Cannot get Docker image to recognize Caffe

AI & Data Science Deep Learning (Training & Inference) DIGITS (Locked)
1

I am trying to run the Caffe version of the nvcr.io/nvidia/digits:19.10-caffe Docker image but can’t get it to recognize Caffe. Can someone please let me know what I’ve got wrong?

Under the Info menu it shows:
DIGITS version:
6.1.1
Caffe version:
Unavailable
Caffe flavor:
Unavailable

Here is the log:
pelizani@pelizani-OptiPlex-9010-AIO:~/GitHub/csce590$ docker run -p 8888:5000 -v /home/pelizani/digits/data:/data -v /home/pelizani/digits/jobs:/workspace/jobs nvcr.io/nvidia/digits:19.10-caffe

============
== DIGITS ==

NVIDIA Release 19.10 (build 8221577)
DIGITS Version 6.1.1

Container image Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
DIGITS Copyright (c) 2014-2019, NVIDIA CORPORATION. All rights reserved.

Various files include modifications (c) NVIDIA CORPORATION. All rights reserved.
NVIDIA modifications are covered by the license terms that apply to the underlying project or file.

WARNING: The NVIDIA Driver was not detected. GPU functionality will not be available.
Use ‘nvidia-docker run’ to start this container; see
nvidia docker · NVIDIA/nvidia-docker Wiki · GitHub .

| _ / | | / |
| |) | | ( || | | | _
|/_|| || |__/ 6.1.1

Caffe support disabled.
Reason: A valid Caffe installation was not found on your system.
cudaRuntimeGetVersion() failed with error #35
Tensorflow support disabled.
2019-12-03 08:02:22 [INFO ] Loaded 9 jobs.
2019-12-03 08:02:22 [WARNING] Failed to load 3 jobs.
2019-12-03 08:02:22 [DEBUG] 20191203-040958-b674 - ImportError: No module named absl.logging
2019-12-03 08:02:22 [DEBUG] 20191203-043738-fca8 - ImportError: No module named absl.logging
2019-12-03 08:02:22 [DEBUG] 20191203-062847-cb92 - ImportError: No module named absl.logging

2

Peter,

Looks like you aren’t hooking in the GPU capability to docker. See more at GitHub - NVIDIA/nvidia-docker: Build and run Docker containers leveraging NVIDIA GPUs but in short, add --gpus to your command line. For example:
% docker run --gpus all nvidia/cuda:9.0-base nvidia-smi

or in your case
% docker run --gpus -p 8888:5000 -v /home/pelizani/digits/data:/data -v /home/pelizani/digits/jobs:/workspace/jobs nvcr.io/nvidia/digits:19.10-caffe

3

Thanks, @michaelo!

The machine that I am running it on doesn’t have GPUs. Is it required to run with a GPU? The Tensorflow version of the image works.

Peter

4

very helpful document…looking for it for a long time.

5

Hi,

On the same line, trying to run DIGITS on GCP, facing caffe support disabled issue, please advise

sudo docker run --gpus all -it --rm -v /home/workspace/nvprotrain/dataset:/home/workspace/nvprotrain/dataset nvcr.
io/nvidia/digits:20.02-tensorflow-py3

== DIGITS ==

NVIDIA Release 20.02 (build 9979494)
DIGITS Version 6.1.1
Container image Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
DIGITS Copyright (c) 2014-2019, NVIDIA CORPORATION. All rights reserved.
Various files include modifications (c) NVIDIA CORPORATION. All rights reserved.
NVIDIA modifications are covered by the license terms that apply to the underlying project or file.

| _ / | | / |
| |) | | ( || | | | _
|/_|| || |__/ 6.1.1
Caffe support disabled.
Reason: A valid Caffe installation was not found on your system.
2020-03-06 05:43:16.961125: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.2