Hacker News new | ask | show | jobs
by wildzzz 60 days ago
Dreadfully slow for integer math but probably some similar performance to something like a CORDIC for specific operations. If you can build an FPU that does exp() and ln() really fast, it's simple binary tree traversal to find the solution.
1 comments
AlotOfReading 60 days ago
You already have an FPU that approximates exp() and ln() really fast, because float<->integer conversions approximate the power 2 functions respectively. Doing it accurately runs face-first into the tablemaker's dilemma, but you could do this with just 2 conversions, 2 FMAs (for power adjustments), and a subtraction per. A lot of cases would be even faster. Whether that's worth it will be situational.
link
xpe 59 days ago
What are some of your favorite sources to dig into this field as a whole?
link
AlotOfReading 59 days ago
Anything discussing fast inverse sqrt will go over the logarithmic side [0], as it's the key insight behind that code. Exp is just the other direction. It's not widely documented in text otherwise, to my knowledge.

[0] https://github.com/francisrstokes/githublog/blob/main/2024/5...

link
wildzzz 59 days ago
Right, I meant to say fast and accurate. You could run into quite a lot of floating point error using this for everything.
link