[illumos-Developer] Floating-point formats supported on S390/S390x... / was: Re: Ugh... xdr_{float, double} problems

[Roland Mainz]

On Tue, Feb 1, 2011 at 12:34 AM, Garrett D'Amore <garrett at nexenta.com> wrote:
[snip]
> It sounds like all the main CPU targets support IEEE 754.  The only
> reasonable (?) question IMO is S390.  Roland?  Neale?

I'm not sure whether Neale is subscribed here... CC:'ing him and Mike
Cowlishaw as the experts on the (IBM) floating-point stuff.

AFAIK (Neale/Mike, please correct me if I am wrong) S390/S390x support
both IEEE 754 floating point and the IBM floating points format. But
I'm not sure what the Solaris S390 port supports or will support...
this depends on the software (e.g.gcc, libm, kernel) and minimum
hardware (e.g. newer IBM Z/Series machines even support the decimal
floating-point parts of IEEE 754-2008 in hardware and the S390 port
only runs on rather new hardware (and newer versions of the Hercules
emulator) but Solaris libm has no support for that (yet) and I didn't
remeber seeing any support for that in the S390 port) ...

BTW: Interesting question: Is there a way (e.g. helper function in gcc
for S390/S390x) to convert between IBM and IEEE754 floating-point
formats on the fly ?

----

Bye,
Roland

-- 
  __ .  . __
 (o.\ \/ /.o) roland.mainz at nrubsig.org
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 641 3992797
 (;O/ \/ \O;)