Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: I am surprised: under XP programs run up to 10% faster than under 98!

Author: Matt Taylor

Date: 09:01:15 04/01/03

Go up one level in this thread


On April 01, 2003 at 11:30:21, Russell Reagan wrote:

>On March 31, 2003 at 23:58:47, Jouni Uski wrote:
>
>>My new PC is now fixed (they replaced whole mainboard). I compared some test
>>runs under XP Pro to old Windows98 SE test and surprised: specially Ruffian now
>>runs 10% faster and also Fritz8 2-5% faster (including Fritzmark). I sure didn't
>>except this. Only minor drawback is, that XP takes more RAM, but it's only
>>10-15MB more. So Microsoft has something right, when they claim programs run
>>faster in XP.
>
>The only thing I can think of is that XP manages memory better than the older
>Win9x's. "Better than Win9x" isn't saying much though.

The stability of the Windows NT kernel is approximately equal to the stability
of the Linux kernel (which is acclaimed for this fact). What keeps NT from being
as stable as Linux is drivers.

There are many reasons why XP might be faster than Win9x, the first of which is
the fact that XP does not copy 2 KB of page tables on every thread switch.
(That's probably why XP can switch twice as often, too.) Microsoft also boasts
about syscall speed advantages, though I think that is a bombastic claim. I've
measured the frequency of interrupt 2E generation (kernel calls). If anything,
XP will load itself and programs faster than its predecessors. Actual int 2E
rates during the execution of programs are quite low. When there is no
mouse/keyboard activity, I have recorded fewer than 10 per second. Hopefully
Chess engines aren't causing tons of int 2E's during their execution.

-Matt



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.