computational time

edit

WIth sagemath cloud, the performance depends on the load that depends on how many users use it at the same time (unless you pay for dedicated ressource). On your own computer, you can shut down all other ressource-consuming programs, and specify the architecture (processor, ram,...) on the paper so that readers can extrapolate and check on their own architecture. Hence i would recommend to prefer these results.

Howver, beyond that naive argument, note that benchmarking has tons of pitfalls, and requires some methodology especially if you want to compare the implementations of two different algorithms (sorry that i can not help here, i am not a specialist, but specialists do exist an produce teaching material).

While i am at it, let me suggest to consider the citationmodule which allows to also cite the libraries that Sage used to provide your answer: http://doc.sagemath.org/html/en/refer...

If you use the timeit command to profile your code, adjusting the command options (e.g., preparse, number, repeat, precision) adequately, you should obtain more or less the same timing every time you measure your code execution time, either in sagemath cloud (cocalc) or running it through the terminal.