Another user's comment (mathematician this time).
Some of my own experiences using a Micron 90MHz pentium machine
wiith DPT eata-dma card and quantum atlas (7200rpm). 32Mb.
Stand alone.
I have found the swapping performance good. As illustrations:
Test 1: ppmtogif, needed 105Mb swap. Time < 5 minutes. Concurrently running
Motif. During test constantly ran free and moved windows.
Free response fast, windows reasonable.
Comparison: SGI Indigo (1993) same ram, no nfs essentially
standalone About 20 minutes. Very poor and slow response to
other commands.
Comparison Micron setup, WD IDE drive instead. SLOW (25 minutes).
Disk close to frying...
Test 2: ppmtogif. Needed 350 Mb swap (!). Switching consoles to run free
(ie not running X). Time about 20 minutes. In my view
fantastic.
Comparisons on IDE and SGI: None. Probably would have fried
the disk on IDE.
Agree strongly with Peter Jaeckel's suggestion about a compile time
option. My own experience - with SUN and IBM RISC/6000 machines
is that swapping performance can be very variable.
-- ************************************************* * * * Extraordinaire comme les mathematiques * * vous aident a vous connaitre. * * * * "Molloy" --Samuel Beckett * *************************************************Mike Field, email: mf@uh.edu; phone: 713-743-3470