search

manuelruder / artistic-videos

Torch implementation for the paper "Artistic style transfer for videos"
Other
1.75k stars 252 forks source link

deepmatching-static segfaults on ubuntu 16.04 #38

Closed Teepareep closed 7 years ago

Teepareep commented 7 years ago

Hello,

I migrated my setup to a fresh machine and am running into a strange issue with deepmatching segfaulting shortly after running (10 minutes?).

Ubuntu 16.04 Linux mlart 4.4.0-59-generic #80-Ubuntu SMP Fri Jan 6 17:47:47 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux GTX Titan X (Maxwell) CPU: Intel(R) Core(TM) i7-5775C CPU @ 3.30GHz

I installed systemd-coredump and it spins up a bunch of processes as soon as deepmatching-static starts. Here's an example coredump. I'm not very familiar with debugging these, so sorry for the noob question here.

$ coredumpctl info 8618 PID: 8618 (deepmatching-st) UID: 0 (root) GID: 0 (root) Signal: 11 (SEGV) Timestamp: Tue 2017-01-17 12:16:17 PST (30min ago) Command Line: ./deepmatching-static /content/vid/example/frames/frame_0004.ppm /content/vid/example/frames/frame_0005.ppm -nt 0 Executable: /root/artistic-videos/deepmatching-static Control Group: /docker/922e2593770bb0190376fdae5482d60ff373a0aa32565f3d84e33b482e1a0c3b Slice: -.slice Boot ID: 156c400d834845b2a36f1e00da8dcf63 Machine ID: 57adaac235924da8aeaa7dd1b3a478e9 Hostname: mlart Coredump: /var/lib/systemd/coredump/core.deepmatching-st.0.156c400d834845b2a36f1e00da8dcf63.8618.1484684177000000000000.xz Message: Process 8618 (deepmatching-st) of user 0 dumped core.

            Stack trace of thread 329:
            #0  0x0000000000457bb0 n/a (/var/lib/docker/aufs/diff/f9fada9ab7f83d8797c8b483087cce5fdfb9f61eabe3e9a8f5fa4649103e7989/root/artistic-videos/deepmatching-static)

I'm running this with nvidia-docker, which was working great on my previous machine. I thought I'd ping this repo about possible solutions before I go ordering more RAM and start swapping hardware.

Teepareep commented 7 years ago

Extremely strange but when I do the quarter-res optical flow calculations mentioned in issue 34 I no longer seem to be having this issue. Going to test further.

manuelruder commented 7 years ago

Deepmatching is very memory intensive. quarter-res reduces memory consumption of course. If your video has a high resolution, I don't see a reason why not just stick with quarter-res.

Teepareep commented 7 years ago

I think this is just a hardware issue. Closing for now.