![]() NVRM version: NVIDIA UNIX x86_64 Kernel Module 440.82 Wed Apr 1 20:04: | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ‘ nvida-smi‘ command to verify your GPU is accessible. ![]() Run the following commands (thanks to this page) So here is how to freeze cuda versions, so they don’t automatically get upgraded $ sudo apt-mark hold cuda-10.1Įxport LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/extras/CUPTI/lib64 Verify NVidia StackĪfter doing the above, be sure to reboot your machine So we don’t want CUDA updated when we update the ubuntu system (by using sudo apt update & sudo apt upgrade) It is very important to maintain CUDA libraries at the supported version. This is OK, Tensorflow works with CUDA 10.2 as well. Note: CUDA might be updated to 10.2, when you update the Ubuntu system (sudo apt-get update sudo apt-get upgrade). Sudo apt-get install -y -no-install-recommends libnvinfer6=6.0.1-1+cuda10.1 \ Requires that libcudnn7 is installed above. Sudo apt-get install -no-install-recommends \ # Install development and runtime libraries (~4GB) Check that GPUs are visible using the command: nvidia-smi Sudo apt-get install -no-install-recommends nvidia-driver-430 I am copying the code here for completeness. Tensorflow v2.1 works with CUDA 10.1 (and 10.2) as of this writing It is * very important* that you install the right version of NVidia stack. Hardware : Nvidia RTX 2070 8GB ( see available products on Amazon) Here is the final setup to help out anyone who is looking to do the same. It took a lot of effort, a lot of Googling and a lot of experimenting. I recently got GPU version of Tensorflow working on my ubuntu machine. This guide is verified as of 2020 May, with Tensorflow version 2.1.0 Background
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |