GPU Support for Deep Learning

KNIME Extensions
1

I’m trying to activate GPU support to run the examples in chapters 7 & 8 in the Codeless Deep Learning book. The models run, but don’t use the GPU. I have CUDA installed. My Keras and Tensorflow environments are attached. I’m an analyst, not a computer expert. Any help would be appreciated.

Keras Environment.txt (7.3 KB)

Tensorflow Environment.txt (8.4 KB)

2

Hi,

Do you already know the Deep Learning Guide, in it the setup of GPU Support is described, e.g. for Keras: https://docs.knime.com/2020-07/deep_learning_installation_guide/index.html#keras_gpu_support

This should help you in setting the environment up. An overview over the compatible graphic cards is also linked.

Best,
Michael

3

I’ve read the installation guide and obviously its not much help since I’m still having problems. Here’s my Deep Learning installation. It points to the Keras and Tensorflow environments I sent you previously.

Deep Learning Setup
Deep Learning Setup915×972 111 KB

4

Hi @rfeigel,

This line in Keras Environment.txt:

tensorflow-mkl 1.12.0 h4fcabd2_0

suggests that your Keras environment is set up to use the CPU (the GPU-enabled version of TensorFlow would be named tensorflow-gpu).

Could you please try to create a new GPU-enabled Keras environment via “New environment…” > “Create new GPU environment” in the Python Deep Learning preferences and see if this resolves the issue?

Marcel

1 Like
5

Thanks. I created the new environment and according to Windows Task Manager, the GPU is active. Its not much faster, but I suspect that’s because of my low level GPU.

2 Likes
6

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.