Is your Python script refusing to use all CPU scores on your server, and just using 1 CPU core out of the 16 CPU cores?

It could be that the core affinity is messed up.

As it turns out, some modules mess with the core affinity on import, which end up forcing your Python script to use only one CPU core in the end. Just add this before you run the Python script:

<span class="pln">os</span><span class="pun">.</span><span class="pln">system</span><span class="pun">(</span><span class="str" style="color: #800000;">"taskset -p 0xffffffff %d"</span><span class="pun">%</span><span class="pln"> os</span><span class="pun">.</span><span class="pln">getpid</span><span class="pun">())</span>

And then voila! It looks right now!

10495829_10204248476553916_8213806856117433374_o

You can find more details in this StackOverflow thread.