guile-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

GC success report


From: Han-Wen
Subject: GC success report
Date: Sun, 4 Aug 2002 16:06:33 +0200

Hi there,

I managed to dig up some old timings for lilypond  (linked against
1.5.6), and the smallest change (the trigger in scm_malloc) actually
brought the biggest improvement:

1.5.6, lilypond on the K.V.498 piano-part, a moderately large score (heap
size: 2 M cells / 6 M malloced) is down from

  * (gc-times . 128)

to cvs, newgc: 

  * (gc-times . 32)

The improvements in running time seem to match this; they are down by
approximately 20 to 30%, and now lilypond spends only 1/10th of the
running time in GC -- which seems reasonable to me.  The numbers are
very sketchy, but that's because I keep forgetting to store my
benchmark results.

The bad news that now the need for Gen GC doesn't seem so pressing
anymore.

-- 

Han-Wen Nienhuys   |   address@hidden   |   http://www.cs.uu.nl/~hanwen 



reply via email to

[Prev in Thread] Current Thread [Next in Thread]