Hi all
I am a new user to ALPS. Unfortunately I am running into some difficulties with "make tests" for the libraries. Basically one of the tests is failing:
[fixed_capacity] ./fixed_capacity_vector | diff fixed_capacity_vector.output - PASS: fixed_capacity/fixed_capacity_vector [fixed_capacity] ./test_deque | diff test_deque.output - PASS: fixed_capacity/test_deque [fixed_capacity] ./test_vector | diff test_vector.output - 1d0 < Test done successfully. FAIL: fixed_capacity/test_vector
The other 39 seem to pass well enough though :)
I am using:
Alps 1.2.2 gcc 3.3.5 on Linux lp_solve 4.0 boost 1_32 (I run into compile problems with 1_33_1) Opteron
I am just testing ALPS right now so I have compiled it without support for MPI and hdf5.
Is it OK to ignore this failed test? If not is there anything I can do e.g. compile options to get the test to past?
Thanks very much for your help and for ALPS!
Pearl Louis
Dear Pearl Louis,
That seems to me not a bug in the libraries, but a result of optimization by the compiler unforseen in the test suite. I'm almost certain that you can safely ignore this failed test.
Meanwhile I would like to ask you to help us for solving this failure. Could you run the test by hand and see the output? I.e.,
$ cd test/fixed_capacity $ ./test_vector
Thank you in advance for your cooperation.
Best regards, Synge Todo
From: Pearl Louis louis@sci.osaka-cu.ac.jp Date: Tue, 10 Oct 2006 17:07:01 +0900
Hi all
I am a new user to ALPS. Unfortunately I am running into some difficulties with "make tests" for the libraries. Basically one of the tests is failing:
[fixed_capacity] ./fixed_capacity_vector | diff fixed_capacity_vector.output - PASS: fixed_capacity/fixed_capacity_vector [fixed_capacity] ./test_deque | diff test_deque.output - PASS: fixed_capacity/test_deque [fixed_capacity] ./test_vector | diff test_vector.output - 1d0 < Test done successfully. FAIL: fixed_capacity/test_vector
The other 39 seem to pass well enough though :)
I am using:
Alps 1.2.2 gcc 3.3.5 on Linux lp_solve 4.0 boost 1_32 (I run into compile problems with 1_33_1) Opteron
I am just testing ALPS right now so I have compiled it without support for MPI and hdf5.
Is it OK to ignore this failed test? If not is there anything I can do e.g. compile options to get the test to past?
Thanks very much for your help and for ALPS!
Pearl Louis
Dear Synge,
Thank-you for the reply. If it is about compiler optimizations I am running the stock-standard gcc that comes with SUSE 9.3 Professional for 64-bit systems, gcc-3.3.5-5. It is a dual opteron system.
I ran the test by hand and unfortunately I am not able to tell you much because the only output was "Aborted".
I had a quick look through the config.log produced by the ALPS configure command. According to it, the SUSE gcc is compiled with the following options:
Reading specs from /usr/lib64/gcc-lib/x86_64-suse-linux/3.3.5/specs Configured with: ../configure --enable-threads=posix --prefix=/usr --with-local-prefix=/usr/local --infodir=/usr/share/info --mandir=/us r/share/man --enable-languages=c,c++,f77,objc,java,ada --disable-checking --libdir=/usr/lib64 --enable-libgcj --with-slibdir=/lib64 --wi th-system-zlib --enable-shared --enable-__cxa_atexit x86_64-suse-linux
It looks like the compile options that ./configure ended up with are:
CC='gcc' CCDEPMODE='depmode=gcc3' CFLAGS='-pthread -w -O3' COMPILER='gnu' CPP='gcc -E' CPPFLAGS=' -DNDEBUG' CXX='g++' CXXCPP='g++ -E' CXXDEPMODE='depmode=gcc3' CXXFLAGS='-pthread -w -ftemplate-depth-150 -O3 -fexceptions' CYGPATH_W='echo' DEFS='-DHAVE_CONFIG_H' DEPDIR='.deps' ECHO='echo' ECHO_C='' ECHO_N='-n' ECHO_T='' EGREP='grep -E'
Thanks for your help! Pearl
---------- Forwarded Message ----------
Subject: Re: [ALPS-users] Alps 1.2.2: make tests fails on fixed_capacity/test_vector Date: Wed October 11 2006 10:17 am From: Synge Todo wistaria@comp-phys.org To: louis@sci.osaka-cu.ac.jp Cc: comp-phys-alps-users@phys.ethz.ch
Dear Pearl Louis,
That seems to me not a bug in the libraries, but a result of optimization by the compiler unforseen in the test suite. I'm almost certain that you can safely ignore this failed test.
Meanwhile I would like to ask you to help us for solving this failure. Could you run the test by hand and see the output? I.e.,
$ cd test/fixed_capacity $ ./test_vector
Thank you in advance for your cooperation.
Best regards, Synge Todo
From: Pearl Louis louis@sci.osaka-cu.ac.jp Date: Tue, 10 Oct 2006 17:07:01 +0900
Hi all
I am a new user to ALPS. Unfortunately I am running into some difficulties with "make tests" for the libraries. Basically one of the tests is failing:
[fixed_capacity] ./fixed_capacity_vector | diff fixed_capacity_vector.output - PASS: fixed_capacity/fixed_capacity_vector [fixed_capacity] ./test_deque | diff test_deque.output - PASS: fixed_capacity/test_deque [fixed_capacity] ./test_vector | diff test_vector.output - 1d0 < Test done successfully. FAIL: fixed_capacity/test_vector
The other 39 seem to pass well enough though :)
I am using:
Alps 1.2.2 gcc 3.3.5 on Linux lp_solve 4.0 boost 1_32 (I run into compile problems with 1_33_1) Opteron
I am just testing ALPS right now so I have compiled it without support for MPI and hdf5.
Is it OK to ignore this failed test? If not is there anything I can do e.g. compile options to get the test to past?
Thanks very much for your help and for ALPS!
Pearl Louis
Dear Pearl,
From: Pearl Louis louis@sci.osaka-cu.ac.jp Date: Wed, 11 Oct 2006 10:52:32 +0900
I ran the test by hand and unfortunately I am not able to tell you much because the only output was "Aborted".
Mmm... Then, could you add the following line at the very beginning of test/fixed_capacity/test_vector.C
#define VORBOSE
and compile and run it again?
cd test/fixed_capacity make test_vector ./test_vector
Thank you again for you cooperation.
Best, Synge
Dear Synge,
I dumped the output to file but it's 2.3MB so I won't post it to the mailing list. Would you like me to send it to you directly?
The results of using #define VERBOSE are quite interesting. Not only does it output a whole lot of data as expected but for some reason test_vector starts working! From the end of the output:
[quote] 9997 insert sequence (0.587807,0.587807,0.587807) (0.299336,0.299336,0.299336) (0.126336,0.126336,0.126336) (0.148921,0.148921,0.148921) (0.621902,0.621902,0.621902) (0.356793,0.356793,0.356793) (0.12183,0.12183,0.12183) (0.847102,0.847102,0.847102) (0.154197,0.154197,0.154197) (0.528198,0.528198,0.528198) (0.386415,0.386415,0.386415) (0.653625,0.653625,0.653625) (0.35809,0.35809,0.35809) (0.722514,0.722514,0.722514) (0.212357,0.212357,0.212357) 9998 clear
9999 erase 0 0
Test done successfully.[/quote]
Very strange results for a simple verbose option. If I remove the #define VERBOSE and recompile, test_vector fails again.
Thanks, Pearl
---------- Forwarded Message ----------
Subject: Re: [ALPS-users] Fwd: Re: Alps 1.2.2: make tests fails on fixed_capacity/test_vector Date: Wed October 11 2006 11:25 am From: Synge Todo wistaria@comp-phys.org To: louis@sci.osaka-cu.ac.jp Cc: comp-phys-alps-users@phys.ethz.ch
Dear Pearl,
From: Pearl Louis louis@sci.osaka-cu.ac.jp Date: Wed, 11 Oct 2006 10:52:32 +0900
I ran the test by hand and unfortunately I am not able to tell you much because the only output was "Aborted".
Mmm... Then, could you add the following line at the very beginning of test/fixed_capacity/test_vector.C
#define VORBOSE
and compile and run it again?
cd test/fixed_capacity make test_vector ./test_vector
Thank you again for you cooperation.
Best, Synge
-------------------------------------------------------
Dear Pearl,
From: Pearl Louis louis@sci.osaka-cu.ac.jp Date: Wed, 11 Oct 2006 11:48:40 +0900
I dumped the output to file but it's 2.3MB so I won't post it to the mailing list. Would you like me to send it to you directly?
No, thank you :-)
Very strange results for a simple verbose option. If I remove the #define VERBOSE and recompile, test_vector fails again.
That suggests a bug in gcc. So please ignore the test failure, or consider for updating your gcc if it bothers you.
Thanks, Synge
comp-phys-alps-users@lists.phys.ethz.ch