On Wed, Sep 5, 2012 at 12:16 AM, Roland Mainz <roland.ma...@nrubsig.org> wrote:
> Hi!
> ----
> Mostly offtopic for this list:
> TheRegister.co.uk has a small article about the ne SPARC64-X processor
> - see http://www.theregister.co.uk/2012/09/04/fujitsu_sparc64_x_processor/
> Noteable (for us) feature is the Decfloat accelerator... which is IMHO
> an indicator that Decimal floating point has become more important
> (remember the old proposal for typeset -D and typeset -lD to define
> |decimal64| and |decimal128| datatypes (agghllrr... does anyone
> remember which printf format letters are reserved for DecFloat ?))

The 'reserved' letters in printf formats for decimal floating point
are %D and %H, it will be used to represent decimal floating point
formats in the future (i.e. the decimal floating point length
modifiers for printf and scanf floating point conversion specifiers
(a, A, e, E, f, F, g, G) are DF for _Decimal32, DD for _Decimal64, and
DL for _Decimal128).

- http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1312.pdf

For those who think decimal floating point support is far away: AIX
and HP-UX already support decimal floating point in libc and one major
x86 vendor is going to add hardware support in its next generation of
CPUs. No surprises here. But seeing this in Fujitsu SPARC64 was even a
surprise for me since Fujitsu is very conservative in adding new ABI
related features. If they add it then it must have become very

ast-developers mailing list

Reply via email to