> THIS ----------- is rediculous!!!!!!!! Why should we have to
>spend a thousand dollars every year or so to keep up with this when I
>am sure there is a way to make this work on machines slower than what
>they are saying it will.
> I am frustrated and fed up. This BS has to end somewhere. I have
>two kids and a family to support. I work for a living and hard at
>that --------- but I certainly don't appreciate the fact that these
>CHARACTERS that design this stuff keep putting it so far into
>HYPERSPACE that you almost have to have a leftover NASA computer to
>run it.
It really hurts me to read this. Are you a software engineer? How do
you know there is a way to make GPL work on slower machines? Can you
explain how?
I am a software engineer and a former race driver, who took physics in
college. In my humble opinion, Dave Kaemmer is one of the great
software engineers in the world. GPL's physics engine is a software
engineering marvel. It is modeling an incredible number of details, and
doing so incredibly efficiently.
For example, GPL models the inertia of the drivetrain components. Put
the car in nuetral, foot on the brake, and rev the engine (if you don't
have separate brake and gas pedal axes, park on a level space on The
Long Straight, after the loop). You will see the chassis rock from side
to side as you blip the throttle. This is due to the inertia of the
engine and drivetrain. Now push in the clutch (keypad Del key). When
you blip the throttle, the engine will reach readline a little quicker.
That's because the engine now does not have to accelerate the inertia of
the transmission input shaft.
This kind of attention to detail permeates the physics engine. It is
simply incredible. I believe that few software engineers or physicists
in the world possess the combination of skills necessary to make this
happen correctly.
Furthermore, I believe that even fewer possess the skills to make this
happen with the performance necessary to run on computers that are
widely available today. Dave has worked very hard to make the GPL
physics engine efficient enough to work acceptably on the machines of
today, and the multitude of posts in this newsgroup show quite clearly
that his efforts have been a resounding success.
The amount of calculations going on in GPL every second boggles the
mind. Dave completely recalculates *everything* going on in the car -
drivetrain inertia, suspension movement, tire friction, tire rotating
inertia, chassis movement and rotation in all three axis, on and on -
two hundred and eighty-eight times a second (compared with around 30 hz
for ICR2). Empirical testing showed that this was the minimum frequency
necessary to avoid oscillations. Imagine the CPU power necessary to do
this, even with extremely efficient code.
Now add in the processing power required to construct the visual world,
calculating positions of track, landscape, objects, then drawing
polygons (in the case of the Voodoo Graphics and software-only
implementations) and shooting everything over to the 3D card (on
Rendition and Voodoo Graphics machines). Other racing sims, with far
less sophisticated physics models, can devote a much greater percentage
of their processor time to dealing with graphcs. GPL must belt out all
the graphics in the small spaces left between the all-important physics
calculations.
In my humble opinion, GPL is an engineering masterpiece.
GPL is far from being another GP2, which could not run well with full
graphics on anything available at the time of its release and for years
afterwards. GPL can run very well on an entry-level system available
today for roughly $1000, with the addition of an inexpensive (under
$100) Rendition 2x00 card. Even many older machines like P-133's
(provided they have L2 cache) can be upgraded for a few bucks' worth of
memory, a Rendition 2x00 card, and an inexpensive AMD K6-200, and
they'll run GPL quite well.
I find this simply amazing.
How can anyone complain about this?? Instead of berating the folks at
Papyrus, I say we applaud them loudly.
Eagle Woman
Remove the spam blocker NOSPAM to email me.
http://www.nh.ultranet.com/~alison/