[gs-bugs] [Bug 689966] c329.bin hang with bitcmyk in clist code

bugs.ghostscript.com-bugzilla-daemon at ghostscript.com bugs.ghostscript.com-bugzilla-daemon at ghostscript.com
Tue Jul 22 10:37:33 PDT 2008


http://bugs.ghostscript.com/show_bug.cgi?id=689966





------- Additional Comments From ray.johnston at artifex.com  2008-07-22 10:37 -------
Created an attachment (id=4232)
 --> (http://bugs.ghostscript.com/attachment.cgi?id=4232&action=view)
c329.log.gz

Some of these pages are MUCH slower than others (see below). The total time
on my machine with MAX_PATTERN_BITMAP_SIZE = 32Mb (rev >= 8851).

In particular Page 19 trips over the memory based clist threshold for doing
compression (my build had BAND_LIST_STORAGE=memory).

I think we need to understand these pages, but the original REALLY excessive
(hours) time issue is now fixed.

Page 8:
%  done : time = 14.234, pages = 7
[:] Width = 5100,	Height = 6600
[:]clist_end_page at cfile=338645651, bfile=7416
[:]tot_raw=338653043, tot_compressed=0
[:]width=5100, band_width=5100, band_height=1251, nbands=6
%  done : time = 85.516, pages = 8

Page 19:
%  done : time = 108.266, pages = 18
[:]Beginning compression
[:] Width = 5100,	Height = 6600
[:]clist_end_page at cfile=608688708, bfile=9936
[:]Number of raw buffers allocated=65
[:]tot_raw=608698620, tot_compressed=13734762
[:]width=5100, band_width=5100, band_height=1251, nbands=6
%  done : time = 329.234, pages = 19

Page 23:
[:]width=5100, band_width=5100, band_height=1251, nbands=6
%  done : time = 354.594, pages = 22
[:] Width = 5100,	Height = 6600
[:]clist_end_page at cfile=340487816, bfile=3864
[:]tot_raw=340491656, tot_compressed=0
[:]width=5100, band_width=5100, band_height=1251, nbands=6
%  done : time = 423.234, pages = 23

Page 24:
%  done : time = 423.234, pages = 23
[:] Width = 5100,	Height = 6600
[:]clist_end_page at cfile=470047107, bfile=5304
[:]tot_raw=470052387, tot_compressed=0
[:]width=5100, band_width=5100, band_height=1251, nbands=6
%  done : time = 572.797, pages = 24

--------------------------------------------------

It's a side issue, but the code also leaks like a sieve. There were a 1477
objects 'leaked' during this 62 page test, total 700 Kbytes. This is partly
captured in another bug assigned to Igor (bug 689822).



------- You are receiving this mail because: -------
You are the QA contact for the bug, or are watching the QA contact.



More information about the gs-bugs mailing list