[Symphony-tickets] [SYMPHONY] #121: wrong optimum with bdver1 or bdver2 options
SYMPHONY
coin-trac at coin-or.org
Mon Aug 5 19:56:08 EDT 2013
#121: wrong optimum with bdver1 or bdver2 options
---------------------------+-------------------
Reporter: charlesBrixko | Owner: tkr
Type: defect | Status: new
Priority: normal | Milestone:
Component: Not Sure | Version: 5.5.0
Keywords: |
---------------------------+-------------------
I tested the file "newdano.mps" from MIPLIB2010 on Fedora 19 64 bits with
gcc 4.8.1.
The correct optimal is 65.666667
With Symphony-5.5.0 compiled without environment variable definition, the
correct optimum is found in 23159 sec. on a 8350.
With Symphony-5.5.0 compiled with ADD_CFLAGS="-march=bdver2" and
ADD_CXXFLAGS="-march=bdver2", SYMPHONY-5.5.0 finds 66.272727 in 2335 sec.
on a 8350.
With Symphony-5.5.0 compiled with ADD_CFLAGS="-march=bdver1" and
ADD_CXXFLAGS="-march=bdver1", SYMPHONY-5.5.0 finds 66.272727 in 2676 sec.
on a 8150.
When giving a wrong value, I noticed that the Lower Bound goes higher than
the correct optimum earlier than the end.
--
Ticket URL: <https://projects.coin-or.org/SYMPHONY/ticket/121>
SYMPHONY <http://projects.coin-or.org/SYMPHONY>
The SYMPHONY framework for parallel branch-and-cut algorithms.
More information about the Symphony-tickets
mailing list