Open in new window / Try shogun cloud
--- Log opened Sun Sep 11 00:00:17 2011
shogun-buildbotbuild #209 of csharp_modular is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/csharp_modular/builds/209  blamelist: blackburn91@gmail.com00:11
shogun-buildbotbuild #218 of octave_modular is complete: Failure [failed compile]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_modular/builds/218  blamelist: blackburn91@gmail.com00:15
blackburn Speicherzugriffsfehler00:16
blackburnI guess it is something like PieceOfShitError?00:16
blackburn:D00:16
shogun-buildbotbuild #219 of octave_modular is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_modular/builds/21900:25
shogun-buildbotbuild #198 of ruby_modular is complete: Failure [failed test_1]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/ruby_modular/builds/198  blamelist: blackburn91@gmail.com01:02
shogun-buildbotbuild #199 of ruby_modular is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/ruby_modular/builds/19901:16
-!- blackburn [~blackburn@31.28.44.65] has quit [Quit: Leaving.]01:18
shogun-buildbotbuild #210 of csharp_modular is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/csharp_modular/builds/21001:26
-!- blackburn [~blackburn@31.28.44.65] has joined #shogun10:33
@sonney2kblackburn, yes12:05
CIA-3shogun: Soeren Sonnenburg master * rb1edffb / src/shogun/kernel/LocalAlignmentStringKernel.cpp : use CALLOC instead of MALLOC - http://git.io/Xo1Xag20:25
CIA-3shogun: Soeren Sonnenburg master * r73e9943 / examples/undocumented/python_modular/graphical/svm.py : use axis tight to improve figure - http://git.io/05RWKg20:25
blackburnsonney2k: wow did you fix that?20:26
@sonney2kblackburn, did you already commit your fast LLE version?20:26
blackburnsonney2k: yes20:26
@sonney2kblackburn, don't know20:26
blackburnsonney2k: I've spent the whole day trying to get packing numbers work nicely, no success yet20:26
blackburnnow continued integrating superlu20:27
@sonney2kI don't know what packing numbers are20:27
blackburnsonney2k: not important, just told you what is I'm working on :)20:28
shogun-buildbotbuild #219 of r_static is complete: Failure [failed git]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/r_static/builds/219  blamelist: sonne@debian.org20:29
shogun-buildbotbuild #217 of octave_static is complete: Failure [failed git]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_static/builds/217  blamelist: sonne@debian.org20:29
shogun-buildbotbuild #231 of cmdline_static is complete: Failure [failed git]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/cmdline_static/builds/231  blamelist: sonne@debian.org20:29
shogun-buildbotbuild #212 of python_static is complete: Failure [failed git]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_static/builds/212  blamelist: sonne@debian.org20:29
blackburnsonney2k: local alignment still failing20:51
@sonney2kblackburn, with which version did you compare last?20:52
blackburn0.9.020:52
@sonney2kI mean did it work in 0.8.sth?20:52
blackburnsonney2k: I didn't test it20:53
@sonney2kblackburn, I guess it might be sufficient to do a git diff agains 0.820:53
shogun-buildbotbuild #13 of nightly_default is complete: Success [build successful]  Build details are at http://www.shogun-toolbox.org/buildbot/builders/nightly_default/builds/1321:52
blackburnsonney2k: lle is really fast now - 223s for 10000 vectors22:01
blackburnbut superlu to go22:01
@sonney2kso is it up to par to scikits now?22:02
@sonney2kor faster?22:02
blackburnsonney2k: much faster22:02
@sonney2khow fast is scikits?22:02
blackburnwell I didn't test on 10k but on 8k it was 2 times slower22:02
blackburnsonney2k: hidden feature of my impl is parallelization - it can really impact on high-dimensional data22:03
blackburnI got crazy with superlu, kind of complicated thing22:04
blackburnsonney2k: 3k examples:22:07
blackburnSHOGUN Took 9.007753s22:07
blackburnSKLEARN Took 11.569473s22:07
@sonney2kabout the same22:09
blackburnsonney2k: wait a minute for 6k22:09
@sonney2kblackburn, if you write a paper you should have a figure like this showing how much faster your stuff is somewhere (potentially even in shogun docs for dim red stuff)22:10
blackburnI know22:10
blackburnsonney2k: I will improve perfomance with superlu, then it would be faster, may be significantly22:11
blackburnSKLEARN Took 91.296676s22:11
blackburnSHOGUN Took 61.362044s22:11
blackburnsonney2k: ^ not really the same22:11
blackburnand now I turn on parallel things22:12
blackburnSHOGUN Took 50.653053s22:13
@sonney2k2 cores?22:13
blackburnyes22:14
blackburnI believe it would be faster on really multiprocessor machine22:14
@sonney2kwhat is the bottleneck currently?22:14
blackburnsonney2k: sparse dot and eigenproblem almost the same22:15
blackburnneighbor searching is fast22:15
@sonney2ksparse dot is parallel?22:15
blackburnyes22:15
@sonney2kbut eigenproblem is not right?22:15
blackburneigenproblem is not22:15
@sonney2kis this with lapack?22:15
blackburnarpack22:15
@sonney2kahh - is arpack much faster?22:15
blackburnyes, twice22:16
blackburnscikits uses arpack too22:16
@sonney2k(because there is the parallel scalapack)22:16
@sonney2kok22:16
blackburnsonney2k: arpack could impact from parallel lapack22:16
blackburnit uses lapack ops22:16
blackburne.g. it provides a pointer to a vector you should product with matrix :)22:17
blackburnor solve the system22:17
blackburnnow I'm using LU/Cholesky decomposition from lapack22:17
blackburnit is pretty fast but doesn't take any advantages of matrix being \sparse22:18
blackburnsonney2k: well finally, arpack have parallel version too :)22:19
blackburnP_ARPACK22:19
blackburnno idea how to work with it though22:19
blackburnit lacks any docs22:19
@sonney2knite22:31
blackburnsee you22:32
-!- blackburn [~blackburn@31.28.44.65] has quit [Quit: Leaving.]23:01
--- Log closed Mon Sep 12 00:00:22 2011