Date: Thu, 30 Nov 2017 19:59:57 +0100 From: Szabolcs Nagy <nsz@...t70.net> To: musl@...ts.openwall.com Subject: Re: remquo - underlying logic * Damian McGuckin <damianm@....com.au> [2017-12-01 05:11:40 +1100]: > Is there anywhere this is described in detail. > > As part of some research, I was studying an implementation > > https://opensource.apple.com/source/Libm/Libm-315/Source/ARM/remquo.c > > which uses ideas which avoids the fixed point arithmetic. > > While there are some issues with this implementation, some of its ideas may > produce cleaner and likely shorter code than the current remquo. > maybe not so much with double but for anything else high, yes. > > Has anybody tried this before? i think it does similar things than the current implementation but uses float arithmetics instead of ints. i hadn't considered using floats, it may be faster i don't know, you have to benchmark etc. i can add this to my math todo items, but i can't promise to look into it soon unless there is a good reason to.
Powered by blists - more mailing lists
Powered by Openwall GNU/*/Linux - Powered by OpenVZ