procedure THIPBlur.Simple;
var rc:PThreadRec;
    i,c:integer;
    id:LongWord;
    lpSystemInfo:_SYSTEM_INFO;
    FEvents:array of cardinal;
    lst:PList;
begin
   bmp.PixelFormat := pf24bit;
   src.PixelFormat := pf24bit;
   if(bmp.Width > 256)and(bmp.Height > 256)then
     begin
       GetSystemInfo(lpSystemInfo); 
       c := lpSystemInfo.dwNumberOfProcessors;
   //!! Выясняет количество доступных ядер/cpu  
     end
   else c := 1;
   
   lst := NewList;
   SetLength(FEvents, c);
   for i := 1 to c do
     begin
       new(rc);
       rc.src := src;
       rc.bmp := bmp;
       rc.size := bmp.height div c;
       rc.Start := (i-1)*rc.size;
       rc.handle := CreateThread(0, 0, @Simple_proc, rc, 0, id);
         //..и раскидывает задачу по потокам количество которых равно количеству ядер    
       FEvents[i-1] := rc.handle;
       lst.Add(rc); 
       SetThreadPriority(rc.handle, THREAD_PRIORITY_HIGHEST);
     end;
   WaitForMultipleObjects(c, PWOHandleArray(@FEvents[0]), true, cardinal(-1));
   for i := 0 to c-1 do
     begin
       CloseHandle(FEvents[i]);
       dispose(PThreadRec(lst.Items[i]));
     end;
   lst.Free; 
end;

function Simple_proc(l:pointer):Integer; {$ifdef F_P}stdcall;{$endif}

...

rc.handle := {$ifdef F_P}CreateThread{$else}BeginThread{$endif}rc.handle(0, 0, @Simple_proc, rc, 0, id);