--- Log opened Sat Mar 05 00:00:23 2016 | ||
-!- ebadawy [~ebadawy@197.163.32.61] has joined #shogun | 00:10 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 00:33 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 00:33 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has left #shogun [] | 00:59 | |
shogun-buildbot | build #3580 of deb1 - libshogun is complete: Failure [failed git] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/3580 blamelist: EdgeX <tonmoysaikia@live.com>, Tonmoy Saikia <tonmoysaikia@live.com> | 01:06 |
---|---|---|
shogun-buildbot | build #3581 of deb1 - libshogun is complete: Failure [failed git] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/3581 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, EdgeX <tonmoysaikia@live.com>, Tonmoy Saikia <tonmoysaikia@live.com> | 01:29 |
-!- Tanishq [~tanishq@1.186.16.201] has quit [Ping timeout: 240 seconds] | 01:43 | |
-!- bollu [~Adium@1.186.133.195] has joined #shogun | 03:50 | |
-!- bollu [~Adium@1.186.133.195] has quit [Client Quit] | 03:51 | |
shogun-buildbot | build #1083 of nightly_default is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_default/builds/1083 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Bj?rn Esser <me@besser82.io>, EdgeX <tonmoysaikia@live.com>, Tonmoy Saikia <tonmoysaikia@live.com>, Viktor Gal <viktor.gal@maeth.com> | 04:17 |
shogun-buildbot | build #960 of nightly_all is complete: Failure [failed compile] Build details are at http://buildbot.shogun-toolbox.org/builders/nightly_all/builds/960 blamelist: Heiko Strathmann <heiko.strathmann@gmail.com>, Bj?rn Esser <me@besser82.io>, EdgeX <tonmoysaikia@live.com>, Tonmoy Saikia <tonmoysaikia@live.com>, Viktor Gal <viktor.gal@maeth.com> | 04:25 |
-!- lambday [6a33f123@gateway/web/freenode/ip.106.51.241.35] has joined #shogun | 04:50 | |
-!- mode/#shogun [+o lambday] by ChanServ | 04:50 | |
-!- besser82 [~besser82@fedora/besser82] has joined #shogun | 05:15 | |
-!- mode/#shogun [+o besser82] by ChanServ | 05:15 | |
-!- ebadawy [~ebadawy@197.163.32.61] has quit [Remote host closed the connection] | 05:22 | |
-!- aditya123 [uid148851@gateway/web/irccloud.com/x-myfufpprzirjrmaz] has quit [Quit: Connection closed for inactivity] | 05:39 | |
-!- rjtshrma [6725c94b@gateway/web/freenode/ip.103.37.201.75] has quit [Ping timeout: 252 seconds] | 05:56 | |
-!- red_devil [~rahul__@49.207.189.134] has joined #shogun | 06:46 | |
-!- red_devil is now known as Guest53760 | 06:47 | |
-!- Ameya [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has joined #shogun | 07:03 | |
Ameya | Hello! I am Ameya from IIIT Hyderabad, India. I was trying to build shogun but ran into this error. | 07:03 |
Ameya | http://pastebin.com/caTWdhRr | 07:03 |
Ameya | The cmake settings used by me were - cmake -DPythonModular=ON .. | 07:04 |
Ameya | It is a simple correction in line 389 in shogun/src/.doxy2swig.py , changing it from o.write("".join(self.pieces)) -> o.write("".join(self.pieces)).encode('utf-8) | 07:05 |
Ameya | How should I go about reporting it? | 07:06 |
@wiking | shogun-buildbot: force build --branch=develop 'deb1 - libshogun' | 07:20 |
shogun-buildbot | build forced [ETA 6m41s] | 07:20 |
shogun-buildbot | I'll give a shout when the build finishes | 07:20 |
shogun-buildbot | build #3582 of deb1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/3582 | 07:32 |
@wiking | sonney2k: around? | 07:58 |
@besser82 | Ameya, I addressed your change in https://github.com/shogun-toolbox/shogun/pull/3039 | 07:59 |
@besser82 | Ameya, in future you should open an issue on Github --> https://github.com/shogun-toolbox/shogun/issues | 07:59 |
@wiking | besser82: mornign | 08:05 |
@besser82 | wiking, mornin' ^^ | 08:14 |
@besser82 | Ameya, changes for doxy2swig are merged in development branch now | 08:39 |
-!- lambday_ [6a338432@gateway/web/freenode/ip.106.51.132.50] has joined #shogun | 08:50 | |
-!- mode/#shogun [+o lambday_] by ChanServ | 08:50 | |
-!- lambday [6a33f123@gateway/web/freenode/ip.106.51.241.35] has quit [Ping timeout: 252 seconds] | 08:50 | |
shogun-buildbot | build #2774 of deb3 - modular_interfaces is complete: Failure [failed python modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/2774 blamelist: Bj?rn Esser <me@besser82.io> | 08:52 |
shogun-buildbot | build #743 of deb4 - python3 is complete: Failure [failed python modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb4%20-%20python3/builds/743 blamelist: Bj?rn Esser <me@besser82.io> | 08:56 |
Ameya | Thank you Mr. Esser! | 08:56 |
@besser82 | Ameya, you're welcome ^^ | 08:59 |
@wiking | mmm | 09:07 |
@wiking | besser82: that one failed :) | 09:07 |
@besser82 | wiking, cannot be my fault... :P | 09:07 |
@wiking | i dunno who put some none-ascii stuff | 09:07 |
@wiking | into the fucking thing... | 09:08 |
@wiking | and who let it to be merged | 09:08 |
Ameya | oh yes | 09:12 |
Ameya | o.write("".join(self.pieces).encode('utf-8)) | 09:13 |
Ameya | My bad | 09:13 |
@besser82 | Ameya, oops... will fix quickly | 09:19 |
shogun-buildbot | build #457 of trusty - libshogun - viennacl is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/457 blamelist: Bj?rn Esser <me@besser82.io> | 09:19 |
@besser82 | Ameya, wiking: fix is pushed :P | 09:23 |
Ameya | Mr. Essar, did you make the pr? | 09:28 |
Ameya | I just made one fixing the issue. It was declined | 09:28 |
Ameya | https://github.com/shogun-toolbox/shogun/pull/3041 | 09:31 |
Ameya | Is this okay? | 09:31 |
Ameya | Oh, wait, I see you fixed it in https://goo.gl/zdmoC5. Was hoping I could get a PR in myself :) | 09:33 |
@besser82 | Ameya, next time ^^ | 09:35 |
shogun-buildbot | build #3585 of deb1 - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/3585 blamelist: Bj?rn Esser <me@besser82.io> | 09:45 |
shogun-buildbot | build #336 of trusty - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun/builds/336 blamelist: Bj?rn Esser <me@besser82.io> | 09:51 |
shogun-buildbot | build #458 of trusty - libshogun - viennacl is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun%20-%20viennacl/builds/458 | 09:52 |
-!- Guest53760 is now known as rahul | 09:58 | |
-!- rahul is now known as Guest4503 | 09:58 | |
-!- Guest4503 [~rahul__@49.207.189.134] has quit [Quit: Leaving] | 10:01 | |
-!- cache-nez [~cache-nez@93.175.19.64] has joined #shogun | 10:37 | |
-!- lambday_ [6a338432@gateway/web/freenode/ip.106.51.132.50] has quit [Ping timeout: 252 seconds] | 11:00 | |
-!- jack [a4a41f06@gateway/web/freenode/ip.164.164.31.6] has joined #shogun | 11:15 | |
jack | hi | 11:15 |
-!- jack is now known as Guest22344 | 11:15 | |
Guest22344 | I have some problem with loading large data set(100k) to ridgeregression.sg | 11:16 |
Guest22344 | so I just checked with loading just 5K data set and it works, but if I go beyond that it gets loading loading and loading | 11:18 |
Guest22344 | and there is no result till 2 days | 11:18 |
Guest22344 | is there some limitation with loading data sets? | 11:19 |
Ameya | Hello again! I wanted to run the tests and fix the tests to all run <1second . But, as specified by Heiko in #3005, "run make test with Python modular enabled to get a list of timings", I tried running that command but the output is: make: *** No rule to make target `test'. Stop. | 11:31 |
@besser82 | Ameya, run `ctest .` in the build-dir ;) | 11:36 |
Ameya | Oh! I tried that. Got this output - 'Test project /home/ameya/shogun/build No tests were found!!!' | 11:42 |
@besser82 | Ameya, Then you need to rebuild / reconfigure using cmake -DENABLE_TESTING=ON | 11:42 |
-!- cache-nez [~cache-nez@93.175.19.64] has quit [Ping timeout: 244 seconds] | 11:49 | |
-!- Tanishq [~tanishq@1.186.12.154] has joined #shogun | 12:24 | |
-!- lambday [6a338432@gateway/web/freenode/ip.106.51.132.50] has joined #shogun | 12:31 | |
-!- mode/#shogun [+o lambday] by ChanServ | 12:31 | |
-!- cache-nez [~cache-nez@93.175.2.172] has joined #shogun | 12:54 | |
-!- lzc [~lzc@115.27.197.134] has joined #shogun | 12:56 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 13:00 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 13:00 | |
-!- besser82 [~besser82@fedora/besser82] has quit [Ping timeout: 240 seconds] | 13:02 | |
-!- Guest22344 [a4a41f06@gateway/web/freenode/ip.164.164.31.6] has quit [Quit: Page closed] | 13:04 | |
-!- cache-nez [~cache-nez@93.175.2.172] has quit [Ping timeout: 240 seconds] | 13:27 | |
@wiking | HeikoS: ping | 13:51 |
@HeikoS | wiking: pong | 13:51 |
@wiking | ok so last questions | 13:52 |
@HeikoS | shoot | 13:52 |
@wiking | i've fixed most of the stuff in that branch | 13:52 |
@wiking | now about the place to store | 13:52 |
-!- wanderer_ [67157d4e@gateway/web/freenode/ip.103.21.125.78] has joined #shogun | 13:52 | |
wanderer_ | hey | 13:52 |
@wiking | a) do we really want to have the cookbook updated per commit? wouldn't be it enough as part of a nightly | 13:52 |
@wiking | ? | 13:52 |
@wiking | b) shouldn't we store the cookbooks the same place as the website? | 13:53 |
@HeikoS | a) yes nightly is fine for develop | 13:53 |
@HeikoS | b) yes I agree | 13:53 |
@wiking | mmmm | 13:53 |
@wiking | ok | 13:53 |
@HeikoS | c) for PR, it would be good per commit, dont you think? | 13:53 |
@wiking | yeah PR is fine | 13:53 |
@HeikoS | ehm per PR | 13:53 |
@HeikoS | okok | 13:53 |
@wiking | i can do that | 13:53 |
@wiking | i was just wondeirng about the 'latest' | 13:53 |
@wiking | that should be enough per nightly | 13:53 |
wanderer_ | wiking: i tried compiling the latest version with g++-4.9, but it always gives an error here | 13:53 |
@wiking | just as part of doxygen update | 13:53 |
@HeikoS | wiking: yeah sure, just like the notebooks/doxygen | 13:53 |
@HeikoS | totally fine | 13:54 |
@wiking | wanderer_: use pastebin.com | 13:54 |
@wiking | HeikoS: ok lemme do those changes in .travis.yaml | 13:54 |
@HeikoS | wiking: sweet | 13:54 |
wanderer_ | here - http://pastebin.com/uDwnSJS8 | 13:55 |
@HeikoS | wiking: btw I have no idea why the c# examples wont be compiled anymore, it used to work with the *same* CMakeLists.txt in the examples/meta/csharp dir | 13:55 |
@wiking | good question | 13:55 |
@wiking | dunno yet | 13:55 |
@wiking | but i wanna work on io stuff now | 13:56 |
@wiking | so i fix this PR thing | 13:56 |
@wiking | but then it's your stuff | 13:56 |
@HeikoS | wiking: ok | 13:56 |
wanderer_ | and for your reference, my makefile - http://pastebin.com/knytSw0i | 13:56 |
@HeikoS | wiking: great, thanks! | 13:56 |
@wiking | makefile? | 13:56 |
wanderer_ | one generated by ccmake | 13:56 |
@wiking | yeah well nobody cares about that | 13:56 |
wanderer_ | okay | 13:56 |
@wiking | dunno never seen this error before | 13:57 |
@wiking | and certainly we are compiling shogun with 4.9 | 13:57 |
@HeikoS | wanderer_: I think this might be compiler version | 13:57 |
wanderer_ | my g++ --version shows 4.9 | 13:57 |
@HeikoS | wiking: good stuff, looking forward to this | 13:57 |
@HeikoS | wiking: btw do you have an opinion how to test all the meta example outputs? | 13:58 |
@HeikoS | wiking: I did not really like the tester.py script | 13:58 |
@HeikoS | could do similar, or something different | 13:58 |
@wiking | HeikoS: ? | 14:00 |
@wiking | what do you mean how to test the output? | 14:00 |
@HeikoS | wiking: the test build runs all meta examples | 14:01 |
@HeikoS | meta examples store all numerical variables into a single file | 14:01 |
@HeikoS | from each language | 14:01 |
lisitsyn | ok I woke up! | 14:01 |
@wiking | lisitsyn: :DDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDDD | 14:01 |
@wiking | lol | 14:01 |
@wiking | HeikoS: but why would you do that? :) | 14:01 |
@HeikoS | now want to compare that the example cpp/java/py all produce the same file, that is the same as a reference file | 14:01 |
@wiking | ah i see | 14:01 |
@HeikoS | wiking: a) implicitly test typemaps b) integration testing | 14:02 |
@wiking | and these are examples | 14:02 |
@wiking | :) | 14:02 |
@wiking | mmm | 14:02 |
@wiking | good question | 14:02 |
lisitsyn | whats the status of this thing | 14:02 |
@wiking | lemme see first where the hell should we store those PR cookbooks | 14:02 |
@HeikoS | lisitsyn: how to compare the example output | 14:02 |
@wiking | btw | 14:02 |
@HeikoS | wiking: ok | 14:02 |
@wiking | somebody fix all the fucking broken stuff | 14:02 |
@wiking | that has been introduced in the last 24 hours | 14:02 |
@wiking | shogun-buildbot: force build --branch=develop 'deb1 - libshogun' | 14:03 |
shogun-buildbot | build forced [ETA 3m33s] | 14:03 |
shogun-buildbot | I'll give a shout when the build finishes | 14:03 |
@wiking | plz be more vigilant with PRs | 14:03 |
@HeikoS | wiking: yeah ok, will do, it is hard as there are so many | 14:03 |
@wiking | level of codes lately is not the high standard :D | 14:03 |
@HeikoS | wiking: you have to shoot people who merge things early ;) | 14:04 |
@HeikoS | but I agree | 14:04 |
shogun-buildbot | build #3586 of deb1 - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb1%20-%20libshogun/builds/3586 | 14:04 |
@wiking | shogun-buildbot: force build --branch=develop 'trusty - libshogun' | 14:05 |
shogun-buildbot | build forced [ETA 14m16s] | 14:05 |
shogun-buildbot | I'll give a shout when the build finishes | 14:05 |
@wiking | shogun-buildbot: force build --branch=develop 'deb3 - modular_interfaces' | 14:05 |
shogun-buildbot | build forced [ETA 33m08s] | 14:05 |
shogun-buildbot | I'll give a shout when the build finishes | 14:05 |
@wiking | [14:06][buildbot@7nn:~/masters/shogun/public_html]$ du -sh | 14:06 |
@wiking | 9.9G. | 14:06 |
@wiking | lol | 14:06 |
@wiking | :) | 14:06 |
@wiking | i better clean up some memcheck static analysis files | 14:06 |
@wiking | or better yet, create a crontab there | 14:06 |
wanderer_ | umm .... any thoughts? | 14:07 |
@wiking | HeikoS: ok so as far as i remember my BSD licensing agreement got flushed from static folder | 14:08 |
@wiking | once the website is updated | 14:08 |
@HeikoS | wiking: pitty | 14:08 |
@HeikoS | wiking: but we decided to do it anyways | 14:08 |
@HeikoS | gunnar overrode all his employees ;) | 14:08 |
@wiking | yeah yeah | 14:09 |
@HeikoS | only the alglib stuff is problematic as it is gpl | 14:09 |
@HeikoS | the rest shoul dbe ok | 14:09 |
@wiking | but the question is | 14:09 |
@wiking | where should i place | 14:09 |
@wiking | the PRed cookbooks | 14:09 |
@HeikoS | we announce and give ppl a chance to tell us if they dont agree | 14:09 |
@HeikoS | aj | 14:09 |
@wiking | on the website | 14:09 |
@HeikoS | dont know | 14:09 |
@HeikoS | maybe? | 14:09 |
@wiking | because if i put in the static dir | 14:09 |
@wiking | it'll be flushed | 14:10 |
@wiking | if somebody updates the website | 14:10 |
@wiking | or maybe not | 14:10 |
@wiking | drwxr-xr-x 5 shogun shogun 4096 Jul 10 2012 admin | 14:11 |
@wiking | im certain since then stuff got updated | 14:11 |
@HeikoS | yes definitely | 14:11 |
@wiking | ok i'll put it there then | 14:12 |
@wiking | cookbook_pr | 14:12 |
@HeikoS | wiking: it should be flushed every now and then anyways right? | 14:12 |
@HeikoS | when PR is closed | 14:12 |
@HeikoS | or something | 14:12 |
@wiking | HeikoS: cron | 14:12 |
@wiking | it'll flush it every week once | 14:12 |
@HeikoS | once a week? | 14:12 |
@HeikoS | good | 14:12 |
@wiking | yeah something like that | 14:12 |
@wiking | lisitsyn: can we remove that ssh key from shogun | 14:12 |
@wiking | that you've created? | 14:12 |
lisitsyn | yes I already removed it I think | 14:13 |
lisitsyn | I realized it is pretty much useless in this case | 14:14 |
@wiking | :) | 14:14 |
@wiking | k | 14:14 |
@wiking | mmm | 14:14 |
@wiking | OSX buildbot | 14:14 |
@HeikoS | osx builbot would be a blast | 14:15 |
@HeikoS | and a nightly binary ;) | 14:15 |
@wiking | OS X El Capitan Lite | 14:15 |
@wiking | $24.99/mo | 14:15 |
@wiking | :< | 14:15 |
@HeikoS | and cygwin bot would be nice too | 14:15 |
@HeikoS | arhh | 14:15 |
@wiking | f*ing 25 usd | 14:15 |
@HeikoS | what about gunnars machine? | 14:15 |
@HeikoS | did that ever work? | 14:15 |
@wiking | i get access to a matlab machine | 14:15 |
@wiking | mmm something weird is with LaRank.* unit tests | 14:16 |
@wiking | takes a long time | 14:16 |
@wiking | it's a quite recent change | 14:16 |
@wiking | because that's why recently unit tests starts to timeout | 14:17 |
@lambday | whoa | 14:21 |
@HeikoS | wiking: what are you doing with this file reader input stuff? | 14:21 |
@HeikoS | lambday: hey there! | 14:21 |
@lambday | HeikoS: hola | 14:21 |
* lambday scratches his head | 14:21 | |
@HeikoS | lambday: whats up? :) | 14:21 |
@lambday | HeikoS: I needed to discuss some design related things regarding the new flexy-framework.. | 14:22 |
@lambday | have been working on it | 14:22 |
@lambday | HeikoS: shall I write a mail in detail? | 14:22 |
@HeikoS | lambday: maybe yes, or on github? | 14:22 |
@wiking | HeikoS: which file reader? | 14:22 |
@lambday | I changed quite a few things in the POC - tried to make it sexier https://github.com/lambday/flash | 14:22 |
@wiking | buildsl+ 24091 99.9 0.0 225460 29332 ? R 14:10 11:19 /home/buildslave/trusty_-_libshogun/build/build/tests/unit/shogun-unit-test --gtest_filter=LaRank.* | 14:22 |
@HeikoS | wiking: you mentioned something you were doing, just curious | 14:22 |
@wiking | HeikoS: io | 14:23 |
@wiking | HeikoS: refactoring io code | 14:23 |
@wiking | because it's shitty | 14:23 |
@HeikoS | lambday: ok cool ,send me a brief summary of the changes maybe per mail? | 14:23 |
@wiking | adding libarchive + hdfs support | 14:23 |
@HeikoS | wiking: ah yes please!!! | 14:23 |
@wiking | and fixing the circular buffer | 14:23 |
@HeikoS | lambday: im preparing a talk for monday, so keep it brief ;) Monday I have more time :) | 14:23 |
@HeikoS | wiking: and then also csv if possible | 14:23 |
@wiking | that's why we have a problem reading cvs 10M+ | 14:23 |
@lambday | HeikoS: cool.. I am worried about sending a PR.. it is too huge.. | 14:23 |
@lambday | HeikoS: okay monday would be better | 14:23 |
@wiking | HeikoS: csv is due to circular buffer bug | 14:24 |
@lambday | let me write the mail before I forget the details | 14:24 |
@HeikoS | lambday: maybe send me a mail with the main changes | 14:24 |
@HeikoS | lambday: and then just push to feature branch | 14:24 |
@wiking | so we fix that csv is fixed | 14:24 |
@lambday | HeikoS: alright | 14:24 |
@HeikoS | and then put tests there, and if all is good we merge this thing | 14:24 |
@HeikoS | lambday: you are right a PR is too big with this | 14:24 |
@lambday | massive | 14:24 |
@HeikoS | wiking: great | 14:24 |
@HeikoS | lambday: haha ,) | 14:24 |
@lambday | HeikoS: I also thought, that until all the components get added, I'd keep it under shogun/statistics/experimental... old stuffs would also be there | 14:25 |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has joined #shogun | 14:25 | |
@HeikoS | lambday: just feature branch it | 14:25 |
@HeikoS | nobody is working on this | 14:25 |
@lambday | okay | 14:25 |
@HeikoS | lambday: curious what you did | 14:27 |
@HeikoS | looking forward, let me know when you sent the email | 14:27 |
@HeikoS | procastinating here ;) | 14:27 |
@lambday | lisitsyn: https://github.com/lambday/flash/blob/develop/src/flash/statistics/HypothesisTest.cpp#L58 tribute to a few tricks I learned from you over the years | 14:27 |
@lambday | HeikoS: alright | 14:27 |
lisitsyn | lambday: hah | 14:28 |
Saurabh7 | lambday, wahts the difference between https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/mathematics/Math.cpp#L90 and https://github.com/shogun-toolbox/shogun/blob/develop/src/shogun/mathematics/linalg/internal/implementation/Dot.h#L138 | 14:28 |
@lambday | when I shogunify it, gotta "some"/"unique" the hell out of it | 14:28 |
lisitsyn | lambday: why don't you call 'impl' as 'self'? ;) | 14:28 |
@lambday | lisitsyn: I will | 14:29 |
Saurabh7 | lambday, same thing i asked few days ago :) | 14:29 |
lisitsyn | lambday: just curious what you think | 14:29 |
Saurabh7 | lambday, but i am definetly getting perfromancedifference between those too | 14:29 |
Saurabh7 | *two | 14:29 |
@lambday | Saurabh7: can you please benchmark it? it's weird | 14:30 |
@lambday | lisitsyn: I like self.. not sure about "some".. | 14:30 |
shogun-buildbot | build #337 of trusty - libshogun is complete: Failure [failed test] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun/builds/337 | 14:30 |
@lambday | I get what you mean.. but "shared" is clearer | 14:30 |
lisitsyn | lambday: good | 14:30 |
lisitsyn | lambday: what I was thinking about 'some' | 14:30 |
lisitsyn | is that 'some' is a good name for something pooled | 14:30 |
lisitsyn | like give me some available classifier | 14:31 |
@lambday | lisitsyn: coming from some other source - I don't own it I won't delete it | 14:31 |
@lambday | that's what you mean | 14:31 |
@wiking | shogun-buildbot: force build --branch=develop 'trusty - libshogun' | 14:31 |
shogun-buildbot | build forced [ETA 14m16s] | 14:31 |
shogun-buildbot | I'll give a shout when the build finishes | 14:31 |
lisitsyn | lambday: I am also good with shared | 14:31 |
lisitsyn | don't mind renaming it in shogun as well | 14:31 |
lisitsyn | Shared as a class, and 'shared' function to produce them | 14:32 |
@lambday | yeah... like make_shared | 14:32 |
@lambday | on the other hand, make_some is weird :D | 14:32 |
@lambday | get_some :D | 14:32 |
@lambday | haha | 14:32 |
lisitsyn | I don't think make_shared is short enough | 14:33 |
@wiking | shogun-buildbot: force build --branch=develop 'deb4 - python3' | 14:33 |
shogun-buildbot | The build has been queued, I'll give a shout when it starts | 14:33 |
lisitsyn | I mean if I call shared<SomeClass>(..) | 14:33 |
lisitsyn | it is already clear that I request shared SomeClass | 14:33 |
@lambday | yeah.. | 14:34 |
@lambday | true that | 14:34 |
lisitsyn | the only danger if that become a keyword | 14:34 |
lisitsyn | :) | 14:34 |
@lambday | haha don't worry that won't happen anytime soon | 14:34 |
@lambday | this dot issue is weird! these two pieces of code that Saurabh7 shared is *almost* identical.. | 14:35 |
@lambday | except for the fancy map operator that's used in linalg dot | 14:35 |
@lambday | and it's for SG | 14:35 |
Saurabh7 | its dynamic size though | 14:35 |
@lambday | Saurabh7: I don't get it.. it basically does the same thing! | 14:36 |
Saurabh7 | in linalg | 14:36 |
lisitsyn | lambday: are they are different performance wise? | 14:36 |
@lambday | lisitsyn: as Saurabh7 says, yes! | 14:36 |
@lambday | Saurabh7: the other one is also for dynamic size, right? | 14:36 |
lisitsyn | Saurabh7: what's faster? | 14:36 |
@lambday | as far as eigen is concerned | 14:37 |
Saurabh7 | onst Eigen::VectorXd> ev1(v1,n) | 14:37 |
Saurabh7 | const Eigen::VectorXd> ev1(v1,n) | 14:37 |
Saurabh7 | this ones faster | 14:37 |
@wiking | yeah because it's consst | 14:37 |
@wiking | HeikoS: ok i gotta go out | 14:37 |
@wiking | i see sun for the first time since i'm europe | 14:37 |
lisitsyn | hmm could be because of that ^ | 14:37 |
@HeikoS | wiking: no sun here :( | 14:38 |
@wiking | maybe w ecannot use travis | 14:38 |
@wiking | i'll have to see | 14:38 |
@wiking | but in worst case | 14:38 |
@HeikoS | wiking: no travis? | 14:38 |
@wiking | i can put a hook for a buildbot | 14:38 |
@HeikoS | ah ok | 14:38 |
@wiking | that'll do the same thing for pr | 14:38 |
@lambday | ah | 14:38 |
@wiking | yeah i mean if its const | 14:39 |
@wiking | you dont need to do any memcpy or anything | 14:39 |
@wiking | right? | 14:39 |
Saurabh7 | lambday, maybe we can pass vector size and map to constant in linalg | 14:39 |
Saurabh7 | rather than dynamic everytime | 14:39 |
@lambday | wiking: map doesn't do any memcpy anyway so far I know | 14:40 |
lisitsyn | it doesn't | 14:40 |
lisitsyn | it is either const somehow | 14:40 |
lisitsyn | or alignment | 14:40 |
@wiking | lambday: yeah but if it's not const | 14:41 |
@wiking | static T compute(shogun::SGVector<T> a, shogun::SGVector<T> b) | 14:41 |
@wiking | that thing is not a pass by reference | 14:41 |
@wiking | you pass there the whole class | 14:41 |
@wiking | right? | 14:41 |
@lambday | that just increases the refcount | 14:42 |
@wiking | right? | 14:42 |
@wiking | on the stack? | 14:42 |
@lambday | doesn't do memcpy | 14:42 |
@wiking | ok write your code | 14:42 |
@wiking | gcc -o# | 14:42 |
@wiking | gcc -O3 | 14:42 |
@wiking | and see what it'll do on the stack for you | 14:42 |
@wiking | objdump -D | 14:42 |
@lambday | yes the object is on the stack but the data ptr is dynamically allocated for SGVector | 14:42 |
shogun-buildbot | build #338 of trusty - libshogun is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/trusty%20-%20libshogun/builds/338 | 14:42 |
@lambday | which just copies the ptr and increases the refcount | 14:43 |
@lambday | no memcpy | 14:43 |
@wiking | again | 14:43 |
@lambday | lisitsyn: ^^ can you confirm this? | 14:43 |
@wiking | compile | 14:43 |
@wiking | and objdump | 14:43 |
@wiking | you'll see | 14:43 |
lisitsyn | sgvector could be copied but not the data | 14:43 |
@wiking | isn't that enough already? | 14:43 |
lisitsyn | no it is very small | 14:43 |
lisitsyn | a few bytes | 14:44 |
@lambday | how a ptr copy and refcount++ be so much overhead? | 14:44 |
lisitsyn | comparing to dot operation that is a few thousand of multiplications | 14:44 |
@wiking | Saurabh7: what's the runtime diff? | 14:44 |
@lambday | we never pass SGVector/SGMatrix via ref in shogun.. | 14:45 |
@lambday | afaik | 14:45 |
Saurabh7 | i am using it in kmeans for k= 100 and it goes from ~30 sec to ~100 s | 14:46 |
@lambday | Saurabh7: -O3? | 14:46 |
@lambday | okay let me try a quick benchmark | 14:47 |
wanderer_ | can anybody please help me with a build error? | 14:47 |
@wiking | mmm lol that's quite a diff | 14:47 |
@wiking | ttyl | 14:48 |
lisitsyn | wanderer_: what's the error? | 14:48 |
wanderer_ | http://pastebin.com/uDwnSJS8 | 14:49 |
wanderer_ | i am using g++-4.9 with -std=c++11 in the cmake | 14:49 |
lisitsyn | wanderer_: what os? | 14:49 |
wanderer_ | linux mint 17 | 14:49 |
wanderer_ | its based on ubuntu 14.04 i think | 14:50 |
shogun-buildbot | build #2775 of deb3 - modular_interfaces is complete: Success [build successful] Build details are at http://buildbot.shogun-toolbox.org/builders/deb3%20-%20modular_interfaces/builds/2775 | 14:51 |
shogun-buildbot | build forced [ETA 9m28s] | 14:51 |
shogun-buildbot | I'll give a shout when the build finishes | 14:51 |
lisitsyn | HAVE_CXX11 is not set for you somehow | 14:51 |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has joined #shogun | 14:52 | |
lisitsyn | it would call proper setter of real part if HAVE_CXX0X or HAVE_CXX11 was set | 14:52 |
lisitsyn | they are set according to a set of tests | 14:52 |
lisitsyn | wanderer_: how do you pass that '-std=c++11' ? | 14:53 |
wanderer_ | in CMAKE_CXX_FLAGS while doing ccmake . | 14:54 |
shogun-buildbot | build #744 of deb4 - python3 is complete: Failure [failed python modular] Build details are at http://buildbot.shogun-toolbox.org/builders/deb4%20-%20python3/builds/744 | 14:55 |
wanderer_ | lisitsyn: any ideas on how to set the HAVE_CXX11 then? | 14:56 |
lisitsyn | wanderer_: why do you want to pass it to flags? | 15:01 |
wanderer_ | lisitsyn: then how do i build it? | 15:02 |
lisitsyn | wanderer_: I mean why to pass std to flags directly? | 15:03 |
lisitsyn | don't it compile without it? | 15:03 |
wanderer_ | lemme check. back in a minute | 15:04 |
wanderer_ | lisitsyn: it doesn't actually | 15:08 |
@lambday | lisitsyn: Saurabh7 https://gist.github.com/lambday/7cbd77eb0530ad41d9f5 | 15:08 |
@lambday | this is negligible difference | 15:08 |
wanderer_ | lisitsyn: it actually produces a c++-11 error | 15:10 |
wanderer_ | lisitsyn: http://pastebin.com/THyrApVA | 15:10 |
wanderer_ | that's why i thought enabling it via flags was a good idea | 15:10 |
@lambday | Saurabh7: https://gist.github.com/lambday/938eaf0b0a7962769c46 | 15:16 |
@lambday | lisitsyn: ^^ | 15:16 |
lisitsyn | wanderer_: ok I see | 15:18 |
lisitsyn | wanderer_: our build system believes your gcc doesn't support C++11 (for some reason) | 15:18 |
lisitsyn | and meanwhile non-C++11 build is broken | 15:19 |
lisitsyn | hah | 15:19 |
-!- cache-nez [~cache-nez@93.175.2.172] has joined #shogun | 15:19 | |
@lambday | Saurabh7: can you run that last benchmark in your machine and let us know the difference? | 15:20 |
@lambday | Saurabh7: I used these flags : g++ -O3 -std=c++14 dot_benchmark.cpp -I/usr/include/eigen3 -lshogun -lhayai_main | 15:20 |
wanderer_ | lisitsyn: i saw the cmakelists.txt file | 15:20 |
wanderer_ | and this is what i found | 15:20 |
wanderer_ | IF (NOT ((CYGWIN AND ENABLE_TESTING) OR (DARWIN AND COMPILE_MODULAR_INTERFACE AND SWIG_VERSION VERSION_LESS "2.0.12") OR MatlabStatic)) | 15:21 |
wanderer_ | INCLUDE(CheckCXX11Features) | 15:21 |
-!- Tanishq [~tanishq@1.186.12.154] has left #shogun [] | 15:25 | |
Saurabh7 | lambday, same not much difference | 15:34 |
Saurabh7 | lambday, but | 15:34 |
Saurabh7 | can you try this https://gist.github.com/Saurabh7/9e83fca8763b45fd48cf | 15:34 |
Saurabh7 | more iterations as in kmeans but dimensions not that big | 15:34 |
@lambday | Saurabh7: are you using the compiler optimization flag? | 15:35 |
Saurabh7 | g++ -O3 -std=c++11 benchmark.cpp -I/usr/include/eigen3 -lshogun -lhayai_main | 15:35 |
@lambday | linalg is all templates, so without -O3 it is doomed | 15:35 |
@lambday | Saurabh7: let me try | 15:35 |
@lambday | what's the result in your machine? | 15:35 |
Saurabh7 | sec posting | 15:36 |
Saurabh7 | lambday, https://gist.github.com/Saurabh7/41dc7227dd3380dace2b | 15:36 |
Saurabh7 | looks like for smaller dimensions dynamic is slower | 15:37 |
@lambday | Saurabh7: yeah I see.. for smaller vectors it is taking way too much time | 15:37 |
@lambday | let me try with the naked linalg::T* thing | 15:37 |
Saurabh7 | and many runs as required for kmeans distance computations | 15:37 |
-!- curiousguy13 [~curiousgu@59.177.198.117] has joined #shogun | 15:43 | |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has quit [Remote host closed the connection] | 16:06 | |
@lambday | Saurabh7: for small dims, I noticed some really weird stuffs... check the 2nd set of benchmarks.. https://gist.github.com/lambday/7cbd77eb0530ad41d9f5 | 16:07 |
@lambday | dot_benchmark2.cpp and output2.log | 16:07 |
@lambday | will check it in details | 16:07 |
@lambday | for small dims, eigen map is an overkill | 16:08 |
@lambday | in that case, the best solution seems to be to have a wrapper for naked ptr in linalg and use NATIVE | 16:10 |
Saurabh7 | lamday, hmm looks weird native is just loops right | 16:12 |
-!- travis-ci [~travis-ci@ec2-54-226-94-57.compute-1.amazonaws.com] has joined #shogun | 16:12 | |
travis-ci | it's Heiko Strathmann's turn to pay the next round of drinks for the massacre he caused in shogun-toolbox/shogun: https://travis-ci.org/shogun-toolbox/shogun/builds/113893881 | 16:12 |
-!- travis-ci [~travis-ci@ec2-54-226-94-57.compute-1.amazonaws.com] has left #shogun [] | 16:12 | |
Saurabh7 | but we need a generic solution since we dont knwo dimensions | 16:13 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 16:13 | |
@lambday | Saurabh7: not even the ballpark? for your knn use-case? | 16:14 |
@lambday | can it be really huge, that's what I am asking | 16:14 |
* lambday afk | 16:14 | |
Saurabh7 | yeah i mean the data can be any dimensions | 16:15 |
Saurabh7 | not very huge that would be impractical | 16:15 |
Saurabh7 | lambday, but good performance for say dimensions 10 to 1000 would be nice | 16:16 |
Saurabh7 | lambday, let me know if i can help update something , i ahve to use it in this pr https://github.com/shogun-toolbox/shogun/pull/3042 | 16:18 |
Saurabh7 | gtg | 16:18 |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has quit [Quit: Leaving] | 16:18 | |
-!- wanderer_ [67157d4e@gateway/web/freenode/ip.103.21.125.78] has quit [Quit: Page closed] | 16:19 | |
-!- cache-nez [~cache-nez@93.175.2.172] has quit [Ping timeout: 264 seconds] | 16:21 | |
@lambday | lisitsyn: https://gist.github.com/lambday/7cbd77eb0530ad41d9f5 | 16:27 |
@lambday | any clue? | 16:27 |
@lambday | :/ | 16:27 |
@lambday | check dot_benchmark2.cpp and output2.log | 16:27 |
-!- lzc [~lzc@115.27.197.134] has quit [Quit: Leaving.] | 16:31 | |
-!- Nebula_ [~chatzilla@1.39.60.142] has joined #shogun | 16:40 | |
-!- Nebula_ [~chatzilla@1.39.60.142] has quit [Client Quit] | 16:41 | |
-!- curiousguy13_ [~curiousgu@120.56.229.207] has joined #shogun | 16:46 | |
-!- curiousguy13 [~curiousgu@59.177.198.117] has quit [Ping timeout: 260 seconds] | 16:48 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 16:57 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 16:57 | |
lisitsyn | lambday: which results should I check? | 16:58 |
@lambday | lisitsyn: check dot_benchmark2.cpp and output2.log | 16:58 |
lisitsyn | lambday: linalg_native is slow | 16:59 |
lisitsyn | is it? | 16:59 |
@lambday | lisitsyn: na the raw ptr is extraordinarily fast | 17:00 |
@lambday | also, native calls the same but it is slow | 17:00 |
lisitsyn | lambday: incredibly hard to read this thing | 17:00 |
lisitsyn | :D | 17:01 |
@lambday | hehe yeah | 17:01 |
@lambday | wait I am removing other two irrelevant files | 17:01 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Client Quit] | 17:01 | |
@lambday | lisitsyn: check now | 17:02 |
lisitsyn | 0.002 mus per iteration? | 17:04 |
lisitsyn | lambday: wasn't it optimized out somehow? | 17:05 |
@lambday | I suspect the same | 17:05 |
@lambday | how do I stop it from happening without compromising with -O3 | 17:06 |
@lambday | add the result? | 17:06 |
lisitsyn | yeah put it into volatile variable | 17:06 |
lisitsyn | lambday: I guess when you specify it even more it gives more ability to optimize it out | 17:08 |
@lambday | lisitsyn: check this https://gist.github.com/lambday/7cbd77eb0530ad41d9f5 | 17:09 |
@lambday | this makes much more sense.. except the fact that, the cmath one is still faster while it uses the same as eigen_map test | 17:10 |
@lambday | but I don't understand why linalg is so slow | 17:11 |
-!- cache-nez [~cache-nez@93.175.2.172] has joined #shogun | 17:24 | |
-!- wanderer_ [67157d4e@gateway/web/freenode/ip.103.21.125.78] has joined #shogun | 17:31 | |
wanderer_ | hey | 17:31 |
wanderer_ | so i am trying to include mosek support in shogun | 17:32 |
wanderer_ | in the cmake file there's only one MOSEK_LIBRARY | 17:32 |
wanderer_ | but according to the developers there are three *.so libraries | 17:32 |
wanderer_ | am i doing something wrong? | 17:32 |
wanderer_ | lisitsyn: are you free? | 17:40 |
@wiking | lambday: so what was the cause? | 17:46 |
@lambday | wiking: bummer - no clue | 17:47 |
@lambday | wiking: from the benchmarks, the ref thing didn't make any (significant) difference | 17:47 |
@wiking | but there's a diff? | 17:48 |
@wiking | i mean between cmath:: and linalg:: ? | 17:48 |
@lambday | cmath is faster (?) for small dimensions | 17:49 |
@wiking | mmm ok | 17:49 |
@lambday | for large enough dimensions, both are same | 17:49 |
@lambday | that's understandable, since cmath one works with naked ptrs | 17:50 |
@lambday | but I fail to understand why so much difference is there for smaller dimension, especially when both use eigen3 backend | 17:50 |
@lambday | in earlier benchmarks I have always taken 1bil sized vectors... so never noticed the difference for smaller sized ones | 17:52 |
-!- cache-nez [~cache-nez@93.175.2.172] has quit [Ping timeout: 244 seconds] | 17:54 | |
-!- lzc [~lzc@115.27.197.134] has joined #shogun | 17:55 | |
-!- cache-nez [~cache-nez@93.175.2.172] has joined #shogun | 18:00 | |
-!- rahul_red [~rahul__@183.83.51.4] has joined #shogun | 18:00 | |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has joined #shogun | 18:15 | |
-!- Ameya [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has quit [Ping timeout: 252 seconds] | 18:27 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 18:29 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 18:29 | |
-!- cache-nez [~cache-nez@93.175.2.172] has quit [Ping timeout: 252 seconds] | 18:35 | |
-!- cache-nez [~cache-nez@93.175.2.172] has joined #shogun | 18:40 | |
-!- Ameya [0e8b5206@gateway/web/freenode/ip.14.139.82.6] has joined #shogun | 18:44 | |
-!- cache-nez [~cache-nez@93.175.2.172] has quit [Ping timeout: 244 seconds] | 18:51 | |
-!- cache-nez [~cache-nez@93.175.2.172] has joined #shogun | 18:57 | |
@lambday | HeikoS: sent you the mail.. I haven't CC-ed Arthur for this one since this is too much design/implementation details.. Let me know your thoughts when you get some free time.. | 19:02 |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has joined #shogun | 19:03 | |
-!- lzc [~lzc@115.27.197.134] has quit [Quit: Leaving.] | 19:07 | |
@HeikoS | lambday: nice I saw it | 19:08 |
@HeikoS | will read it a bit later today | 19:08 |
@HeikoS | lambday: hows the linalg going you were discussing with the others? | 19:09 |
@lambday | HeikoS: trouble in paradise.. for vectors with smaller dimensions, eigen3 linalg method is performing superbad whereas cmath one (which uses the same thing) performs better | 19:10 |
@lambday | made multiple benchmarks but still not sure about the issue | 19:11 |
@lambday | I always tested with larger dim vectors so never noticed the issue with small ones | 19:11 |
@lambday | gotta check | 19:11 |
@HeikoS | lambday: is it maybe the pass by value of SGVector? | 19:11 |
@lambday | nope | 19:11 |
@HeikoS | good to spot such things | 19:11 |
@lambday | that hardly makes a difference | 19:12 |
@lambday | yes thanks to Saurabh7 | 19:12 |
@HeikoS | huh? | 19:12 |
@HeikoS | what else can it even be then? | 19:12 |
@HeikoS | SIMD? | 19:12 |
@lambday | it's calling eigen3 dot.. | 19:12 |
@HeikoS | maybe eigen3 hinders the c++ optimizer? | 19:12 |
@lambday | both of 'em | 19:12 |
@HeikoS | ah | 19:12 |
@HeikoS | ok | 19:12 |
@HeikoS | so it is linalg | 19:12 |
@lambday | yep | 19:12 |
@HeikoS | not eigen | 19:12 |
@HeikoS | I see | 19:12 |
@HeikoS | thats a good catch! | 19:12 |
@lambday | but how I don't get | 19:13 |
@HeikoS | Saurabh7: nice one | 19:13 |
@HeikoS | no ideas at all? | 19:13 |
@lambday | it's the *almost* exact same code | 19:13 |
@HeikoS | lisitsyn has any ideas? | 19:13 |
@HeikoS | and all the -O3 stuff is the same? | 19:13 |
@lambday | none yet :'( | 19:13 |
@HeikoS | virtual calls or something? | 19:13 |
@lambday | no virtual in linalg | 19:13 |
@HeikoS | mmh | 19:13 |
@lambday | everything is compile-time.. it's supposed to be superfast | 19:14 |
@HeikoS | what about comparing the generated code | 19:14 |
@HeikoS | for a single dot call? | 19:14 |
@lambday | yeah I'll check taht | 19:14 |
@HeikoS | interesteing stuff | 19:14 |
@HeikoS | ha! | 19:14 |
@HeikoS | always surprises in these things | 19:14 |
@lambday | the difference is significant.. like 2x | 19:14 |
@lambday | but if you use vectors of dims = 1000 or more, it virtually becomes 0 | 19:15 |
@HeikoS | ok | 19:17 |
@HeikoS | I see | 19:18 |
Saurabh7 | HeikoS, :) | 19:18 |
Saurabh7 | http://eigen.tuxfamily.org/index.php?title=Benchmark | 19:18 |
Saurabh7 | he reason why the values are typically low for small sizes, is that in this benchmark we deal with dynamic-size matrices which are relatively inefficient for small sizes | 19:18 |
Saurabh7 | dunno the reason but even they have to say the same | 19:18 |
@HeikoS | Saurabh7: but if both methods call the eigen code, there cannot be a difference | 19:20 |
@HeikoS | both methods use eigen dynamic matrices right? | 19:21 |
Saurabh7 | cmath uses different one Eigen::Map<const Eigen::VectorXd> ev1(v1,n); | 19:21 |
@HeikoS | Saurabh7: so that might be it then | 19:22 |
@HeikoS | lambday: thoughts? | 19:22 |
-!- bollu [~Adium@122.179.44.227] has joined #shogun | 19:28 | |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has quit [Ping timeout: 248 seconds] | 19:34 | |
@lambday | aah | 19:34 |
@lambday | he went offline | 19:34 |
@lambday | the d in VectorXd is for dynamic | 19:34 |
@lambday | it's all dynamic afaik | 19:34 |
@lambday | just the const won't make much of a differenc | 19:35 |
@HeikoS | mmh ok | 19:35 |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has joined #shogun | 19:35 | |
@HeikoS | lambday: probably some template programming stuff | 19:35 |
@HeikoS | very implicit | 19:35 |
@HeikoS | so looking at the compiler output might be best | 19:35 |
@lambday | HeikoS: that's something to think about... | 19:35 |
@lambday | HeikoS: I'll try to get objdump | 19:36 |
@lambday | Saurabh7: | 19:36 |
@lambday | the d in VectorXd is for dynamic | 19:36 |
@lambday | it's all dynamic afaik | 19:36 |
Saurabh7 | oh | 19:37 |
@lambday | oh sorry, the d is for double | 19:37 |
Saurabh7 | or is it double ? | 19:37 |
@lambday | the X is for dynamic | 19:37 |
Saurabh7 | yeah xd xf | 19:37 |
@lambday | Eigen::Dynamic, -1 | 19:37 |
@lambday | for static, things are way too fast | 19:38 |
@lambday | Saurabh7: can you try an ugly thing? try linalg::implementation::dot<linalg::Backend::EIGEN3>::compute(a, b) | 19:39 |
@lambday | without the wrapper | 19:39 |
@lambday | sorry it would be | 19:40 |
Saurabh7 | where ? | 19:40 |
@lambday | linalg::implementation::dot<linalg::Backend::EIGEN3, SGVector<float64_t>>::compute(a, b) | 19:40 |
@lambday | instead of linalg::dot(a, b) | 19:40 |
Saurabh7 | lambday, https://gist.github.com/Saurabh7/6ab7c68501697f8de0d6 | 19:44 |
@lambday | :'( | 19:45 |
@lambday | ok tomorrow I am profiling | 19:45 |
@lambday | this is ridiculous | 19:45 |
@lambday | but did the direct call give any advantages over the wrapper? | 19:46 |
@lambday | Saurabh7: ^^ | 19:46 |
Saurabh7 | lambday, looks like it did actually | 19:47 |
@lambday | how much? | 19:48 |
Saurabh7 | sec | 19:48 |
Saurabh7 | https://gist.github.com/Saurabh7/1df1ccbb6bfccd7fff94 | 19:48 |
Saurabh7 | this is before | 19:48 |
@lambday | aha... wrapper is the villain | 19:49 |
Saurabh7 | gottcha | 19:49 |
@lambday | I'd anyway do massive refactoring with linalg soon.. | 19:49 |
@lambday | just the API calls would be same | 19:49 |
@lambday | linalg::dot(...) allways | 19:49 |
@lambday | so let's see | 19:49 |
@lambday | as of now, I'd suggest that you use naked ptr | 19:50 |
@lambday | or cmath | 19:50 |
Saurabh7 | yup sounds good :) | 19:50 |
Saurabh7 | let me knw if u want help | 19:50 |
@lambday | Saurabh7: absolutely | 19:51 |
@lambday | clueless about the reason btw.. this method sets Vector=SGVector<T> and then calls compute... the other method does just that.. | 19:53 |
@lambday | and it's static method, it's not like it creates an object of dot every single time! | 19:53 |
@HeikoS | found it guys? | 19:53 |
@HeikoS | wrapper method | 19:53 |
@lambday | HeikoS: found the culprit.. but not sure why | 19:54 |
@lambday | (!) | 19:54 |
@HeikoS | ok good stuff | 19:56 |
@HeikoS | first step ;) | 19:56 |
@lambday | profiler to the rescue | 19:56 |
-!- cache-nez [~cache-nez@93.175.2.172] has quit [Ping timeout: 244 seconds] | 19:56 | |
-!- cache-nez [~cache-nez@93.175.2.172] has joined #shogun | 19:57 | |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has quit [Remote host closed the connection] | 19:58 | |
-!- leagoetz_ [~leagoetz@host-92-0-162-192.as43234.net] has joined #shogun | 20:03 | |
-!- bollu [~Adium@122.179.44.227] has quit [Quit: Leaving.] | 20:07 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 20:08 | |
-!- leagoetz_ [~leagoetz@host-92-0-162-192.as43234.net] has quit [Remote host closed the connection] | 20:10 | |
-!- bollu [~Adium@122.179.44.227] has joined #shogun | 20:11 | |
-!- cache-nez [~cache-nez@93.175.2.172] has quit [Ping timeout: 252 seconds] | 20:28 | |
-!- lambday [6a338432@gateway/web/freenode/ip.106.51.132.50] has quit [Quit: sleeping] | 20:32 | |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has quit [Ping timeout: 244 seconds] | 20:40 | |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has joined #shogun | 20:42 | |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has quit [Remote host closed the connection] | 20:42 | |
-!- Saurabh7 [~Saurabh7@103.225.100.51] has joined #shogun | 20:42 | |
-!- rahul_red [~rahul__@183.83.51.4] has quit [Quit: Leaving] | 20:45 | |
-!- bollu [~Adium@122.179.44.227] has quit [Quit: Leaving.] | 20:56 | |
-!- wanderer_ [67157d4e@gateway/web/freenode/ip.103.21.125.78] has quit [Quit: Page closed] | 21:01 | |
Ameya | I compiled the tests. 182 of the 512 are failing, most relating to python modular code. I did compile it with cmake -DENABLE_TESTING=ON -DPythonModular=ON .. as required by the README.developer. | 21:16 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 21:19 | |
Ameya | Is this a problem specific to my computer, or is it really a common issue? I tried building it again and verifying that I should not have commited any obvious mistakes. I thought I should really ask it out. | 21:19 |
-!- mode/#shogun [+o HeikoS] by ChanServ | 21:19 | |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has joined #shogun | 21:19 | |
-!- shubham__ [uid57367@gateway/web/irccloud.com/x-ualbppdvjhuqfjrf] has joined #shogun | 21:31 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 21:46 | |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has quit [Remote host closed the connection] | 21:49 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 22:22 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:22 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 22:38 | |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has joined #shogun | 22:38 | |
-!- mode/#shogun [+o HeikoS] by ChanServ | 22:38 | |
@HeikoS | lisitsyn: jojo | 23:09 |
@HeikoS | I tried to fix this but failed | 23:09 |
@HeikoS | https://travis-ci.org/shogun-toolbox/shogun/jobs/113893887#L3121 | 23:09 |
@HeikoS | maybe you can have a look | 23:09 |
@HeikoS | ? | 23:09 |
@HeikoS | thanks! | 23:09 |
-!- HeikoS [~heiko@host-92-0-162-192.as43234.net] has quit [Quit: Leaving.] | 23:09 | |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has joined #shogun | 23:24 | |
-!- albedo [~textual@191.189.12.228] has joined #shogun | 23:27 | |
-!- leagoetz [~leagoetz@host-92-0-162-192.as43234.net] has quit [Remote host closed the connection] | 23:34 | |
--- Log closed Sun Mar 06 00:00:25 2016 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!