A few build and dependency problems with 2.2.alpha

[justapost]

Ideally on the mailing list (http://phoronix-test-suite.com/mailm...test-suite.com) or to email to michael [at] phoronix.com.

On it's way to the mailinglist.
Unfortunately I found one small error, install.sh needs an backslash infront of $NUM_CPU_CORES in line 231

[Michael]

Ideally on the mailing list (http://phoronix-test-suite.com/mailm...test-suite.com) or to email to michael [at] phoronix.com.

[justapost]

Ok I got it working with result caching, who can I submit the profile?

[Michael]

Hmmm, you mean as in the cache feature in 2.2? It should work if you write it out during the parse-results process. Try it out and let me know but it should

[justapost]

Of course I'll send you the profile. Thank's for the tip units work fine now.
Hmm, it not work with cached results. Can pts-results-scale be generated during parsing?

[Michael]

When you're done, you're welcome to submit the test profile to inclusion in mainline

Yes, during the testing process you can write the units to ~/pts-results-scale during the testing process. If no scale is defined within the XML spec, pts-core will read it from pts-results-scale. To Git master I have just committed the same support similarly for pts-results-proportion, but pts-results-scale has been supported for a release or two.

[justapost]

Thanks, will work on some changes for that. Patches also welcome

Have you been working on some test profiles or something? You seem to be running PTS a lot lately (or at least judging by the posts)?

I started with an new test for hpcc today. http://icl.cs.utk.edu/hpcc/index.html
Atm I use an static config suitable for four cores and I build against atlas and openmpi coming with debian. These tests include linpack and I'm thinking about a good way to calculate the problem size based on the available memory and a way to calculate an good grid size based on the number of cpu's and cores.
The benchmark results end up in an text file which I copy to $LOGFILE. Then I use a slightly modified parsing from the stream benchmark.
The results have different units like GFlops, GUPs,TFlops,GB/s and seconds. Is it possible to define a separate unit and proportion for each test result or can this be defined only once for an test?

[Michael]

Thanks, will work on some changes for that. Patches also welcome

Have you been working on some test profiles or something? You seem to be running PTS a lot lately (or at least judging by the posts)?

[justapost]

This is what I get when I try to build a 32bit version of tachyon

gcc -m32 -Wall -O3 -fomit-frame-pointer -ffast-math -DLinux -DTHR -D_REENTRANT        -I../src -I../demosrc -o ../compile/linux-thr/tachyon ../compile/linux-thr/libtachyon/main.o ../compile/linux-thr/libtachyon/getargs.o ../compile/linux-thr/libtachyon/parse.o ../compile/linux-thr/libtachyon/nffparse.o ../compile/linux-thr/libtachyon/glwin.o ../compile/linux-thr/libtachyon/spaceball.o ../compile/linux-thr/libtachyon/trackball.o ../compile/linux-thr/libtachyon/mgfparse.o ../compile/linux-thr/libtachyon/ac3dparse.o  -L../compile/linux-thr   -L. -ltachyon    -lm -lpthread
/usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/4.3.4/libgcc.a when searching for -lgcc
/usr/bin/ld: skipping incompatible /usr/lib/gcc/x86_64-linux-gnu/4.3.4/libgcc.a when searching for -lgcc
/usr/bin/ld: cannot find -lgcc
collect2: ld returned 1 exit status

Seems like you have an 32bit gcc build envirnoment installed on ubuntu.

That patch I posted for yafray is part of the current debian unstable source package (yafray-0.0.9+dfsg).
Errors without the patch

g++ -o src/yafraycore/bound.os -c -Wall -DHAVE_CONFIG_H -D_PTHREADS -O3 -ffast-math -fomit-frame-pointer -fPIC -Isrc/yafraycore -I. src/yafraycore/bound.cc
In file included from src/yafraycore/bound.h:30,
                 from src/yafraycore/bound.cc:26:
src/yafraycore/vector3d.h: In function ‘int yafray::ourRandomI()’:
src/yafraycore/vector3d.h:267: warning: integer overflow in expression
src/yafraycore/vector3d.h: In function ‘yafray::PFLOAT yafray::ourRandom()’:
src/yafraycore/vector3d.h:279: warning: integer overflow in expression
src/yafraycore/vector3d.h: In function ‘yafray::PFLOAT yafray::ourRandom(int&)’:
src/yafraycore/vector3d.h:291: warning: integer overflow in expression
g++ -o src/yafraycore/buffer.os -c -Wall -DHAVE_CONFIG_H -D_PTHREADS -O3 -ffast-math -fomit-frame-pointer -fPIC -Isrc/yafraycore -I. src/yafraycore/buffer.cc
In file included from src/yafraycore/buffer.cc:22:
src/yafraycore/buffer.h: In constructor ‘yafray::gBuf_t<T1, T2>::gBuf_t(int, int)’:
src/yafraycore/buffer.h:46: error: there are no arguments to ‘exit’ that depend on a template parameter, so a declaration of ‘exit’ must be available
src/yafraycore/buffer.h:46: error: (if you use ‘-fpermissive’, G++ will accept your code, but allowing the use of an undeclared name is deprecated)
src/yafraycore/buffer.h: In member function ‘void yafray::gBuf_t<T1, T2>::set(int, int)’:
src/yafraycore/buffer.h:62: error: there are no arguments to ‘exit’ that depend on a template parameter, so a declaration of ‘exit’ must be available
src/yafraycore/buffer.h: In constructor ‘yafray::Buffer_t<T>::Buffer_t(int, int)’:
src/yafraycore/buffer.h:156: error: there are no arguments to ‘exit’ that depend on a template parameter, so a declaration of ‘exit’ must be available
src/yafraycore/buffer.h: In member function ‘void yafray::Buffer_t<T>::set(int, int)’:
src/yafraycore/buffer.h:178: error: there are no arguments to ‘exit’ that depend on a template parameter, so a declaration of ‘exit’ must be available
scons: *** [src/yafraycore/buffer.os] Error 1
scons: building terminated because of errors.
cp: Aufruf von stat f?r „yafray_/etc/gram.yafray“ nicht m?glich: Datei oder Verzeichnis nicht gefunden

And I get this error for jgxbat without JAVA_HOME set to /usr/share/doc/sun-java6-jdk in my case with sun-java6-demo installed.

###########################################################
Java Graphics Basic Acceptance Test:                       
Phoronix Test Suite v2.2.0a2                               


Final: java full version 1.6.0_16-b01
Unable to access jarfile /usr/demo/jfc/Java2D/Java2Demo.jar
Unable to access jarfile /usr/demo/jfc/Java2D/Java2Demo.jar
FAIL ()

[Michael]

Hmm, Tachyon is running fine here on 64-bit Ubuntu without that command.

With Yafray I am also hitting this on the build process:

src/yafraycore/buffer.h: In constructor ‘yafray::gBuf_t<T1, T2>::gBuf_t(int, int)’:
src/yafraycore/buffer.h:46: error: there are no arguments to ‘exit’ that depend on a template parameter, so a declaration of ‘exit’ must be available
src/yafraycore/buffer.h:46: error: (if you use ‘-fpermissive’, G++ will accept your code, but allowing the use of an undeclared name is deprecated)
src/yafraycore/buffer.h: In member function ‘void yafray::gBuf_t<T1, T2>::set(int, int)’:
src/yafraycore/buffer.h:62: error: there are no arguments to ‘exit’ that depend on a template parameter, so a declaration of ‘exit’ must be available
src/yafraycore/buffer.h: In constructor ‘yafray::Buffer_t<T>::Buffer_t(int, int)’:
src/yafraycore/buffer.h:156: error: there are no arguments to ‘exit’ that depend on a template parameter, so a declaration of ‘exit’ must be available
src/yafraycore/buffer.h: In member function ‘void yafray::Buffer_t<T>::set(int, int)’:
src/yafraycore/buffer.h:178: error: there are no arguments to ‘exit’ that depend on a template parameter, so a declaration of ‘exit’ must be available