> 
> > From: Tom Trebisky <tom@kofa.as.arizona.edu>
> > Date: Thu, 15 May 1997 10:53:21 -0700 (MST)
> > X-Mailer: ELM [version 2.4 PL25]
> > Content-Type: text
> > Content-Length: 449
> > 
> > > 
> > > Why bother with the FPA?  Lotta pain for not much gain...
> > 
> > I hate these sorts of answers.
> 
> Yeah, I know...  Sorry to discourage you.

No no, I expect at least one "why don't you stop messing with that
old thing and get a ______" type of answer every day.  :-)

> >       Of course you could say the
> > exact thing about sun3 machines overall. Just buy a pentium
> > system and pitch the boatanchor into the dumpster.
> 
> Except that the machine works just fine without using the FPA,
> so you don't have to pitch it into the dumpster.

Absolutely, I love them.

> The FPA was interesting when the mc68881 was an expensive option,
> but these days, FPA support means building special executables to
> make use of the FPA.  It's such a hassle, I wouldn't bother.
> (Just buy an mc68881 if the machine needs one! 8^)

Well, all my  machines have 68881 (except maybe the 3/50 xterminal
I am typing on now).  But the FPA still has utility over the 68881,
and I believe you may be mistaken about their role in the sun3
scheme of things.

Based on benchmarking info in the sun floating point programmers guide,
the FPA runs roughly 5 times faster than the 68881.  The 68881 cranks
roughly 0.2 Mflops and the FPA will do around 1.0 Mflops, so I think
it is worth messing with.  Of course my pentium or even my 5x86 will
do 4-5 Mflops without any fussing around, perhaps more.

> > But to answer your question ... because it is there!
> 
> Well, have fun!

I will -- I enjoy things like this because of the "hack factor".
I.e. the enjoyment of learning how something works and making it
do something useful.  The FPA has a high hack factor.

	Tom

-- 
	Tom Trebisky			Steward Observatory
	ttrebisky@as.arizona.edu	University of Arizona
	http://kofa.as.arizona.edu/	Tucson, Arizona 85721
					(520) 621-5135