This may be an operating system issue: Windows 9x (including ME) vs.
Windows NT (including 2000/XP). Nikos reports terrible start times
under Windows ME, and Mike and CY report terrible start times under
Windows 98. My start times are fine (about 2 secs) under Windows 2000
(1.0 Ghz Athlon with 256 MB of RAM). XP is supposedly faster for things
like this, but I don't have it.
A very quick Web search shows that Windows 9x is still very widely used.
Among users of Shockwave, for example, W9x/WNT are about 50/50 as of
December
2002(http://www.macromedia.com/software/player_census/shockwaveplayer/us
er_profile.html). Of course, we don't have statistics about our "target
market" of math hackers, but the fact that at least three people on the
list are still using 9x shows that it *is* relevant.
Details:
On my machine, 5.9.0rc4 takes < 2 sec to start, and 5.5 takes < 1 sec to
start. These are warm start times, that is, that is, there was a
previous invocation of Maxima which I killed -- so DLLs and so on may
still be memory-resident. Restart time on 5.9.0 is about 1 second. I
also tried to simulate a cold start by using a large amount of memory in
some other processes (which I then killed), and in that case the start
times were both about 3 seconds. I suppose I should try after
rebooting, but I'm too lazy....