[Home] [Groups] - Message: [Prev in Group] [Next in Group]
10583: Re: [MUD-Dev] Text Parsing
[Full Header] [Plain Text]
From: Joey Hess <joey@kitenet.net>
Newsgroups: nu.kanga.list.mud-dev
Date: Wed, 9 Jun 1999 20:31:22 -0700
References: [1] [2] <-newest
Organization: Kanga.Nu
J C Lawrence wrote:
> I've scaled using simulated users under very artificial conditions
> and a rather minimal world to just over 700 simultaneous players on
> a fairly mid-range box. Performance at that point really became a
> kernel function due to poor MT work on IO under Linux (single lock).
When you speak of simultaneous users, how active were they? Did they issue 1
command per second, a constant stream of commands, or what?
I do frequent benchmarking on my moo, but I've so far been using a
simplistic test that calculates the average number of commands per second.
(Perlmoo can manage about 170/sec right now on a P-II 350, I figure that
corresponds to more than 5x that many normal active users.) Of course this
eliminates any overhead due to threading or network IO, etc, but that can be
a good thing when you have little control over those factors anyway.
--
see shy jo
_______________________________________________
MUD-Dev maillist - MUD-Dev@kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev