You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Raised by Louise Bowler/Jost Migenda (@JostMigenda) (KCL RSE), that the PredatorPrey code/exercise from the profiling section could be used to demonstrate many of the patterns in practice.
Currently there are no interactive exercises within the optimisation part of the course (this was highlighted by one of their attendees!).
General plan discussed would be to de-optimise the initial exercise's code, then create exercises that allowed attendees to apply identified optimisations and see the impact.
e.g. they already identified it would be possible to optimise the grass eaten function with vectorisation.
Due to concerns over timing (and course becoming over a day's worth of content), the exercises wouldn't ask attendees to profile the code themselves (although this could be left as optional homework), instead providing them with the relevant snippet of code and potentially profiler output.
The predator prey code could also be expended to emit a built in runtime after execution for convenience.
TODO
Identify several appropriate optimisations
a)....
b)....
c)....
Draft exercises for these optimisations
a)....
b)....
c)....
Review impact on profiling exercises
Does it make it significantly more expensive to profile on bad uni pcs?
Do the results of profiling change, what is being taught/explained.
Review delivery
e.g. what's the implication on timing/flow?
The text was updated successfully, but these errors were encountered:
Raised by Louise Bowler/Jost Migenda (@JostMigenda) (KCL RSE), that the PredatorPrey code/exercise from the profiling section could be used to demonstrate many of the patterns in practice.
Currently there are no interactive exercises within the optimisation part of the course (this was highlighted by one of their attendees!).
General plan discussed would be to de-optimise the initial exercise's code, then create exercises that allowed attendees to apply identified optimisations and see the impact.
e.g. they already identified it would be possible to optimise the grass eaten function with vectorisation.
Due to concerns over timing (and course becoming over a day's worth of content), the exercises wouldn't ask attendees to profile the code themselves (although this could be left as optional homework), instead providing them with the relevant snippet of code and potentially profiler output.
The predator prey code could also be expended to emit a built in runtime after execution for convenience.
TODO
The text was updated successfully, but these errors were encountered: