No clue why clippy throws a fit, this PR does not touch those failing files at all.

There's new clippy complaints every time a new clippy version comes out. Don't worry about them.

Do you know the performance implications of this change? I'm not sure performance was great to start with, but I don't want to unintentionally add guarantees that will make it unnecessary slow forever.

There's new clippy complaints every time a new clippy version comes out. Don't worry about them.

Do you know the performance implications of this change? I'm not sure performance was great to start with, but I don't want to unintentionally add guarantees that will make it unnecessary slow forever.

https://godbolt.org/z/7318KPd4f Not great, but im not sure what else to do

It should be possible to replace clamp with custom comparison that handles NaN at zero cost:

if !(x < 1.0) { 1.0 } handles both NaN and >1.0.

It should be possible to replace clamp with custom comparison that handles NaN at zero cost:

if !(x < 1.0) { 1.0 } handles both NaN and >1.0.

Something like this appears to produce better codegen indeed https://godbolt.org/z/qnbjWhr3c

There's new clippy complaints every time a new clippy version comes out. Don't worry about them.

Do you know the performance implications of this change? I'm not sure performance was great to start with, but I don't want to unintentionally add guarantees that will make it unnecessary slow forever.

I have a few ideas for speeding things up, too much for this PR though. Ill experiment and open a PR when i have something that might make sense.