5 ms·

I was just taking a look and couldn't help but notice the switch statement for your operator[], which likely causes a lot of unnecessary bad speculation at runt

by amadio 2y ago

I was just taking a look and couldn't help but notice the switch statement for your operator[], which likely causes a lot of unnecessary bad speculation at runtime: https://github.com/RandyGaul/cute_headers/blob/755849fc2819d4e61e7d999aa036d79706d832b0/cute_math.h#L99 https://github.com/RandyGaul/cute_headers/blob/755849fc2819d... I fixed this exact problem in a highly used library in high energy physics: https://gitlab.cern.ch/CLHEP/CLHEP/-/commit/5f20daf0cae911793d9aa5fc62987d6dc83647ab https://gitlab.cern.ch/CLHEP/CLHEP/-/commit/5f20daf0cae91179... Many believe the C++ compiler will magically optimize the switch away, but in some cases, like the example above for CLHEP, it doesn't happen, so you end up with bad performance.

amadio 2y ago
Since you left this "optimize me" comment here: https://github.com/RandyGaul/cute_headers/blob/755849fc2819d4e61e7d999aa036d79706d832b0/cute_math.h#L667 https://github.com/RandyGaul/cute_headers/blob/755849fc2819d... See an optimized quaternion multiplication implementation in SSE by me here: https://stackoverflow.com/questions/18542894/how-to-multiply-two-quaternions-with-minimal-instructions/18542951 https://stackoverflow.com/questions/18542894/how-to-multiply...