Another blog post:
Interlocked min/max on HLSL single precision floats
https://www.jeremyong.com/graphics/2023/09/05/f32-interlocked-min-max-hlsl/
@ninepoints nice! So yeah, exact same trick as what is used to use radix sort on floats :) It can also be done branchless, see http://stereopsis.com/radix.html
@aras yes that's a good one, once you have the concept of an order preserving map, things like a radix sort become pretty easy (and it helps to know what happens to nans and stuff).
@aras I added a link to this in an added section in the post, thanks! The branchless version is what I use also