TsotsosLab
[Docker][SalGAN][eDN][oSALICON] Models are not GPU accelerated
Docker models SalGAN, eDN and oSALICON are not GPU accelerated.
Observed slow processing speed per image and verified by testing with nvidia-smi -l 1:
| NVIDIA-SMI 384.130 Driver Version: 384.130 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GeForce GTX 1070 Off | 00000000:B3:00.0 On | N/A |
| 0% 52C P8 8W / 151W | 525MiB / 8112MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| 0 14534 G /usr/lib/xorg/Xorg 306MiB |
| 0 14755 G 13MiB |
| 0 15173 G compiz 203MiB |
+-----------------------------------------------------------------------------+
While all other docker models are able to reach between 40%-100% utilization, these models report utilization of 0%-4% meaning the gpu is not being utilized.
That's odd, they run on the GPU on my machine. Can you please send me the experiment file / run command you are using? You can email it to me if you prefer not posting it here.
I ran via the run command, but I observe the same thing when running from a yaml config file (default settings).
smiler run -m "salgan,edn,osalicon" Exp/12-17/input Exp/12-17/output-docker