This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


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

Re: fcmove on Alpha not qualified with /su when compiling -mieee - is that correct ?


Richard Henderson wrote:

> On Mon, Jul 03, 2000 at 10:40:11PM +0200, I wrote:

> > Does this mean that, without /s, it is OS-implementation-defined
> > what happens with denormals-on-input on the Alpha ?

> At least, all existing operating systems deliver SIGFPE.  One
> could suppose an OS that didn't, though that would be silly
> since your program has now got garbage in one of its registers.

Is that because without /s all bets are off anyway w.r.t. denormals on
Alpha ?  It seems the instruction could just trap and do nothing when it
detects a denormal input operand - and with precise exceptions (like on
ev6 and beyond) this means that it still could be "patched up", no ?

> Though I'm not clear enough
> on the physical sciences to know what sort of water vapour you should
> be expecting even at -140 C.  I'd have thought exactly zero...

Ah, no - at any temperature, at least as long as water molecules are
free to leave the ice/water surface, there's room for the concept of
"saturation pressure" - it's defined as the amount of vapor that's in
balance with the ice/water surface in the sense that as many molecules
leave the ice/water surface as return to it.

The reason the buildup of the table in our software broke down is that
the formula used for that balance computed a denormal saturation vapor
pressure at around 15 K (so it *was* close to zero ....).  However:

> > the funny thing about 13 K is that there is no `air' in which vapor
> > pressure has a meaning).
> 
> Well, you've still got helium, have you not?  I didn't think
> that liquifed until about 4 K.  ;-)

Ah, yes, but at these temperatures a number of assumptions commonly used
in Numerical Weather Prediction code breaks down:

1. The above formula for satuaration vapor pressure is not valid any-
   way there (even disregarding the change in the environment of the 
   ice/water surface).

2. NWP models deal with a *gaseous* atmosphere throughout.  The concept
   of liquid oxygen or nitrogen is simply not catered for in their
   governing equations.

3. And a couple of others that I don't feel like digging up now ...

Happy Fourth of July, ye of the Left Side !

-- 
Toon Moene - mailto:toon@moene.indiv.nluug.nl - phoneto: +31 346 214290
Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
GNU Fortran 77: http://gcc.gnu.org/onlinedocs/g77_news.html
GNU Fortran 95: http://g95.sourceforge.net/ (under construction)

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