search

CliMA / ClimaLand.jl

Clima's Land Model
Apache License 2.0
32 stars 8 forks source link

add RichardsModel to benchmarks #648

Closed juliasloan25 closed 2 weeks ago

juliasloan25 commented 3 weeks ago

Purpose

closes #647

Questions

The resolution here is nelements = (101, 15) (copied from richards_runoff), but the other benchmark (bucket) uses nelements = (100, 10). Do we prefer to be consistent with richards_runoff, or with the bucket benchmark?

Sbozzolo commented 2 weeks ago
  • The resolution here is nelements = (101, 15) (copied from richards_runoff), but the other benchmark (bucket) uses nelements = (100, 10). Do we prefer to be consistent with richards_runoff, or with the bucket benchmark?

It is not important to be consistent with the bucket benchmark. You can increase the bucket benchmark to (101, 15) if you prefer.

Sbozzolo commented 2 weeks ago

How often is data read from file in the benchmark?

kmdeck commented 2 weeks ago

How often is data read from file in the benchmark?

I think it is every "updateat" - every dt in this case. that is how often we do the driver callback update which calls the time varying input update

kmdeck commented 2 weeks ago

How often is data read from file in the benchmark?

I think it is every "updateat" - every dt in this case. that is how often we do the driver callback update which calls the time varying input update

(we could get away with less frequent)

Sbozzolo commented 2 weeks ago

Having very frequent is perfect because it allows us to stress-test IO.