[geeks] Best Vista story I've seen

[Charles Shannon Hendrix]

Tue, 20 Feb 2007 @ 15:00 -0500, Francois Dion said:

> On 2/20/07, Charles Shannon Hendrix <shannon at widomaker.com> wrote:
> > Miles 2D (the primary wave mixing 3D API) is nice if that's all your hardware
> > will support, but it is nothing like full EAX support.
> 
> I was pointing out in fact how GUS demos would kick the behind of
> software mixing demos, hence *everybody* added direct GUS hardware
> support to their demo. I was in the demo scene eons ago, and having a
> GUS was just like having a second CPU, since a full tracker on a 486

Ah, OK.  

I used to have a Gravis Ultrasound, maxxed out with 2MB of RAM.  

> would take half the CPU. In comparison to PC speed now, hardware
> acceleration of sound might buy you 10-15% sustainable "free" cpu, not
> 75%+, so

It's more than just how much CPU power it represents. I would agree with
you if you meant raw power, but that's not the whole story.

Doing realtime audio on your CPU wastes a lot of cycles because of
required idle periods to meet deadlines exactly. Sound is highly
sensitive to deadline issues and bus latency.

Also, there are a number of effects supported by Audigy and X-FI that
you are very hard on a CPU, and some that are not possible, at least not
while doing any other processing.

For example, things like occlusion eat a lot of power on a CPU, and a
sound processor does a better job of it. In fact, occlusion effects
don't get really good until you use something like the 400MHz X-Fi sound
processor.

You lose a lot of effects without hardware accleration, because doing it
on the CPU would eat so much CPU, you'd have little left for the game
engine and graphics.

Of course, with quad-core CPUs coming out, we might see a software sound
engine start doing more effects. That would be a good way to use extra
cores I guess.

Man... supercomputing power is now being used for mindless
entertainment.

Can you imagine the look someone would have given you if you asked them,
in 1986, if you could use their Cray to play Doom?

> Now they want a bunch of mp3 so they can have more variety. Of course,
> the actual score is not altered in real time based on the action, at
> least in my experience playing recent games (The Sims, AD1701 etc). At
> best, they change tracks to alter the mood.

Yes, but I think that is mostly a quality issue. Generated music can't
compete with a professional studio or orchestral group.

> > Trivial, but subject to latency penalties and game stuttering when
> > your system is very busy, and it can't offer many features.
> 
> Not arguing that. Just pointing out that software companies dont
> really care about this stuff. Just give them an interface and that's
> it. 

I think they care quite a bit, and that's certainly what they say all
the time, and what they tell people in their support forums.

It's also a huge part of the budget for any game in the last five years,
with few exceptions where 3D sound isn't as important.

If they didn't care then it seems they'd have avoided EAX and just wrote
all their software to use D3D or Miles.

> If they cared about technology and extracting the max speed out of the
> hardware, BeOS for example would have been a very popular platform and
> Windows would never have gotten any games... :)

Nah...

BeOS was great in theory, but in practice it was buggy and incomplete.

I think that is why it didn't take off.


-- 
shannon           | All of us get lost in the darkness, dreamers turn to look 
                  | at the stars.
                  |         -- Rush