--- Log opened Sat Aug 20 00:00:06 2011 | ||
CIA-87 | shogun: Soeren Sonnenburg master * r57c1c84 / examples/undocumented/java_modular/check.sh : improve java's check.sh - http://git.io/nKDWSA | 07:20 |
---|---|---|
CIA-87 | shogun: Soeren Sonnenburg master * r7d3b613 / examples/undocumented/csharp_modular/check.sh : improve mono tests - http://git.io/1AVYCw | 07:22 |
shogun-buildbot | build #75 of r_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/r_modular/builds/75 | 10:09 |
-!- blackburn [~blackburn@188.122.238.50] has joined #shogun | 11:40 | |
CIA-87 | shogun: Sergey Lisitsyn master * r2ca6531 / .gitignore : Updated gitignore - http://git.io/G53bhQ | 12:06 |
CIA-87 | shogun: Sergey Lisitsyn master * r569d85e / (19 files in 9 dirs): Yet another warnings removal - http://git.io/btXRNA | 12:46 |
-!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 14:32 | |
shogun-buildbot | build #76 of r_modular is complete: Failure [failed test_1] Build details are at http://www.shogun-toolbox.org:5347/builders/r_modular/builds/76 blamelist: blackburn91@gmail.com | 17:17 |
shogun-buildbot | build #75 of python_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/python_modular/builds/75 | 17:30 |
shogun-buildbot | build #77 of r_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org:5347/builders/r_modular/builds/77 | 17:30 |
-!- alesis-novik [~alesis@188.74.87.206] has joined #shogun | 18:04 | |
alesis-novik | Hey sonney2k , where's the warning log again (or undocumented stuff log) | 18:10 |
-!- blackburn [~blackburn@188.122.238.50] has joined #shogun | 18:19 | |
@sonney2k | http://shogun-toolbox.org:5347/waterfall | 18:56 |
@sonney2k | alesis-novik, e.g http://shogun-toolbox.org:5347/builders/python_modular/builds/74/steps/compile/logs/warnings%20%2853%29 | 18:56 |
-!- blackburn [~blackburn@188.122.238.50] has quit [Quit: Leaving.] | 18:59 | |
-!- f-x [~user@117.192.222.10] has joined #shogun | 19:21 | |
-!- alesis-novik [~alesis@188.74.87.206] has quit [Quit: Leaving] | 19:22 | |
-!- blackburn [~blackburn@188.122.238.50] has joined #shogun | 19:51 | |
blackburn | sonney2k: what's up with buildbots? | 19:52 |
@sonney2k | blackburn, ? | 19:52 |
blackburn | it seems interfaces builds are out-of-date | 19:52 |
blackburn | I fixed a couple of warnings but they are still here | 19:54 |
CIA-87 | shogun: Soeren Sonnenburg master * r226853f / src/configure : fail when java/ruby/lua are enabled but not found - http://git.io/uoLPJw | 19:58 |
@sonney2k | blackburn, my computer is too unstable :( | 19:58 |
blackburn | why? | 19:58 |
@sonney2k | I cannot figure out why | 19:58 |
@sonney2k | spent lots of time w/ trying | 19:59 |
@sonney2k | (newer linux kernel version, very conservative options, fan running full speed, unplugged usb extensions, no more pci cards, removed / added memory ...) | 20:00 |
@sonney2k | random reboots are the evil of it all | 20:01 |
@sonney2k | If I don't figure out why these happen I will have to buy a new machine | 20:02 |
blackburn | sad | 20:08 |
@sonney2k | in this case it will certainly have enough power to be a buildbot, core i7 / 16G memory - but thats another 400 EUR and more fiddling time :/ | 20:09 |
blackburn | sonney2k: macos failure.. | 21:16 |
@sonney2k | ? | 21:16 |
blackburn | reported on trac | 21:16 |
@sonney2k | blackburn, he should try git | 21:16 |
@sonney2k | that is for shogun 0.10 | 21:16 |
blackburn | ah | 21:16 |
blackburn | sonney2k: should I answer? | 21:17 |
@sonney2k | please | 21:17 |
blackburn | sonney2k: but how you know it is 0.10? | 21:20 |
@sonney2k | blackburn, believe me I know | 21:20 |
blackburn | okay | 21:20 |
CIA-87 | shogun: Sergey Lisitsyn master * r37c5df6 / (4 files): Got rid of custom distance usage in Isomap and MDS - http://git.io/-Fxj6Q | 22:02 |
CIA-87 | shogun: Sergey Lisitsyn master * rba59f81 / (2 files): Updated dimension reduction examples - http://git.io/jH9jbA | 22:02 |
blackburn | sonney2k: is there any define I could use for detecting if pthreads are here? | 22:22 |
@sonney2k | blackburn, I ordered a new machine now | 22:43 |
blackburn | uh | 22:43 |
blackburn | expensive | 22:43 |
@sonney2k | so somewhen next week I hope we will have some better buildbot | 22:44 |
@sonney2k | 400 EUR | 22:44 |
blackburn | I have added HAVE_PTHREAD flag | 22:44 |
blackburn | will commit soon | 22:44 |
@sonney2k | 3.4 GHz Quad-Core i7-2600K | 22:45 |
@sonney2k | will certainly be fater | 22:45 |
@sonney2k | faster | 22:45 |
@sonney2k | blackburn, err but osx *has* pthreads | 22:46 |
@sonney2k | blackburn, are you using sth non-standard? | 22:46 |
blackburn | spinlock | 22:46 |
blackburn | may be it is non standart | 22:46 |
@sonney2k | why spinlock? | 22:46 |
blackburn | I don't know | 22:46 |
blackburn | faster | 22:46 |
@sonney2k | isn't mutex sufficient? | 22:46 |
@sonney2k | blackburn, but they do busy loops or? | 22:46 |
blackburn | I don't know really | 22:47 |
blackburn | but it is faster | 22:47 |
@sonney2k | blackburn, how much faster? | 22:47 |
blackburn | hmm I have tested on 5000 vectors | 22:47 |
blackburn | it was ~4.1s for serial | 22:47 |
blackburn | ~4.8s for mutex | 22:47 |
blackburn | and 3.8 for spinlock | 22:47 |
@sonney2k | so all the same | 22:48 |
@sonney2k | have you tested on 20k vectors or more? | 22:48 |
@sonney2k | I mean one second more or less doesn't matter | 22:48 |
blackburn | nope | 22:48 |
blackburn | do you suggest to use mutex anyway? | 22:48 |
@sonney2k | blackburn, just find out if spinlocks are nonstandard | 22:49 |
@sonney2k | if they are then better use mutexes | 22:49 |
@sonney2k | but you should test if it is faster then | 22:49 |
@sonney2k | I mean if the overhead is say 1-2 seconds for using mutexes (independent of # vectors that you use) it doesn't matter | 22:49 |
@sonney2k | but if this overhead is badly scaling with # vectors than it is no good | 22:50 |
blackburn | sonney2k: okay but HAVE_PTHREAD is a good idea anyway, right? | 22:51 |
@sonney2k | blackburn, one question - couldn't you make the use of spinlocks optional? | 22:51 |
@sonney2k | blackburn, that yes | 22:51 |
blackburn | yeah I could but really I should check | 22:51 |
@sonney2k | blackburn, I mean we could just have a macro - then one could define USE_MUTEX and mutexes will be used / otherwise spinlock | 22:52 |
blackburn | pthread have _POSIX_SPIN_LOCKS definition | 22:52 |
@sonney2k | judging from http://www.alexonlinux.com/pthread-mutex-vs-pthread-spinlock using mutexes/spinlocks is basically the same | 22:52 |
@sonney2k | blackburn, google says osx doesn't know spinlocks | 22:54 |
blackburn | oh I see | 22:54 |
@sonney2k | so how about introducing the macro PTHREAD_LOCK | 22:54 |
@sonney2k | and PTHREAD_UNLOCK | 22:54 |
blackburn | wait wait | 22:54 |
blackburn | I need to test overhead | 22:55 |
@sonney2k | that will then be defined as either pthread_spin_lock | 22:55 |
@sonney2k | or pthread_mutex_lock | 22:55 |
@sonney2k | blackburn, what for? | 22:55 |
blackburn | init too | 22:55 |
blackburn | I'm not sure if I really should use spinlock here | 22:55 |
@sonney2k | yes PTHREAD_LOCK_T pthread_spinlock_t | 22:55 |
@sonney2k | blackburn, yeah but it makes sense to use mutexes per default | 22:56 |
@sonney2k | and then have a configure option to use spinlocks anyways | 22:56 |
blackburn | where should it be defined? | 22:56 |
@sonney2k | blackburn, btw the good news about the new hardware is that with 16G I can actually run 4 buildbots in parallel | 22:56 |
blackburn | nice | 22:59 |
blackburn | sonney2k: I don't think we need to do it globally | 22:59 |
blackburn | for example SGObject's mutex should stay mutex | 22:59 |
@sonney2k | blackburn, I think it shouldn't | 23:00 |
@sonney2k | I mean it is only for incrementing the reference count | 23:01 |
@sonney2k | very unlikely that two cpus even SG_REF at the same time | 23:01 |
blackburn | The best time to use a spinlock is when the protected critical section of code is performing a simple CPU bound task that should be completed as quickly as possible. An example of a simple CPU bound task would be something like incrementing counters or perhaps reading or modifying a memory mapped IO port.A spinlock should NOT be used if the critical section has code in it that is IO bound. An example of this would be waiting for keyboard input from a u | 23:01 |
@sonney2k | so a spinlock would be faster | 23:01 |
@sonney2k | so a spinlock would make more sense | 23:01 |
@sonney2k | ref++ | 23:01 |
blackburn | hmm okay | 23:02 |
@sonney2k | blackburn, do we have mutexes at other places? | 23:02 |
blackburn | I don't know | 23:02 |
blackburn | where should it be introduced? | 23:02 |
@sonney2k | blackburn, how about Parallel.h ? | 23:03 |
blackburn | hmm yeah seems so | 23:03 |
blackburn | okay will try now | 23:03 |
blackburn | sonney2k: could you remind me how to do parameter in macro? | 23:06 |
blackburn | ah () | 23:07 |
blackburn | okay nevermind | 23:07 |
blackburn | LOCK_PTHREAD_LOCK sounds just like die baby die :D | 23:12 |
blackburn | or something crazy like that | 23:12 |
blackburn | uh done | 23:35 |
@sonney2k | I much prefer burn motherfucker burn. | 23:39 |
blackburn | boom | 23:41 |
CIA-87 | shogun: Sergey Lisitsyn master * ra7f397a / examples/undocumented/python_modular/graphical/swissroll.py : Fixed swissroll example (+5 more commits...) - http://git.io/WnqnXw | 23:41 |
* sonney2k is preparing for the night... | 23:41 | |
* sonney2k ... by killing mosquitos | 23:41 | |
blackburn | no mosquitos here already | 23:42 |
blackburn | too cold :) | 23:42 |
@sonney2k | blackburn, I currently see some major installation issues | 23:45 |
@sonney2k | mostly for the new interfaces but also one other issue | 23:46 |
@sonney2k | one should be able to run all examples once examples are installed | 23:46 |
@sonney2k | this means data has to be shipped somehow too (but where?) | 23:46 |
blackburn | toy data could be included to shogun | 23:47 |
@sonney2k | and also that data has to be in some PATH maybe defined in environment variable | 23:47 |
blackburn | agree | 23:47 |
@sonney2k | also examples should not include ../../../.../something | 23:48 |
@sonney2k | but expect these files under certain directories | 23:48 |
@sonney2k | and another problem: we have modshogun.so as the interface lib for python, java, ... | 23:48 |
@sonney2k | we have to see if we will run into problems because of the naming clash | 23:49 |
blackburn | uh I feel upsat because of no burst with parallel algos :( | 23:50 |
blackburn | is there a way to name it modshogun_java? | 23:50 |
blackburn | or any other modshogun_shit | 23:50 |
@sonney2k | blackburn, that would mean though that one has to write import modshogun_python | 23:52 |
@sonney2k | or whatever | 23:52 |
blackburn | oh nono | 23:52 |
@sonney2k | blackburn, parallel algos need lots of data usually to be really usefull | 23:53 |
@sonney2k | blackburn, for the algos I had parallel algos made a differnce for >1e6 examples | 23:53 |
@sonney2k | but before that hmmhh well not too much | 23:53 |
blackburn | nor MDS, or isomap, or .. can handle 1e6 data hah | 23:54 |
@sonney2k | could be that you overoptimize then | 23:55 |
blackburn | not really | 23:55 |
blackburn | the problem is the bottleneck = eigenproblem is not parallel | 23:55 |
blackburn | other steps burst from parallel a little | 23:56 |
blackburn | but they are compartively fast | 23:56 |
--- Log closed Sun Aug 21 00:00:07 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!