tfjs-node-gpu invalid ELF header on Windows

Question

I'm trying to run tensorflow/tfjs-examples/mnist-node on Windows with GPU.

After cloning the repository and installing the dependencies, I can run the example with CPU. However, when I tried to install @tensorflow/tfjs-node-gpu and run the example, here's what I get:

Registration of backend tensorflow failed
Error: The specified module could not be found.    
\\?\C:\Users\noob\Desktop\tfjs-examples\mnist-node\node_modules\@tensorflow\tfjs-node-gpu\build\Release\tfjs_binding.node
    at Object.Module._extensions..node (internal/modules/cjs/loader.js:718:18)
    at Module.load (internal/modules/cjs/loader.js:599:32)
    at tryModuleLoad (internal/modules/cjs/loader.js:538:12)
    at Function.Module._load (internal/modules/cjs/loader.js:530:3)
    at Module.require (internal/modules/cjs/loader.js:637:17)
    at require (internal/modules/cjs/helpers.js:20:18)
    at bindings (C:\Users\noob\Desktop\tfjs-examples\mnist-node\node_modules\bindings\bindings.js:81:44)
    at C:\Users\noob\Desktop\tfjs-examples\mnist-node\node_modules\@tensorflow\tfjs-node-gpu\dist\index.js:10:60
    at Environment.registerBackend (C:\Users\noob\Desktop\tfjs-examples\mnist-node\node_modules\@tensorflow\tfjs-core\dist\environment.js:188:27)
    at Object.<anonymous> (C:\Users\noob\Desktop\tfjs-examples\mnist-node\node_modules\@tensorflow\tfjs-node-gpu\dist\index.js:9:9)

According to tfjs-node, it seems like I've already fit the requirements. Here's my environments:

Environment:

node v10.9.0
nvcc 9.2 V9.2.148
Windows 10 version 1803 build 17738.1000

NoobTW · Accepted Answer · 2018-09-02 04:57:08Z

1

OK, Problem solved. I have to install CUDA v9.0 instead of v9.2. After uninstall and reinstall it works again.

Here's some tutorial if someone have similar problem.

First download CUDA 9.0 and cuDNN v7 for CUDA 9.0, install CUDA 9.0.
Extract cuDNN files to your CUDA installation folder. (Default to C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v9.0)
Add C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v9.0\bin and C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v9.0\lib\x64 to PATH.
run node main in tfjs-examples\mnist-node, it would work now.

At the first time, it may take about 2~5 minutes stuck at Adding visible gpu devices: 0, just wait for it. I'll work.

BTW, you cannot use GPU in WSL mode yet.

answered Sep 2, 2018 at 4:57

NoobTW

2,6242 gold badges27 silver badges46 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

tfjs-node-gpu invalid ELF header on Windows

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related