[PATCH 09/12] benchtests: Add erff function to bench-math

Paul Zimmermann Paul.Zimmermann@inria.fr
Wed Nov 3 17:07:49 GMT 2021


       Dear Sunil,

> +# Random inputs in [0,b=0x5.ebed23725989cp+0]
> +# where b is the smallest number such that erf(b) rounds to 1 (to nearest)

this comment does not seem to be correct. I find b=0x1.f5a888p1 for the
largest binary32 number such that erf(b) does not round to 1 (to nearest)

> +0x1.351946p+2

this value (like plenty others) yields erf(x) = 1, which is not very
interesting.

Paul




More information about the Libc-alpha mailing list