Run deeplearning4j without GPU?

Question

The documentation says the library runs on GPUs. If my powerful laptop doesn't have a GPU, can I still run Deeplearning4J?

Answer 1

We don't imply anywhere in our docs that we only run on gpus. If for some reason you found that to be the case, could you notify us in a github issue? https://github.com/deeplearning4j/deeplearning4j/issues

You only need an nd4j backend: http://nd4j.org/backend.html

Dl4j unlike a lot of other libraries decouples the hardware implementation from the algorithms. The ticket here is our tensor library nd4j. Nd4j handles all the computation.

Think of it is "tensorflow/theano" if you will.

We deploy all of our native binaries to maven central.

Usually all you need is: http://search.maven.org/#search%7Cga%7C1%7Cnd4j-native-platform

for cpu. GPU, we support the 2 latest cuda versions: http://search.maven.org/#search%7Cga%7C1%7Cnd4j-cuda-7.5-platform

The same exact code as is also works on android and pis: http://deeplearning4j.org/android

Same idea, you don't compile from source, just specify the right backend.