[Coin-discuss] Cbc and miplib3
John J Forrest
jjforre at us.ibm.com
Mon Dec 20 12:22:17 EST 2004
Sven,
Helped by your hints I have been playing with different optimization
levels and debug. If I use -g and NDEBUG is not defined (so asserts are
all active) everything seems fine. If I use -On and NDEBUG is defined all
my tests work. If I use -On and NDEBUG is not defined then I did manage
to trigger some asserts e.g assert (theta>=0.0) when theta was -1.0e-15. I
presume that this is due to the arithmetic being done in two different
ways when optimization is on.
So I have removed some of these asserts so that should help.
There were one or two places were it seemed there was an abort without any
assert. I could only find one likely place for that to happen and it
could not occur if NDEBUG was defined. I have added a message so you can
tell me if that was the problem - then I can look further.
By the way for Clp and Coin as a whole there is a slight performance
penalty if NDEBUG is not defined. the latest Makefile.Linux sets NDEBUG
if any optimization is done which may explain some of the difference
between Linux and Sun/Mac.
keep me informed how things go.
John Forrest
Sven de Vries <devries at ma.tum.de>
Sent by: coin-discuss-bounces at list.coin-or.org
12/19/2004 08:51 AM
Please respond to
Discussions about open source software for Operations Research
To
Discussions about open source software for Operations Research
<coin-discuss at list.coin-or.org>
cc
Subject
[Coin-discuss] Cbc and miplib3
Hi,
since last week it seems that Cbc aborts (without further explanation) on
lots of miplib3 problems (I observe this on sun and macosx).
(using the runtime script from Sbb/Samples...)
sven
_______________________________________________
Coin-discuss mailing list
Coin-discuss at list.coin-or.org
http://list.coin-or.org/mailman/listinfo/coin-discuss
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.coin-or.org/pipermail/coin-discuss/attachments/20041220/c4c991f1/attachment.html>
More information about the Coin-discuss
mailing list