This is the mail archive of the glibc-bugs@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug math/14412] Removal of sysdeps/x86_64/fpu/s_sincos.S causes regressions


http://sourceware.org/bugzilla/show_bug.cgi?id=14412

--- Comment #38 from Ondrej Bilka <neleai at seznam dot cz> 2013-04-29 14:04:37 UTC ---
On Mon, Apr 29, 2013 at 01:28:10PM +0000, bugdal at aerifal dot cx wrote:
> http://sourceware.org/bugzilla/show_bug.cgi?id=14412
> 
> --- Comment #33 from Rich Felker <bugdal at aerifal dot cx> 2013-04-29 13:28:10 UTC ---
> Ondrej, the results are not pure garbage. There are only two correct results
> sin(0x1p1000) can give, either of the nearest representable neighbors (less
> than 1ulp error) of the number 2**1000. Your fallacy is writing PI in your
> example. There is no such floating point number as PI, and this is part of why
> implementing correct trig functions is nontrivial; implementing ones that work
> in degree units, or a base-2 division of the unit circle, would be much easier.
>
Ok, assume you wrote function cos_deg that takes input degree in
degrees. You still have
cosdeg(360*(2**60)) == cosdeg(360*(2**60)+180)
In short when you have inputs with zero significant digits you can
expect only garbage as output.

-- 
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]