eoshough@nmsu.edu (Erik OShoughnessy) (07/11/90)
We have a NeXT running the newest release of the software (1.0a), with 8meg of RAM, and a 330 meg hard drive. This machine acts as an ftp site, and hosts a LP-Mud, which is a multi-user net game. The problem is the machine seems too slow, even for the amount of buisness it's doing, ( which isn't really all that much ). The guru's here speculate that the machine is running out of swap space, which would result in the slow down of the machine, and we suspect the Mud of not relinquishing memory. To solve the problem, we've tried killing the Mud, and allowing it to restart, killing the mud entirely, and then restarting it, asserting the *machine* and then restarting the mud, and then finally, just asserting the *machine* and not restarting the mud for a couple of hours. All of this resulted in absolutely no speed up of the machine. I'm asking the author of the Mud software about the memory allocation that the Mud software does, and I'm asking anyone out there if there's something "funny" the NeXT does in recovering allocated memory. Any pointers would be appreciated. -- ~ | Erik O'Shaughnessy |\/\/\/| ~ | Small Systems | | ~ | New Mexico State University | * | ~ o========== | Las Cruces, NM USA | (x)(x) |_|==== | eoshough@dante.nmsu.edu C _) === |------------------------------------ | ,___| === |"Do not follow the NULL pointer, | / | for that path leads to chaos, /____\ Die Bart! Die! | and madness." - me | \ |
dastrout@miavx1.acs.muohio.edu (root@next1) (07/16/90)
All NeXT's are slow. It's in the union rules. I belive the compiler puts approximately 5E7 NOP's between each call to any function. I would suggest re-writing the MUD in assembler. Along with the OS. Along with the Window Manager. Along with the..... -- Dave Strout dastrout@miavx1.acs.muohio.edu root@next1.acs.muohio.edu 513-523-8245