SPONSORED LINKS
> No, of course not. That's not how it works. But you may have noticed > a speed difference in a heavy duty graphics or mathmatical calculation series. > Or in a long series of processes in a large database. Or networking, sound, > and many other intensive operations. I know how it works. I said it was informal, and I wasn't just talking about zooming the pointer around the screen either. I was including a floating-point and memory intensive speech recognition system running in Linux in that informal speed comparision. I'm not quite sure what I said that was so wrong. I just ran a more scientific experiment in Linux with a 770 with 96MB vs. 64MB. (I simulated 64MB by telling Linux that it only had 64MB.) I ran the same program, and this program is banging on about 8MB. It took 33.0s with 64MB and 37.8s with 96MB. (These were the average of three runs on an unloaded system.) That seems to indicate a slowdown (for this program with this OS) of about 15% increasing memory above the 64MB cacheable limit. I do not know where the Linux kernel or the user program run in memory (above or below the cacheable limit). I also do not know the memory layout of Windows with respect to the cacheable limit. Of course, this slowdown could be a speedup if the additional memory eliminated virtual memory swapping or increased disk caching, so as with most things, YMMV. --Lee Hetherington PS: I am glad the Pentium II systems do not suffer from this silly more memory = slowdown syndrome. Silly Intel with their Pentium chipsets.