Increasing compute performance of Jetson TX 1

First, let me thank the forum users for being so responsive to my questions! I have another one…

Currently I am trying use a Jetson TX 1 board for neural net training using Theano. Finally, I’ve gotten it reasonably configured so that Theano reports that it is using the Jetson’s GPU and is running my code. A benchmark test I have shows that the speedup factor from doing the same training on my Intel Core i5 processor is about 5.7 in going to the Jetson’s GPU cores.

I’m basically posting to feel out–does this seem reasonable, or low? Also, I notice that the Jetson board does not seem to be getting particularly warm, is there some way to increase the computing power?

I noticed this forum post:
https://devtalk.nvidia.com/default/topic/901337/cuda-7-0-jetson-tx1-performance-and-benchmarks/
It has a “max perf script”–a shell script that seems to set options in device driver files. Are these options documented anywhere?

Thanks so much!

Hi cclauson,

You can refer to Jetson performance wiki [url]http://elinux.org/Jetson/Performance[/url]
The “max perf script” is just for you easy to use, please see below post related to the script:
[url]https://devtalk.nvidia.com/default/topic/894945/jetson-embedded-systems/jetson-tx1/post/4740508/#4740508[/url]

Thanks