1

I am trying to install CUDNN to use tensorflow with GPU support on my Macbook Pro. After downloading cudnn from NVDIA website, these are all the instructions i could find:

OS X

cd <installpath>
export DYLD_LIBRARY_PATH=`pwd`:$DYLD_LIBRARY_PATH

Add <installpath> to your build and link process by adding -I<installpath> to your compile
line and -L<installpath> -lcudnn to your link line.

While, the first step is really straightforward, what is the meaning of second statement and how do I finish the second step? Please advise. Thanks.

talonmies
  • 70,661
  • 34
  • 192
  • 269
Shimano
  • 785
  • 1
  • 6
  • 13
  • Does this answer your question? [How to verify CuDNN installation?](https://stackoverflow.com/questions/31326015/how-to-verify-cudnn-installation) – Channa Jan 18 '21 at 06:30

1 Answers1

2

So I was stuck at the same place and what I found was that the cudnn archive basically had just two library files a : .h and .so files. so I figured it was best to put them someplace that is already looked at and I found this :

Untar the folder and copy the necessary files to the existing cuda 
library we installed earlier.

tar -xzvf cudnn-7.0-linux-x64-v4.0-prod.tgz
cp cuda/lib64/* /usr/local/cuda/lib64/
cp cuda/include/cudnn.h /usr/local/cuda/include/

so I guess u should find an equivalent for mac

source:

https://medium.com/@acrosson/installing-nvidia-cuda-cudnn-tensorflow-and-keras-69bbf33dce8a

praneeth mendu
  • 369
  • 3
  • 17
  • This worked for me on my macbook pro (OSX 10.11.6). I just copied the handful of files to their respective directories in the main CUDA installation. For me (using defaults), these were the dirs, '/Developer/NVIDIA/CUDA-8.0/bin' '/Developer/NVIDIA/CUDA-8.0/lib' – user2348114 May 08 '17 at 22:59