Investigate the memory usage of model creation

[abelsiqueira]

What and Why

For the EU case it currently uses 15 GB of memory. It can probably be improved. This is probably easier after #452.

Possible Drawbacks

No response

Related Issues

No response

[abelsiqueira]

After https://github.com/MilesCranmer/AirspeedVelocity.jl/pull/32, we should be able to see the memory benchmark automatically - I hope.

[datejada]

This becomes more relevant with the changes in #583 since the EU case study update uses 86GB of memory

[clizbe]

Hopefully this now works automatically. @abelsiqueira Is this something we should look at before and after the refactor?

[abelsiqueira]

Hopefully this now works automatically. What do you mean?

@abelsiqueira Is this something we should look at before and after the refactor? Together and later. The refactor will hopefully improve many of these things, but the refactor alone will probably not be enough, so this comes after.

[clizbe]

I thought it might be automatic because https://github.com/MilesCranmer/AirspeedVelocity.jl/pull/32, is now closed.

Kay we'll look at it later.

[abelsiqueira]

Ah ok, I got what you mean. The benchmark report now includes the memory use automatically. The investigation of memory usage (where to improve) will come in a second stage. It involves the profiling.