Revert "Density-based θˡⁱ state: fix compressible saturation & temperature inversion (closes #765, supersedes #752)"

[giordano]

Reverts #767.

I want to see if this resolves the slowdown in the Reactant tests observed at #773 (comment). CC @dkytezab @Pangoraw.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[giordano]

It was indeed #767, Reactant tests are back to normal times 😞

[glwagner]

It is probably worth implementing a different solver that uses a fixed number of iterations, to see if this resolves the reactant performance issue. I am not surprised that a state dependent solver like a Newton solver induces a big slowdown -- we have seen that before

[giordano]

The point is that compile time blows up, not simply runtime, XLA reports spending 25 minutes compiling in reactant/weno_compilation

[glwagner]

The point is that compile time blows up, not simply runtime, XLA reports spending 25 minutes compiling in reactant/weno_compilation

Hmm, compile problems make sense, though I’m not sure what it has to do with WENO

[giordano]

though I’m not sure what it has to do with WENO

The ways of XLA are infinite.