From: Ghost Shaidan <ghost@ao.net>
Date: Monday, February 01, 1999 10:23 AM
>If there is a standard for testing this, I'd be happy to perform the tests
>on my mud. We use a home grown ascii pfile system. I have noticed no
>performance problems with them, other than a lag if we do a force all
>save with more than 20 or 25 people on. (Looking for some sort of solution
>for this, but I am not far enough along in my research to make any
>statements.)
I've done this two ways. The hard (but portable) way is to use the
gettimeofday() and timediff() functions in comm.c. I don't have the code
handy, but if you seach comm.c for gettimeofday you should see what I mean
where circle figures out how many milliseconds it needs to wait for the next
tick.
The easier way is to use a profiler, assuming you've got one. That can also
be used to figure out where your lag is coming from more easily than the above
method.
From past experience with the original ascii release, the weak link was
loading one line of data at a time. That's why in 2.0 I added the diskio.c
utilities. Instead of reading a line at a time (actually a character at a
time until an endline), parsing, and repeating, I load the entire file into
memory using a single fread and parse from memory. The increase in
performance was incredible. I do the same thing when saving.
The diskio.c module in the 2.0 release is pretty easy to use if you'd like to
give it a try. It'll require a little rewriting of your load/save routines,
but may be worth the effort if your lag is caused by slow loads/saves.
Sam
+------------------------------------------------------------+
| Ensure that you have read the CircleMUD Mailing List FAQ: |
| http://qsilver.queensu.ca/~fletchra/Circle/list-faq.html |
+------------------------------------------------------------+
This archive was generated by hypermail 2b30 : 12/15/00 PST