This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH][AArch64] update libm-test-ulps
- From: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>
- To: libc-alpha at sourceware dot org
- Date: Tue, 07 Apr 2015 08:50:39 -0300
- Subject: Re: [PATCH][AArch64] update libm-test-ulps
- Authentication-results: sourceware.org; auth=none
- References: <5523BAB1 dot 5020700 at arm dot com>
Hi,
This is ok, small ulps for some function is expected, you need to
watch out only if the precision cross a higher threshold (like the
some for IBM long double) or if your are seeing precision issues
in function that are *not* suppose show any.
On 07-04-2015 08:08, Szabolcs Nagy wrote:
> Regenerated libm-test-ulps for AArch64 using 'make regen-ulps'.
>
> There are some minor differences compared to x86_64
> float, ifloat, double and idouble ulps, is that OK?
>
> I also have
>
> Function: "pow_tonearest":
> float: 1
> ifloat: 1
> ildouble: 1
> ldouble: 1
>
> but no other function has _tonearest, is that OK?
>
>
>