Skip to content

Comment on Speeding up atan2fparent

Comments

Thanks for clarifying!

Without benchmarking, I would expect atan2f to be around 20-30 cycles per element or less with either Intel's or Apple's scalar math library, and proportionally faster for their vector libs.

Thanks for the info.

By the way, your writing on floating point arithmetic is very informative -- I even cite a message of yours on FMA in the post itself!

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.