All geek questions in one place

TensorFlow HVX Acceleration Support

I have successfully built and launched a test application from https://github.com/tensorflow/tensorflow/tree/master/tensorflow/contrib/hvx . Now I would like to compare HVX with the processor implementation https://github.com/tensorflow/tensorflow/tree/master/tensorflow/tools/benchmark and, if possible, a demonstration of the Android camera to find out how much this will help, but I could not find the documentation that describes how to create these applications with HVX support (my assemblies are executed on the processor). I am testing on the Open-Q 820 development platform with Android 7.0.

Does it use HVX acceleration outside the HVX test application, preferably with benchmarking and, possibly, supported Android demos? If yes, can someone please call me in the right direction? Thank!

+4
source share
1 answer

Currently, the Android demo application does not support the HVX environment. But I'm sure you can use the runtime with the Android demo application by replacing the .so file with the HVX version. If you can wait for official support, this will happen soon, but no promises. Let me know if you have any questions :)

+2
source

Source: https://habr.com/ru/post/1672908/

More articles:


All Articles