#define MONOHROM(m) ((COLORREF)(((BYTE)(m)|((WORD)((BYTE)(m))<<8))|(((DWORD)(BYTE)(m))<<16)))

void __fastcall TForm1::Button1Click(TObject *Sender)
{
    Button1 -> Enabled = false;
    DWORD *buftest = (DWORD*) malloc((4096*4096) * sizeof(DWORD));
    const DWORD kol_proh = 1000;
    DWORD tmp1[256] = {0}, tmp2[256] = {0}, tmp3[256] = {0}, FormColor = 0x00808080;

    long blend = 0, RS = 0, GS =  0, BS = 0, RD = 0, GD =  0, BD = 0;
    float fTimeStart = 0, fTimeStop = 0, fTotalTime = 0;
    float Time_test[kol_proh] = {0};

    ListBox1 -> Clear();
    for(DWORD speed_test = 0; speed_test < kol_proh; speed_test++, fTimeStart = fTimeStop = 0)
    {
  for(DWORD i = 0; i < (4096 * 4096); i++)
   {buftest[i] = i;}
  fTimeStart = clock() / (float)CLOCKS_PER_SEC;
  RD = GetRValue(FormColor);
  GD = GetGValue(FormColor);
  BD = GetBValue(FormColor);
  for(DWORD w = 0, h = 0, count = 0; h < 4096; h++)
  {
   for(w = 0; w < 4096; w++, count++)
   {
    buftest[count] = SWAP32(buftest[count]);
    buftest[count] = (buftest[count] >> 8) | (buftest[count] << 24);
    tmp1[(BYTE)count] = buftest[count];
    tmp2[(BYTE)count] = MONOHROM(buftest[count] >> 24);
    RS = GetRValue(buftest[count]);
    GS = GetGValue(buftest[count]);
    BS = GetBValue(buftest[count]);

    RS = (RS - RD) * (long)(buftest[count] >> 24) / 255 + RD;
    GS = (GS - GD) * (long)(buftest[count] >> 24) / 255 + GD;
    BS = (BS - BD) * (long)(buftest[count] >> 24) / 255 + BD;

    blend = RGB(RS, GS, BS);
    tmp3[(BYTE)count] = blend;
   }
  }
  fTimeStop = clock() / (float)CLOCKS_PER_SEC;
  Time_test[speed_test] = fTimeStop - fTimeStart;
  ListBox1 -> Items -> Add(FloatToStrF(Time_test[speed_test], ffGeneral, 10, 10));
  ListBox1 -> ItemIndex = ListBox1 -> Count -1;
  Application -> ProcessMessages();
    }
    free(buftest);

    for(DWORD speed_test = 0; speed_test < kol_proh; speed_test++)
  {fTotalTime += Time_test[speed_test];}
    fTotalTime /= kol_proh;

    ShowMessage("Average Run Time - " + FloatToStrF(fTotalTime, ffGeneral, 10, 10));
    Button1 -> Enabled = true;
}
//---------------------------------------------------------------------------

double GetCPUSpeed()
{
     int DelayTime = 500;

     DWORD TimerHi, TimerLo;
     int PriorityClass, Priority;

     PriorityClass = GetPriorityClass(GetCurrentProcess);
     Priority = GetThreadPriority(GetCurrentThread);

     SetPriorityClass(GetCurrentProcess, REALTIME_PRIORITY_CLASS);
     SetThreadPriority(GetCurrentThread, THREAD_PRIORITY_TIME_CRITICAL);

     Sleep(10);

     asm {
     dw 310Fh
     mov TimerLo, eax
     mov TimerHi, edx
     }

     Sleep(DelayTime);

     asm {
  dw 310Fh
  sub eax, TimerLo
  sbb edx, TimerHi
  mov TimerLo, eax
  mov TimerHi, edx
     }

     SetThreadPriority(GetCurrentThread, Priority);
     SetPriorityClass(GetCurrentProcess, PriorityClass);

     return (TimerLo/(1000.0 * DelayTime));
}

pixel = SWAP32(t); - это инвертирует все байты в пикселе. По принципу (было: 1234, стало 4321).
pixel = (pixel >> 8) | (pixel << 24); - это меняет местами первый и последний байт.