Do profiling for mode_appt_constraints = 2 too

[marghe-molaro]

Perform profiling in #286 for mode_appt_constraints = 2 too

[willGraham01]

Hi @marghe-molaro - for this issue do you mean to produce SnakeViz plots like those in #286 for mode=2 automatically, or just as a one-off? If a one-off, I can set a run off this week and post the output here.

At present, the profiling workflow that runs weekly is running in mode 2 (http://github-pages.ucl.ac.uk/TLOmodel-profiling/profiling.html) and it outputs information about time spent in each function like this. This organises time spent in each function by the call-stack, rather than pooling the time in each function like in the plots of #286.

If you think it would be useful to produce such plots as part of the automated workflow (that gets published to the profiling page), just say so and I can look into it.

[marghe-molaro]

Hi @willGraham01, ideally as we move to mode=2 being used as frequently as mode=1 we'd have the entire profiling framework applied to mode=1 being applied to mode=2 also, and just as frequently. The issue though is one of resources being limited - both to do the profiling runs and to analyse/optimise the code based on those outputs - so I think how many of the available resources should be allocated to the mode=1 vs =2 analysis is one for @tbhallett to answer.

[tbhallett]

@tamuri and @marghe-molaro this is happening now on the dev server, right?