search

yaroslavvb / tensorflow-community-wheels

Place to upload links to TensorFlow wheels
454 stars 35 forks source link

TensorFlow 2.4.1, No AVX, GPU (Compute 6.1), CUDA 11.1, cuDNN 8.0.5, TensorRT 7.2.3, NCCL 2.9.6, Python 3.6, Ubuntu 18.04, x86_64 #186

Open adanhawth opened 3 years ago

adanhawth commented 3 years ago

https://github.com/adanhawth/tensorflow-builds/releases/tag/v2.4.1

profMastermind commented 2 years ago

Tried with CUDA 11.4, works like charm .. many thanks

phirestalker commented 1 month ago

I tried this in a Docker container that is running on Debian 12. The CUDA version on it was 11.8, and the CUDNN version was 8.2.1. It says the wheel is not compatible.

Does anyone know if there is a way to get it to work? Does the wheel need to be compiled for each OS? This is going to be for the community Nextcloud docker image. Any ideas on how to get a wheel for that? Oh and I have a GTX 680 which is only capable of compute 3.0.

adanhawth commented 1 month ago

I tried this in a Docker container that is running on Debian 12. The CUDA version on it was 11.8, and the CUDNN version was 8.2.1. It says the wheel is not compatible.

Does anyone know if there is a way to get it to work? Does the wheel need to be compiled for each OS? This is going to be for the community Nextcloud docker image. Any ideas on how to get a wheel for that? Oh and I have a GTX 680 which is only capable of compute 3.0.

Your requirements seem sufficiently different than those for which these bits were created. I think I'd be more surprised if it did work for you. I'd suggest either one of two options: you can roll your own, or you can create an issue in this repo and ask for bits. Best of luck.