Phoenix MUD, which is an ugly mix of 2.2 and 3.0 currently runs on a 486dx4 100, using up ALL 32 megs of ram on the machine, plus a couple of megs of a 50 meg swap file. We just recently filled a 500 meg disk, although a lot of that were entire backups of the mud. We have been using about 30-40k of bandwidth a day. I consider Phoenix to be a medium sized mud--we only have 8k rooms, and anywhere from 30-60 players a night. We do swap idle zones out, but that does not seem to help very much with the memory constraints. We are currently rewriting the mud from scratch in a pure 3.0 environment, hopefully being a little more conscious about optimization. > > On Sat, 16 Nov 1996, Erik wrote: > > - Heres an example of what changes can do though... Duris, extremely > - tight code, mem management optimized with in-line asm, uses ~40 megs ram, > - 30% of our T-1's bandwidth, and ~150 meg disk counting our player database. > - Compiles and testports are beginning to push the limit, so we've upgraded > - the box to 128 meg ram running on a dual p166 board. Ouch. > > Oh sure brag :P > > I don't think you can compare Duris to a new mud that may have 1 or 2 > intermediate coders. Don't let this guy scare you :) And once you know > what you're doing with the code, you can always optimize for memory, such > as switching idle zones out of memory (you can get HUGE savings this way, > especially when starting a new mud with a small player base). > > Sam > > p.s. You guys on Duris wouldn't happen to be employees of Microsoft, would > you? ;) > > +-----------------------------------------------------------+ > | Ensure that you have read the CircleMUD Mailing List FAQ: | > | http://cspo.queensu.ca/~fletcher/Circle/list_faq.html | > +-----------------------------------------------------------+ > +-----------------------------------------------------------+ | Ensure that you have read the CircleMUD Mailing List FAQ: | | http://cspo.queensu.ca/~fletcher/Circle/list_faq.html | +-----------------------------------------------------------+
This archive was generated by hypermail 2b30 : 12/18/00 PST