[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: lazy sweeping.
From: |
Tom Lord |
Subject: |
Re: lazy sweeping. |
Date: |
Sun, 28 Jul 2002 21:31:40 -0700 (PDT) |
Nothing spectacular. But now checkout ... the startup time!
(this is the time for starting guile, and then pressing ^D)
This is not a very good measure of effective start-up time in
SCM-based (incremental graph translation) Schemes. Odds are you have
a lot of untranslated code in that image, and when I add my
application code, you'll have even more untranslated code. Substitute
anything useful for that ^D, and the real costs will start to become
more apparent, especially if your application involves long-running
processes.
SCM performs pretty damn well if what you want is a fast
edit-rerun-debug cycle for CAE and symbolic math tasks, especially
with the special GC for environments. It's also great if you have an
`eval'-intensive application.
Unexec helps, though I've noted substantial qualitative differences
between unexec _before_ running through most execution paths
vs. _after_: incremental translation in action.
Heck, you want _really_ fast start up time? Read the first line of
input to Guile before doing _anything_ else, doing a short-circuit
exit if the first line is just EOF. That'll be really fast ;-)
-t
Re: lazy sweeping., Han-Wen, 2002/07/30
Re: lazy sweeping.,
Tom Lord <=
Re: lazy sweeping., Mikael Djurfeldt, 2002/07/29