Quote Originally Posted by before View Post
Here's my humble contribution. Purpose was to see what's the difference between CDT and another copy waza method for Spi32M.

So; I've ran 2 32M yesterday @4.8GHz (wished for better clocks but the chip is FSB bugged @about 535FSB ); testing conditions are basically the same (see screenshot for hardware details). Both times, available real mem was about 543k with a difference of let's say 1k which is not significant IMHO.

1) run with CDT = 10m 33s
2) run after reboot with one of my cw .zip file = 10m 29s

Sorry for not having ran an untweaked one, but I'm quite sure it would have been slower than 10m 33s on my P5B.

To be back on topic. As it was said earlier in this thread, if CDT's really 60% of an entire method; so that's logical, it works and after comparison with another cw method, it can be improved.

So what you say is that you haven't seen "THE" gain out of it and it's a bit worse than Copy Waza?....