If you run it in fullscreen and do not use delay, it should get all the CPU time (beside time that was granted to higher / Highest / realtime priority threads) ... if you add delay then you will have serious problems as delay 1 ms can take longer than 1 ms if a different thread is doing something at that point. This can become especially worse on systems with VSync on when you skip the "right" ms for flipping because it then will have influence on later flips as well (for that reason I do flip 0 time based)
I assume you used time difference based interpolation for your framework ... and in that case, these differences shouldn't mather that much ...
I assume you used time difference based interpolation for your framework ... and in that case, these differences shouldn't mather that much ...