Computing Speed of parallel computing using Dask on MyBinder

Now I’m using MyBinderand testing the computing speed of the hydrologic model.
I’m using Dask for parallel computing and applying different parameter values for parallel computing of the hydrologic model on MyBinder.
But I have a question.
When I checked the number of CPUs using “!nproc --all”, MyBinder showed me 8 CPUs.
But When I checked the Dask Worker interface, that showed me 10 CPUs. Also, parallel computing was working, because multiple outputs of the hydrologic model started to show up simultaneously. But it took a long time to finish each simulation.
The problem is even though MyBinder has many CPUs, the computing Speed of MyBinder is slower than my personal computer which has one CPU with Virtual Box (Ubunbu 20.04 LTS) on Window 10.

Can anybody explain the reason?

Checking the CPUs in a mybinder session actually tells you how many CPUs the Kubernetes node has, not how many CPUs are available to the container. We assign each user pod 1 CPU and 0.5-1GB of memory.

3 Likes

is a thread from a while back that has some discussion and code snippets for finding out what CPU (and RAM) limits are in force when you are running on mybinder.org.

Thank you very much for your reply. It is very helpful.