These are chat archives for beniz/deepdetect

25th
Jun 2018
Emmanuel Benazera
@beniz
Jun 25 2018 15:39
@cchadowitz-pf hi, if you rebase your PR, it should trigger a CI test on our side, and then be good for a merge. Thanks!
cchadowitz-pf
@cchadowitz-pf
Jun 25 2018 16:13
:+1: i'll get on that now!
currently investigating a situation where (on GPU) reusing a dlib service (with the model stored in memory) seems to slowly use more GPU memory over time (although between calls the memory usage will fluctuate up and down, the overall memory usage seems to trend upwards)
doesn't seem to happen when i try to replicate it with a simple app using dlib calls (to take deepdetect out of the mix). so i'm not sure if i'm doing something strange when I keep the model in memory in deepdetect
cchadowitz-pf
@cchadowitz-pf
Jun 25 2018 16:37
@beniz should i pay any attention to the travis builds or are those not really relevant at the moment b/c of your local CI tests?
cchadowitz-pf
@cchadowitz-pf
Jun 25 2018 17:10
oops, just noticed the jenkins progress, never mind!
Emmanuel Benazera
@beniz
Jun 25 2018 17:56
Yes unfortunately the Jenkins access is from our vpn
cchadowitz-pf
@cchadowitz-pf
Jun 25 2018 19:10
no problem
looks like the build may have failed though :-/
Emmanuel Benazera
@beniz
Jun 25 2018 19:17
I'll look a it
Emmanuel Benazera
@beniz
Jun 25 2018 19:27
nothing to do with your code I think, I've restarted it. Should take another 25 mins
cchadowitz-pf
@cchadowitz-pf
Jun 25 2018 19:27
:+1: always good to hear ;)