0.12 worked for us and someone else got 1.0 working as well. Some of those patches may not need to be applied for 1.0/0.12 or you may need to do some searching in the files to apply it in the correct place.
Thanks for the link!
Now I’m stuck at building the tensorflow as it requires a lot of memory. I tried several different way to allocate memory, ‘fallocate’ which stops with error Operation not supported and with ‘dd’ I couldn’t activate the memory with ‘sudo swapon’ of function not implemented. Am I missing some package like L4T for Drive PX?
Thanks for the link! But I’m still stuck at the memory allocation. fallocate gives me, “fallocate failed: Operation not supported” and I tried your solution nvok. For those that aren’t aware I’m running on a Drive PX ONE, fresly flashed.
Also, I adjust your resource allocation parameters for “bazel build” to “–local_resources 8192,2.0,1.0”. But it still does not work. The system always crashes at the halfway of installation. Could you help me with this situation?
Does anyone still have the copy to the link stated above, or at least a way to install tensor flow right from the beginning on drive PX2 I have already tried multiple links related to the TX2/TX1/TK1 but all my efforts have been useless as of now. Also the link has become inaccessible at the moment.
It turns out there is a problem with Bazel, as far as I can see. I have already changed the --local_resources variable as stated above. Didn’t work out though. Thanks.
we tried to use TensorFlow with the docker-container before some months and also failed. In this Forum at the Topic [url]https://devtalk.nvidia.com/default/topic/1027904/?comment=5228522[/url] I got the reply, that there is an issue with the DriveInstall 5.0 and this TensorFlow container.
Hello,@serinvarghese, could you tell me how you installed the tensoflow1.7 on the DPX2? Did you use the conmand "pip install tensorflow-1.7.0-cp35-cp35m-linux_aarch64.whl “