SciMLBenchmarks.jl
SciMLBenchmarks.jl copied to clipboard
SciML
Bump Bio for Catalyst v14 and MTK v9
@TorkelE @isaacsas this is the one to watch for the major FBDF changes
I’d imagine many of these need updating for Catalyst 14 (eg ReactionNetworkImorters generated systems need to be marked complete).
@isaacsas can you help with what needs to be changed with this bump?
@ChrisRackauckas LMK if this doesn't work. I think all that was needed was to complete the converted ODESystem, and get the u0 and p from the oprob now (since varmap_to_vars no longer gives the right format for p).
On a sidenote, it would be useful if the parameter to MTKParameter conversion functionality could be factored into an exported user-accessible function so users can manually process parameter maps to the new form. (Essentially a varmap_to_vars replacement.)
On a sidenote, it would be useful if the parameter to MTKParameter conversion functionality could be factored into an exported user-accessible function so users can manually process parameter maps to the new form. (Essentially a varmap_to_vars replacement.)
Interesting point. Open an issue.
How long is this benchmark? It seems to error https://buildkite.com/julialang/scimlbenchmarks-dot-jl/builds/2705#019146bf-645a-4b43-b1db-597856afaf3b/403-614 but is holding up the CI for more than a day :sweat:
Maybe @TorkelE knows how long it should take. It is a long one but I have no idea if it should be days.
Has the BCR sparse Jacobian compiliation gotten worse? 80 minutes now seems pretty bad and worse than I remember.
Also @ChrisRackauckas seems like there are Symbolics related errors in the WP plotting now?
I don't think it's symbolic, but yes I need to investigate what happened there.
@ChrisRackauckas is this good to merge now? @vyudu has another PR to add a new benchmark but we should wait on that one till the updated Project.toml and Manifest.toml are merged.
Okay I'm going to merge and just fix Bikhori separately since that requires a lot less compute.