A new floating point format

Can express numbers from ~ 2^(2^(10^-308)) to 2^(2^(10^308))
https://www.reddit.com/r/frozenvoid/wiki/algorithms/data_encoding/logfloat

>>9

How easily can it be optimised and implemented in ASM? How does it compare in speed with IEEE floating points?

For Frozenvoid, optimization and performance is an afterthought. If it works, who cares!

How about its accuracy for the integer set?

see >>7.

What happened to Sugin?

Posted here for a little bit, then left back to /jp/.