On Tue, Feb 9, 2016 at 7:06 AM, Daπid <davidmen...@gmail.com> wrote: > On 8 February 2016 at 18:36, Nathaniel Smith <n...@pobox.com> wrote: >> I would be highly suspicious that this speed comes at the expense of >> accuracy... My impression is that there's a lot of room to make >> speed/accuracy tradeoffs in these functions, and modern glibc's libm has >> seen a fair amount of scrutiny by people who have access to the same code >> that openlibm is based off of. But then again, maybe not :-). > > > I did some digging, and I found this: > > http://julia-programming-language.2336112.n4.nabble.com/Is-the-accuracy-of-Julia-s-elementary-functions-exp-sin-known-td32736.html > > In short: according to their devs, most openlibm functions are > accurate to less than 1ulp, while GNU libm is rounded to closest > float.
So GNU libm has max error <= 0.5 ULP, openlibm has <= 1 ULP, and OSX is (almost always) somewhere in-between. So, is <= 1 ULP good enough? Cheers, Matthew _______________________________________________ NumPy-Discussion mailing list NumPy-Discussion@scipy.org https://mail.scipy.org/mailman/listinfo/numpy-discussion