[rescue] Survey: what do you want in _your_ SPARC?

[Joshua D Boyd]

On Fri, Jul 19, 2002 at 12:47:23PM -0400, Katherine Strojny wrote:
> > SS5s usually had 16bit sound to my
> > understanding, but I suppose things could be different here.
> 
> Yes, I'm learning that different is often the case, with this system.
> Frinstance the memory bank layout is unique, and I recall reading somewhere
> that SS5s don't require termination for internal drives but this Axil
> definitely does.  (Though reading JW Birdsall's FAQ, sounds like it's not
> that straightforward, and maybe it makes sense that mine requires termination
> b/c the cable is unfortunately longish.)

External drives?
 
> I have doubts about the Axil's performance meeting my other requirements,
> too, so I'm back to looking at U30s.  I'm willing to take a shot at compiling
> audacity on the Axil, depending on how fast I move on a U30.

Well, give it a shot.  After poor performance is still working, and it
should at least be able to do work, render it to disk, then playback
the work.  But ultimately, there is no excuse for not being able to do
basic things in real time.  Audio takes so little bandwidth that you
have most likely (assuming 85mhz or faster) over two thousand CPU
cylces to use for processing each sample of output.  To not be able to
do real work at that speed is simply pathetic  Mixing 4 tracks
together only requires about 10 instructions[1], and that's assuming no
particularly clever optimizations.  I'm not sure at the moment what it
would take to do realtime reverb, delay, or distortion, but you didn't
say anything about expecting that.

-- 
Joshua D. Boyd

[1] But remeber that not all instructions are guaranteed to execute in
only one cycle.