_f32[1] = v14 * v8;
      a2 = (float32x4_t *)((char *)a2 + 8);
    }
  }
  else
  {
    while ( v9 )
    {
      v15 = *result;
      v16 = result[1];
      result += 2;
      --v9;
      v17 = vmlaq_f32(a2[1], v10, vzip2q_s32(v15, v15));
      v18 = vmlaq_f32(a2[2], v10, vzip1q_s32(v16, v16));
      v19 = vmlaq_f32(a2[3], v10, vzip2q_s32(v16, v16));
      *a2 = vmlaq_f32(*a2, v10, vzip1q_s32(v15, v15));
      a2[1] = v17;
      a2[2] = v18;
      a2[3] = v19;
      a2 += 4;
    }
    v20 = 0;
    v7.n64_f32[1] = v8;
    while ( (a3 & 7) != (_DWORD)v20 )
    {
      a2->n128_u64[v20] = vmla_n_f32((float32x2_t)a2->n128_u64[v20], v7, result->n128_f32[v20]).n64_u64[0];
      ++v20;
    }
  }
  return result;
}