verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
< naywhayare>
JaraxussTong: thanks for closing #388, I'd forgotten :)
< JaraxussTong>
^.^
< naywhayare>
zoq: how can I trigger the benchmark runs for allknn? I made some changes to the tree code recently, and I wanted to see if it made a difference
awhitesong has joined #mlpack
pgunnars has joined #mlpack
< pgunnars>
why is installing mlpack with a local boost lib installation such a god damn hassle
< pgunnars>
whattafakkkk
< awhitesong>
naywhayare, stephentu: Hi! Sorry for being off long :) Did you check the MDS code, is it fast enough? I know there are some formating changes to be made according to mlpack. Are the test cases fine? Thanks!
pgunnars has quit [Ping timeout: 246 seconds]
< naywhayare>
awhitesong: sorry, I haven't had a chance to look into it yet... things have been very busy
< naywhayare>
I have some paper deadlines coming up this month, but let me take a quick look now
< naywhayare>
unhappy pgunnars... hope he comes back, it shouldn't be that hard to compile mlpack with local boost...
awhitesong has quit [Ping timeout: 240 seconds]
awhitesong has joined #mlpack
< naywhayare>
awhitesong: I left a comment, so that's at least a place to start
< awhitesong>
naywhayare : thanks for the input. sure will look into it. :)
< naywhayare>
I'm going to try to set aside some time to review this code, but please, remind me if I don't get around to it
< naywhayare>
it's easy for me to just forget things because I have so many things to do...
< awhitesong>
:) sure. Its going a bit busy here too with my exams. Will notify as soon as i get to something. Is superLU kept for a while?
< naywhayare>
SuperLU is a nightmare...
< naywhayare>
I got spsolve() implemented, but getting lu() implemented is very difficult because I can't get SuperLU to give me back a reasonable L matrix
< naywhayare>
I may be able to work around this for generalized eigendecomposition though, but, I won't be able to provide lu() soon
< awhitesong>
ok . ya indeed superlu seems pretty complex. meanwhile with the tests, i could look into ball trees and djikstra's for Isomap. I guess scikit has a decent implementation for Isomap with these. non-metric MDS is still left though. thanks again!
< naywhayare>
yeah, superlu is somewhat poorly designed...
awhitesong has quit [Ping timeout: 244 seconds]
< zoq>
naywhayare: sorry for the slow reponse ... I guess, we can use the main benchmark database? Or another database?
< zoq>
hm, jenkins shows that the shogun benchmark job is still running but the benchmark finished days ago.
< naywhayare>
ah, okay, so maybe this was already done, but I don't remember seeing anything from jenkins-mlpack here in IRC
< naywhayare>
I guess I could look at the nightly benchmark just as easily :)
< zoq>
ah yeah the nightly builds
< zoq>
naywhayare: maybe a jenkins restart on shoeshine solves the problem not sure
< zoq>
naywhayare: btw. the command to run a specific method is: 'make run BLOCK=mlpack LOG=True UPDATE=True COPY=True METHODBLOCK=ALLKNN'
travis-ci has joined #mlpack
< travis-ci>
mlpack/mlpack#107 (master - 148bfca : Ryan Curtin): The build passed.
< zoq>
no benchmark process is currently running on shoeshine, this is strange
< naywhayare>
zoq: ok, thanks for the command
< naywhayare>
I see that the jenkins process is running on shoeshine
< naywhayare>
load average on shoeshine is 7...
< zoq>
I can't see a python benchmark process
< naywhayare>
I meant just the java slave agent
< naywhayare>
what I noticed was that update-svn.sh (for Armadillo) was running in I think six processes
< naywhayare>
all of which had been started in the past six days
< naywhayare>
so I killed those processes, because only the NFS server should be updating armadillo svn
< naywhayare>
and then suddenly the system started compiling
< naywhayare>
I think it's getting ready to run some benchmarks, but I'm not certain
< naywhayare>
I commented out the update-svn.sh cron job (which I should have done months ago...) so hopefully that problem will not happen again
< naywhayare>
ah, yeah, it's starting the benchmark - mlpack - commit process now
< naywhayare>
so I guess the problem was that update-svn.sh was hanging or something like this ? I'm not certain
< zoq>
great, I wonder why the update-svn holds back the entire benchmark build
< naywhayare>
my best guess was that it was hanging because the directory it was trying to update is mounted via NFS, and NFS can be weird and irritating sometimes
< naywhayare>
that's not really a great or precise answer, but it's probably a stab in the right direction :)