--- Log opened Tue Oct 18 00:00:06 2011 | ||
blackburn | sonney2k: how can one use uint64_t? | 00:10 |
---|---|---|
shogun-buildbot | build #238 of octave_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/octave_modular/builds/238 | 00:11 |
-!- blackburn [~blackburn@188.168.4.221] has quit [Quit: Leaving.] | 00:57 | |
-!- Netsplit *.net <-> *.split quits: sonne|work, shogun-buildbot | 02:51 | |
-!- Netsplit over, joins: shogun-buildbot, sonne|work | 02:55 | |
-!- blackburn [~blackburn@188.168.4.77] has joined #shogun | 09:16 | |
CIA-3 | shogun: Soeren Sonnenburg * rdb8a1a4 / src/configure : add int32_t and int64_t to default data types - http://git.io/pQRAfw | 09:24 |
blackburn | sonney2k: I guess you forgot to disable unnecessary preprocessors | 09:26 |
@sonney2k | ? | 09:26 |
blackburn | sonney2k: they are templated | 09:27 |
@sonney2k | blackburn, please do then | 09:27 |
blackburn | yeah | 09:27 |
blackburn | sonney2k: what is the alphabet for uint64? | 09:29 |
@sonney2k | can be any | 09:29 |
blackburn | sonney2k: oh sorry I know what serializable dummy is | 09:33 |
blackburn | how can I forgot that | 09:33 |
blackburn | it is macro adding load and save to datatypes | 09:34 |
blackburn | to make it work in java | 09:34 |
@sonney2k | you mean the serialization hooks? | 09:34 |
@sonney2k | for classes that don't support it right? | 09:34 |
blackburn | sonney2k: yes | 09:34 |
@sonney2k | k | 09:34 |
blackburn | hehe I was too tired yesterday | 09:35 |
@sonney2k | blackburn, if you have a better idea on how to beautify this ifdef USE_FLOAT64 mess - feel free to improve things | 09:39 |
blackburn | sonney2k: no idea | 09:39 |
@sonney2k | maybe a macro that produces the %template with the right name / type or so | 09:39 |
blackburn | sonney2k: I don't think it is necessary | 09:40 |
blackburn | we don't change it too much often | 09:40 |
@sonney2k | when drastically limiting data types we can get it to compile requiring <1.5GB | 09:40 |
@sonney2k | so at least sth | 09:40 |
blackburn | sonney2k: have you measured? | 09:40 |
shogun-buildbot | build #223 of java_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/java_modular/builds/223 | 09:41 |
@sonney2k | sure | 09:41 |
@sonney2k | did you see my posts yesterday? | 09:42 |
@sonney2k | <sonney2k> blackburn, ok we are down to 375kLoC (from 531kLoC) | 09:42 |
@sonney2k | <sonney2k> when compiling with datatypes=char,uint8_t,uint16_t,uint64_t,float32_t,float64_t | 09:42 |
@sonney2k | <sonney2k> but examples use all types... | 09:42 |
@sonney2k | <sonney2k> so it is only a poor mans solution | 09:42 |
@sonney2k | <sonney2k> but yes better than nothing | 09:42 |
blackburn | yes I've seen that | 09:42 |
blackburn | but have you measured required memory? | 09:42 |
blackburn | sonney2k: http://www.shogun-toolbox.org/buildbot/builders/java_modular/builds/223/steps/compile/logs/warnings%20%289%29 | 09:43 |
@sonney2k | yes | 09:43 |
@sonney2k | no idea what that is | 09:43 |
blackburn | sonney2k: I have narrowed CFeatures to CSimpleFeatures at EmbeddingConverter | 09:43 |
blackburn | in return type of apply | 09:43 |
blackburn | i.e. generally converter returns Features | 09:43 |
blackburn | but EmbeddingConverter always returns CSimpleFeatures | 09:44 |
blackburn | and java do not support it | 09:44 |
blackburn | well no idea how to avoid it | 09:44 |
@sonney2k | blackburn, why not always return CFeatures? | 09:45 |
@sonney2k | and add special dynamic casts to return simple features? | 09:45 |
blackburn | sonney2k: where to add dyn cast?? | 09:46 |
@sonney2k | blackburn, you could add a function e.g. get_converted_simplefeatures() or so to your embeddingconverter | 09:47 |
@sonney2k | and that will just call apply() and convert the result of apply to simplefeatures | 09:47 |
blackburn | embed() | 09:48 |
@sonney2k | other naming suggestions - get_applied_as_simplefeatures | 09:48 |
@sonney2k | or embed | 09:48 |
@sonney2k | yes | 09:48 |
blackburn | sonney2k: you're such a big fan of long names ;) | 09:48 |
@sonney2k | I am a fan of names I can make sense of | 09:48 |
@sonney2k | embed doesn't tell me anything | 09:49 |
shogun-buildbot | build #222 of python_modular is complete: Success [build successful] Build details are at http://www.shogun-toolbox.org/buildbot/builders/python_modular/builds/222 | 09:53 |
blackburn | sonney2k: so what is the memory reqs for compilation now? | 09:58 |
@sonney2k | blackburn, depends on the number of types you enable | 10:00 |
@sonney2k | with the few types above it was around 1.5G for octave_modular | 10:00 |
@sonney2k | and around 1G for python_modular | 10:00 |
blackburn | sonney2k: before? | 10:00 |
@sonney2k | ~3.5 GB | 10:02 |
blackburn | oh | 10:02 |
blackburn | nice | 10:02 |
CIA-3 | shogun: Sergey Lisitsyn * r115a78d / src/interfaces/modular/Preprocessor.i : Added checking for enabled types to preprocessor modular interfaces - http://git.io/cNYLQA | 10:20 |
CIA-3 | shogun: Sergey Lisitsyn * rea9ad9a / (26 files in 4 dirs): Added embed method to EmbeddingConverter, adapt structure - http://git.io/X7UWTQ | 10:20 |
blackburn | sonney2k: one more algo | 11:21 |
CIA-3 | shogun: Sergey Lisitsyn * r99027e8 / (6 files in 3 dirs): Introduced NeighborhoodPreservingEmbedding (NPE) - http://git.io/9FAPpQ | 11:21 |
@sonney2k | ...and counting | 11:21 |
blackburn | sonney2k: I'm not sure it is very useful though | 11:22 |
-!- blackburn [~blackburn@188.168.4.77] has quit [Quit: Leaving.] | 11:24 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 18:12 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has quit [Ping timeout: 244 seconds] | 18:40 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 18:40 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has quit [Ping timeout: 256 seconds] | 19:18 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 19:18 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has quit [Ping timeout: 276 seconds] | 19:34 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 19:35 | |
-!- blackburn [~blackburn@188.168.2.167] has joined #shogun | 20:35 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has quit [Ping timeout: 248 seconds] | 20:55 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has joined #shogun | 20:57 | |
-!- alesis-novik [~alesis@cpat001.wlan.net.ed.ac.uk] has quit [Ping timeout: 256 seconds] | 21:04 | |
-!- alesis-novik [~alesis@5ac57106.bb.sky.com] has joined #shogun | 21:27 | |
-!- Netsplit *.net <-> *.split quits: alesis-novik | 22:23 | |
-!- Netsplit over, joins: alesis-novik | 22:24 | |
-!- heiko [~heiko@host109-152-233-34.range109-152.btcentralplus.com] has joined #shogun | 22:27 | |
heiko | sonney2k, around? | 22:27 |
-!- heiko [~heiko@host109-152-233-34.range109-152.btcentralplus.com] has quit [Quit: Leaving.] | 22:33 | |
--- Log closed Wed Oct 19 00:00:12 2011 |
Generated by irclog2html.py 2.10.0 by Marius Gedminas - find it at mg.pov.lt!