Re: [CODE] Parallel Processing

From: Daniel A. Koepke (
Date: 02/21/00

On Mon, 21 Feb 2000, George Greer wrote:

> I think the better question is why?  CircleMUD does not take up much
> processing power at all.  Even a lowly 486 could keep up with it.
> I'm guessing a 386 wouldn't have much trouble either if it had enough
> RAM.

A better indicator: I know of one graphical Mud (none of the big ones like
UO or EQ) which has some fairly computationally expensive server features
but could easily support 2000 or more players on a Pentium III.  Now, of
course you don't have a dedicated Pentium III, but the following is true

    a. None of your muds are getting that many simultaneous connections.

    b. 90% of the time, the mud is inactive in select().

    c. CircleMUD isn't as computationally expensive as a Mud which
       performs calculations of this magnitude.

Or, in other words, you probably don't need a multi-processor machine for
your Mud.  However, compiling with it would be nice.

None of this means that CircleMUD shouldn't, at some time in the future,
offer some sort of threading, even if its just to fork() off some children
(note that distinct processes can be moved to other CPUs by the operating
system and that this a relatively easy way to benefit from MP machines).
I would like to see that option made available in a from-scratch
implementation of CircleMUD just to know that the Mud would scale to
heavier loads.  (Add on top of that a less restricted license and you'd
have CircleMUD used in situations in which the scalability might be an
issue.)  But it's neither practical or useful at this time.  Just get a
faster computer if you really think it's an issue: they're cheaper,


     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
     |  |

This archive was generated by hypermail 2b30 : 04/10/01 PDT