Some of us at
https://CoCalc.com did some poking around and noticed
the following about
gryd.us, in case it is of interest:
"I did play around with them for a while. Looks indeed interesting,
they also just offer the classical jupyter interface, but at least
they've enabled various extensions and seem to also have a strong
focus on teaching.
I'm also always curious about what's under their hood regarding
software. It surprises me that they run a rather old debian jessey
linux, only have tensorflow 1.3 in anaconda, etc. Maybe a bit behind,
but at least it is working.
They say in their free version the cpu is limited. I should run some
sort of benchmark to see if they run it at 50%.
ok, I don't see any 50% cpu capping, but ok, the load is also almost
zero. so maybe they share cpu like we [CoCalc] do and then it's going
to be slower.
Here is publishing a worksheet:
https://beta.gryd.us/notebook/published/ha_sch/py3.ipynb
Seems like they run nbconvert with a template. Also, I noticed that
changing the source file and saving it doesn't update this shared
file. I even waited a minute or two. Unpublish/publish did the trick.
ok, finally some hosting info. looked up their IP (they allow outgoing
network traffic): 54.87.201.190 which is amazon AWS.
The VM I happen to run on is a 2 core Intel(R) Xeon(R) CPU E5-2666 v3 @ 2.90GHz
They also use containers, because I only see this is running and the
notebook has PID 1.
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
gryduser 1 0.1 2.1 386680 82964 ? Ssl 10:34 0:03
/opt/conda/bin/python /opt/conda/bin/jupyter-notebook --NotebookApp
gryduser 38 1.3 1.5 673472 58264 ? Ssl 10:38 0:34
/opt/conda/bin/python3 -m ipykernel_launcher -f /home/gryduser/.loc
gryduser 66 0.0 0.0 4336 768 pts/0 Ss+ 11:21 0:00
/bin/sh -c ps aux w
gryduser 67 0.0 0.0 19100 2428 pts/0 R+ 11:21 0:00 ps aux w
"
-- William (
http://wstein.org)