[Coin-discuss] Osi unitTest core dumps / AMPL driver
Leo Lopes
leo at iems.northwestern.edu
Thu Feb 27 00:11:48 EST 2003
Thanks Guys. Great work.
One small thing. I actually compiled the library with libz and libgz
under cygwin, and when I got a crash there it was bc the Netlib module
was being downloaded to another place. I don't know enough about COIN
yet to know if that is a feature or a bug, but just the heads up...
Thx again.
Leo.
Matthew Saltzman wrote:
> On Wed, 26 Feb 2003, Laszlo Ladanyi wrote:
>
>
>>Hi Leo,
>>
>>I have found the problem and fixed it. Basically, a compiler flag
>>(-fomit-frame-pointer) creeped into COIN/Makefiles/Makefile.Linux and that
>>caused the crash on RH8 w/ g++-3.2. With g++-3.2.2 it does not seem to be a
>>problem on Debian. Strange...
>
>
> I didn't see any obvious reference to this in the gcc release notes or in
> Red Hat's Bugzilla, so yes, it's a bit strange.
>
> It does fix the Coin unit test failure. It doesn't fix the segfaults if
> you don't include libz or gunzip the MPS library.
>
>
>>So I took out all extra optimization flags and set the optimization to -O6. We
>>should probably keep the repository copy clean, and then everyone is welcome
>>to tweak compiler flags in their copy.
>
>
> Yes, it does seem to be easy to screw this up. We must be careful not to
> accidentally update Makefiles. If anyone has an idea that would be more
> reliable than just trying to pay more attention, that would be helpful.
>
>
>>I'm not sure how much extra speed one gets from these special flags anyways.
>>However, I'd be definitely interested if someone decides to play with them :-).
>>The following flags are of interest:
>>-fomit-frame-pointer -ffast-math -fno-math-errno -funsafe-math-optimizations
>>-fno-trapping-math
>>
>>--Laci
>>
>>On Tue, 25 Feb 2003, Leonardo B Lopes wrote:
>>
>>
>>>Sorry if you get this message twice. I sent it last night from a
>>>different computer and there was an email messup.
>>>
>>>Thanks for the quick response. The Platform is RH Linux 8,
>>>g++ 3.2, libc-2.2.93.
>>>
>>>I just tried rebuilding from scratch and got the same results. I also
>>>tried the exact same thing on my windows/cygwin box and everything seems
>>>fine, except that the Netlib collection was not downloaded, and when I
>>>checked out the netlib module, it got into the wrong place. I then
>>>copied it to where /Osi/unitTest expected it to be, and it seems to be
>>>OK. Anyone able to build it on RH8?
>>>
>>>TIA,
>>>Leo.
>>>
>>>JP Fasano wrote:
>>>
>>>>
>>>>
>>>>I just grabbed the latest code from cvs, built, and ran
>>>>COIN\Osi\Test\unitTest.cpp.
>>>>It is running ok for myself.
>>>>What platform are you running on?
>>>>I did this test on windows using MS compiler.
>>>>
>>>>Here is the output:
>>>>--------------------------
>>>>Testing OsiRowCut with OsiClpSolverInterface
>>>>Testing OsiColCut with OsiClpSolverInterface
>>>>Testing OsiRowCutDebugger with OsiClpSolverInterface
>>>>Testing OsiSimplexInterface with OsiClpSolverInterface
>>>>Testing OsiCuts
>>>>Testing OsiClpSolverInterface
>>>>Testing OsiSolverInterface
>>>> processing mps file: 25fv47 (1 out of 90)
>>>>clpSolverInterface 5501.85 = 5501.85 ; okay - took 10.595 seconds.
>>>> processing mps file: 80bau3b (2 out of 90)
>>>>clpSolverInterface 987224 = 987224 ; okay - took 8.293 seconds.
>>>> processing mps file: adlittle (3 out of 90)
>>>>clpSolverInterface 225495 = 225495 ; okay - took 0.08 seconds.
>>>> processing mps file: afiro (4 out of 90)
>>>>clpSolverInterface -464.753 = -464.753 ; okay - took 0.04 seconds.
>>>> processing mps file: agg (5 out of 90)
>>>>clpSolverInterface -3.59918e+007 = -3.59918e+007 ; okay - took 0.211
>>>>seconds.
>>>> processing mps file: agg2 (6 out of 90)
>>>>clpSolverInterface -2.02393e+007 = -2.02393e+007 ; okay - took 0.28
>>>>seconds.
>>>> processing mps file: agg3 (7 out of 90)
>>>>clpSolverInterface 1.03121e+007 = 1.03121e+007 ; okay - took 0.31 seconds.
>>>> processing mps file: bandm (8 out of 90)
>>>>clpSolverInterface -158.628 = -158.628 ; okay - took 0.471 seconds.
>>>> processing mps file: beaconfd (9 out of 90)
>>>>clpSolverInterface 33592.5 = 33592.5 ; okay - took 0.12 seconds.
>>>> processing mps file: blend (10 out of 90)
>>>>clpSolverInterface -30.8121 = -30.8121 ; okay - took 0.081 seconds.
>>>> processing mps file: bnl1 (11 out of 90)
>>>>clpSolverInterface 1977.63 = 1977.63 ; okay - took 1.732 seconds.
>>>> processing mps file: bnl2 (12 out of 90)
>>>>clpSolverInterface 1811.24 = 1811.24 ; okay - took 2.884 seconds.
>>>> processing mps file: boeing1 (13 out of 90)
>>>>clpSolverInterface -335.214 = -335.214 ; okay - took 0.621 seconds.
>>>> processing mps file: boeing2 (14 out of 90)
>>>>clpSolverInterface -315.019 = -315.019 ; okay - took 0.131 seconds.
>>>> processing mps file: bore3d (15 out of 90)
>>>>clpSolverInterface 1373.08 = 1373.08 ; okay - took 0.12 seconds.
>>>> processing mps file: brandy (16 out of 90)
>>>>clpSolverInterface 1518.51 = 1518.51 ; okay - took 0.31 seconds.
>>>> processing mps file: capri (17 out of 90)
>>>>clpSolverInterface 2690.01 = 2690.01 ; okay - took 0.24 seconds.
>>>> processing mps file: cycle (18 out of 90)
>>>>clpSolverInterface -5.22639 = -5.22639 ; okay - took 1.583 seconds.
>>>> processing mps file: czprob (19 out of 90)
>>>>clpSolverInterface 2.1852e+006 = 2.1852e+006 ; okay - took 1.292 seconds.
>>>> processing mps file: d2q06c (20 out of 90)
>>>>....
>>>>
>>>>
>>>>JP Fasano
>>>>jpfasano at us.ibm.com
>>>>(914)945-1324 (tie line 862-1324)
>>>>COIN www.coin-or.org
>>>>
>>>>
>>>>|---------+------------------------------------------------>
>>>>| | Leonardo B Lopes |
>>>>| | <leo at iems.northwestern.edu> |
>>>>| | Sent by: |
>>>>| | coin-discuss-admin at www-124.southbury.|
>>>>| | usf.ibm.com |
>>>>| | |
>>>>| | |
>>>>| | 02/24/2003 08:57 PM |
>>>>| | Please respond to coin-discuss |
>>>>| | |
>>>>|---------+------------------------------------------------>
>>>> >----------------------------------------------------------------------------------------------|
>>>> | |
>>>> | To: coin-discuss at www-124.southbury.usf.ibm.com |
>>>> | cc: |
>>>> | Subject: [Coin-discuss] Osi unitTest core dumps / AMPL driver |
>>>> | |
>>>> >----------------------------------------------------------------------------------------------|
>>>>
>>>>
>>>>
>>>>
>>>>I just downloaded Osi, Clp, Cgl, and Sbb (2/24 ~ 18:30CT). Did my best
>>>>to follow the instructions, but Osi unitTest aborts with a core dump at
>>>>OsiRowCutTest.cpp:246.
>>>>
>>>>Here is the program output:
>>>>
>>>>-----------------------------------------------------------
>>>>[leo at senna Osi]$ ./unitTest
>>>>Testing OsiRowCut with OsiClpSolverInterface
>>>>Coin0001I At line 29 NAME EXAMPLE
>>>>Coin0001I At line 30 ROWS
>>>>Coin0001I At line 37 COLUMNS
>>>>Coin0001I At line 55 RHS
>>>>Coin0001I At line 61 RANGES
>>>>Coin0001I At line 64 BOUNDS
>>>>Coin0001I At line 70 ENDATA
>>>>Coin0002I Problem EXAMPLE has 5 rows, 8 columns and 14 elements
>>>>Coin0008I EXAMPLE read with 0 errors
>>>>Aborted (core dumped)
>>>>-----------------------------------------------------------
>>>>
>>>>and here is the section of Makefile.location I changed:
>>>>
>>>>-----------------------------------------------------------
>>>>CoinDir ?= ${HOME}/devel/fml/COIN
>>>>
>>>>CoinLibsDefined :=
>>>>
>>>>CoinLibsDefined += COIN_libCoin
>>>>CoinLibsDefined += COIN_libOsi
>>>>CoinLibsDefined += COIN_libCgl
>>>>
>>>># CoinLibsDefined += COIN_libVol
>>>># CoinLibsDefined += COIN_libOsiVol
>>>>
>>>>CoinLibsDefined += COIN_libClp
>>>>CoinLibsDefined += COIN_libOsiClp
>>>>
>>>>CoinLibsDefined += COIN_libSbb
>>>>
>>>># CoinLibsDefined += COIN_libCpx
>>>># CoinLibsDefined += COIN_libOsiCpx
>>>>-----------------------------------------------------------
>>>>
>>>>Any help would be much appreciated. There was a question earlier about
>>>>an AMPL driver. At the end of my current project a simple AMPL driver
>>>>should be available.
>>>>
>>>>Cheers,
>>>>Leo.
>>>>--
>>>>=======================================================================
>>>>Leonardo B. Lopes leo at iems.nwu.edu
>>>>Ph.D. Candidate (847)491-8470
>>>>IEMS - Northwestern University http://www.iems.nwu.edu/~leo
>>>>
>>>>_______________________________________________
>>>>Coin-discuss mailing list
>>>>Coin-discuss at www-124.ibm.com
>>>>http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-discuss
>>>>
>>>>
>>>>
>>>>_______________________________________________
>>>>Coin-discuss mailing list
>>>>Coin-discuss at www-124.ibm.com
>>>>http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-discuss
>>>
>>>
>>>--
>>>=======================================================================
>>>Leonardo B. Lopes leo at iems.nwu.edu
>>>Ph.D. Candidate (847)491-8470
>>>IEMS - Northwestern University http://www.iems.northwestern.edu/~leo
>>>
>>>
>>>_______________________________________________
>>>Coin-discuss mailing list
>>>Coin-discuss at www-124.ibm.com
>>>http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-discuss
>>>
>>
>>
>>_______________________________________________
>>Coin-discuss mailing list
>>Coin-discuss at www-124.ibm.com
>>http://www-124.ibm.com/developerworks/oss/mailman/listinfo/coin-discuss
>>
>
>
--
=======================================================================
Leonardo B. Lopes leo at iems.nwu.edu
Ph.D. Candidate (847)491-8470
IEMS - Northwestern University http://www.iems.northwestern.edu/~leo
More information about the Coin-discuss
mailing list