From what I have read, the major performance gains are for MP machines.Check the recent IBM benchmarks that report a 5-1 gain in web pages served for a 24 hour test on an 8-processor machine!
http://www-106.ibm.com/developerworks/linux/library/l-web26/index.html?ca=dgr-lnxw02KernCompare
Ya, but the issue here is.. a desktop system, running basic programs.. I.E. X, xchat-2, mozilla, [your favorite email client here], fluxbox, kde < whatever.. and a few Eterms/xterms
There is not any noticeable improvement on speeds.
See here: http://forums.gentoo.org/viewtopic.php?t=116183
And I QUOTE:
"Lovechild wrote: Love-sources was mainly meant to get a good 2.6 kernel for the desktop. we ship runtime selectable IO schedulers, the Nick Piggin Scheduler Policy patch and a few other goodies - it's based on the -mm kernel so it has all the goodies Andrew is shipping as well as any patches flowing around on lkml right after an -mm release to handle compile problems, etc."
So c'mon.. we are all tired of this bull crap.. let us see some hard evidence that these kernels, esp. the 2.6.X series.. PERIOD are any better than a 2.4.X kernel!!!!
A few of us have solid proof that the 2.6.X kernels actually are very, very much poorer in performance than the 2.4.X kernels!!
Our hdparm, glxgears and other things to test speed improvements are all very bad.
And please dont ask me for my outputs.. I spent quite awhile on this message:
Re: [gentoo-user] bulloney 2.6.X benchmarks (more!)
posted on 02/12/04 @ 13:09 CST - USA
Thank you
Sincerely,
-- TriKster Abacus irc.freenode.net #cllug #gentoo #linuxfriends irc.cotse.com #linux http://www.cllug.org http://www.trikster.homelinux.org http://www.trikster.homelinux.org/contact.html
-- [EMAIL PROTECTED] mailing list
