Caching TensorFlow models to C:\Users\s3886299\Desktop\Fiji.app\models It gives me this error: Using native TensorFlow version: TF 1.15.0 GPU (CUDA 10.0, CuDNN >= 7.4.1) I feel like I’m 95% “there” for being able to use N2V on my GPU, but… Hey there - thanks to everyone for all this troubleshooting and giving so many helpful hints. Now everything seems to work and I can see activity for my GPU at the TaskManager. From Fiji I used the TensorFlow 1.15 and of course I added all necessary paths to the PATH variable. I cannot understand where is the incompatibility… Any help maybe?Īfter uninstalling everything related with nVidia (and after a few reboots) I only installed the CUDA 10.1 and the cuDNN 7.6.5. Status: CUDA driver version is insufficient for CUDA runtime version Whenever I am trying to use the plugin however, I get the following error: Added the necessary variables to my PATH.Installed cuDNN 7.6.5 (I also tried with 7.4.x).Trying to use also our GPUs I did exactly all the steps that you described already: I did use it a bit using only the CPu and the results looked great. First of all, what a great effort for this very useful tool.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |