[Mplayer-users] [mplayer PATCH] fastmemcpy alignment for any cpu

Nick Kurshev nickols_k at mail.ru
Wed Apr 25 12:55:27 CEST 2001


>>To get realistic results imho you should to use Linux-2.4 and descrease number of running 
>> process down to minimum.
>Does that mean, that the 2.4 kernel allows faster copy operations ?

In short: NO

Simply Linux-2.4 has better implemented mechanisms of task scheduling, task switching and requires less
time for itself.
During benchmark test exetutes kernel interrupts it ( > 1000 times) and transfers control into other processes.
When you measure test performance - you measure not pure "time" of fastmemcpy, but fastmemcpy (our 
process) + kernel + other processes. To get realistic results you should descrease number of processes 
which are running simultaneously and use Linux-2.4 instead of 2.2.
(If you misunderstand it then try to find out in internet something about basic concepts of preemptive 

Best regards! Nick

P.S.: I had a bad mail loss during last day. If someone have sent me something during that time, please 
resend it again or send it to this mailing forum directly - I simply haven't received it. My apologies for that.

Mplayer-users mailing list
Mplayer-users at lists.sourceforge.net

More information about the MPlayer-users mailing list