This is the mail archive of the libc-alpha@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]

Re: IEEE128 binary float to decimal float conversion routines


On Mon, 16 Nov 2015, Paul E. Murphy wrote:

> Hi Joseph,
> 
> I think there may have been a question as to where the
> conversion routines between IEEE 128 binary float and
> decimal float should live.
> 
> Observing existing precedent and machinery, I think
> the appropriate place to house them is within libdfp.

I think existing practice is that there's a copy of such functions in 
libgcc and another copy in libdfp - and maybe the libdfp version supports 
exceptions and rounding modes (software or hardware) but the libgcc 
version doesn't?

The BID conversions between binary and decimal float involve several MB of 
tables (whereas libgcc DPD conversions go via strings).  Several MB of 
tables are not of course needed; if doing correctly-rounded conversions 
(required for IEEE 754 conformance) there's a speed/space trade-off in how 
much precomputed data you use versus how much computation you do at 
runtime, and it's up to you what you think the right trade-off for powerpc 
is.

-- 
Joseph S. Myers
joseph@codesourcery.com


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