rec.autos.simulators

Trying desperately to set up GP2 to run properly on my 486DX2/66, i found
the following:
Using a stop watch to compare the actual time to complete a lap versus
the time GP2 claims, having occupancy>100% leads to slow-motion, i.e. it
takes longer to complete a lap than GP2 claims. This is acceptable, as
the computer obviously is in trouble.
But laps with 50%<occupancy<100% resulted in GP2 reporting up to 5% too
long times compared to real time, i.e. GP2 run 5% too fast!
This is disappointing, especially as GP1 was right on time to better than
1%.

I would like to know if faster machines have the same timing problem.

Regards,

Anton Norup Sorensen

http://www.racesimcentral.net/~norup/

Now this is new. I hadn't noticed it myself. I regularly run at about
70% to avoid the slow-motion time effect. Does this mean that I am
making things harder for myself?

GP1 had its own share of the slow-motion effect. The ST version didn't
display the processor occupancy, but "real" lap times were 10% or so
longer than reported by the game.

 I'll have to get the stopwatch out & see!

Richard

BTW, I turned off all texture to make sure my PO's stayed low, but at 17 fps, I
still hit 100%, and a bit above, quite often in the practice laps.  I'm
dissapointed with that kind of performance.  Machine:P133, 16MB, Stealth64 2MB
DRAM.  Maybe too much texture in those trees.

I can confirm this. It is some sort of a let down. But at least
everyone is treated equally(I hope...).

--
--- Terje Wold Johansen

--- http://www.ifi.uio.no/~terjjo/
--- "I am your inferior superior." O.W.

: What GP2 does is it will slow down time within the game when the
: occupancy gets near 100% to allow you to have more control instead of
: having a slide show.  That is why real time is actually longer than in
: the game.

I think the poster realized this.  His real concern is why does the game
speed up time when the CPU is less than 100%?  This compresses our
reaction time and increases the "simulated" speed at which we are
traveling.  Can't make it any easier to simulate F1 driving.  This way
the poor F1 pilot is on some weird kind of medication.  One moment time
is slowed 2-4x reality, the next it has increased 5%.  I certainly
wouldn't want to race against a driver in this state.

Since the game (whoops, I mean "sim") deviates wildly in it's CPU
Occupancy, it will deviate wildly in the fps it delivers.  Therefore it
wildly compresses/expands time.  This is a poorly coded approach to
simulate reality.  I'm sorry, but I see this as a bad call in the "art"
of coding a simulation.  This could explain why I can't seem to get the
flow down during an actual race.  The program is taxing the CPU even
more during a race (as opposed to "practice at any circuit").  This
could even explain the slower speeds during a race vs. practice and
qualifying.  

Will this stop me from going back and trying to find the perfect setup
at Monza?  Not yet, but it may start to wear thin on me after awhile.

--
**************************** Michael E. Carver *************************
     Upside out, or inside down...False alarm the only game in town.

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=<[ /./.  [-  < ]>=-=-=-=-=-=-=-=-=-=-=-=-=-=

: I thought that you only observe this affect of time slowing down
: beyond 100% occupancy and that otherwise the frame rate (passage of
: time remains constant)?  Perhaps, I am wrong.

Well, if I understood the original post, the game compresses time when
the occupancy drops below 100%.  With people setting up their fps to be
2-3 less than the estimated rate, this can really cause wild swings in
time compression/decompression.

I consider slowing down time as a major driving cheat/aid.  I've also
noticed that this can give the human driver a great advantage over the
CC cars.  I haven't fully mastered the art of a standing start (with
out any traction aids).  So even if sit on the front row, most of the
time the CC fly right on by.  However, I am able to out-brake to a huge
degree the CC cars coming into the 1st chicane at Monza.  I can then
drive 2x as fast as they can through the chicane.  Even though there is
only one or two CC's going through the chicane, they are still locked in
slow-mo mode due to the gaggle of cars awaiting their chance.  Since I
am alone on the straight leading to Curva Grande, I make unbelievable
time on the CC cars.  Leaving them struggling to get out of slow-mo.
This is an unreal advantage.  The same goes in reverse.  If I'm mired in
slow-mo time going thru the chicane, any CC cars that got through
"cleanly" are making unbelievable advances.  This just doesn't sim well.

On the other hand, I've experienced difficulty dealing with the same
situations in Papyrus's sims also.  Here the problem isn't with time
slowing down or speeding up, but frame rate vs controller input.  The
car tends to over-react to input.  But, I know it's going to happen and
prepare for it by pretending I am driving on ice through that first turn
(especially on ovals).  With Papyrus's approach, though frustrating, it
can be compensated for and the overall simulation still holds up well to
reality.

Now if GP2 really does compress time with CPU <100%, this is a major
drawback.  Getting better and faster hardware will only enhance this
"feature".  Especially since everyone is striving for low CPU occupancy.
Whereas, with Papyrus's approach the situation only improves when the
hardware gets better/faster.

--
**************************** Michael E. Carver *************************
     Upside out, or inside down...False alarm the only game in town.

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=<[ /./.  [-  < ]>=-=-=-=-=-=-=-=-=-=-=-=-=-=

I've noticed that when I follow a couple of cars heading into a corner
that I get away with braking later than I would have if I was heading
in by myself.

Well, I think the approach the developers chose might have merit.  But
not for a racing sim.  It would have worked better for an air combat
flight sim.  Why?  Because in such games you often respond to what
other planes do and there is a feedback loop.  However, from my
limited experience with racings sims, it seems that the fastest way to
run any track is to anticipate control inputs (steering, gas, and
brake) as opposed to respond to visual inputs.  Thus, feedback loops
are less important than timing and rythm.

Why cant we have a word on this from MICROPROSE!! Gabriel if you cant
answer these type of technical questions please tell someone from
Microprose to monitor this newsgroup!!!!!!!!!!!!!

: >I consider slowing down time as a major driving cheat/aid.  I've also
: >noticed that this can give the human driver a great advantage over the
: >CC cars.  I haven't fully mastered the art of a standing start (with
: >out any traction aids).  So even if sit on the front row, most of the
: >time the CC fly right on by.  However, I am able to out-brake to a huge
: >degree the CC cars coming into the 1st chicane at Monza.  I can then
: >drive 2x as fast as they can through the chicane.  Even though there is
: >only one or two CC's going through the chicane, they are still locked in
: >slow-mo mode due to the gaggle of cars awaiting their chance.  Since I
: >am alone on the straight leading to Curva Grande, I make unbelievable
: >time on the CC cars.  Leaving them struggling to get out of slow-mo.
: >This is an unreal advantage.  The same goes in reverse.  If I'm mired in
: >slow-mo time going thru the chicane, any CC cars that got through
: >"cleanly" are making unbelievable advances.  This just doesn't sim well.
micheal,

there's one problem with this... elongation of time (ie slow mo) is only
dependant on the player's PO.  So, if there are 24 cars going into a
corner yet you are alone on the straight, the 24 cars are NOT left
struggling to get out of slow-mo while you advance.  They are simply
slowing down to avoid collisions.  This is attributable to crappy AI, not
slow-mo.  Also, it is important to note that PO induced time-warping
(sounds like something from Area 51) is simulation-wide.  If you are
driving through a corner and your PO jumps to 150%, a computer car on the
straight before the corner suffers slow-mo just as you are.  So altho
when you are out of the corner the computer car takes the corner at a
compressed/normal time-scale, he suffered a proportional loss on the
straight, and gains nothing on you.  Especially since CC's essentially
have no reaction time.

: >Now if GP2 really does compress time with CPU <100%, this is a major
: >drawback.  Getting better and faster hardware will only enhance this
: >"feature".  Especially since everyone is striving for low CPU occupancy.
: >Whereas, with Papyrus's approach the situation only improves when the
: >hardware gets better/faster.

agreed.
this desperately needs to be fixed in GP2, it is HORRIBLE, obselete way
to code games, (it was even horrible for GP1).  however i'm willing to
bet quite a bit that it will *NEVER* be fixed.  Think about it, GP1 had
this, people saw the problems.  How many years did Geoff have?  is it
fixed? no.  is it going to be fixed?  i doubt it.  pitty such a great sim
is allowed to have such a fatal flaw simply because the head coder is
hard-headed "must do it my way" type of guy.

NOTE to MICROPROSE: enough geoff worship.  tell him to fix it, or let one
of your other coders fix it.  crammond is no god, he's not even a carmack.

i love this game, but it pisses me off.

-------
Carlos Ribas
MalSoft
http://www.nol.net/~draconis

I still haven't gotten round to measuring this on my machine, but
whilst I quickly noticed this at p/o's of > 100%, I haven't noticed it
at all at < 100%. If it does have an affect below 100% it must be
pretty small.

Well, just set the detail / frame rate to be < 100%. That way you are
not cheating yourself.

That still leaves the issue of hot-laps, but to be honest, I tried
having a p/o above 100% and it was hopeless. How anyone can set a
quick time driving through that temp***soup, I haven't a clue.

Get the revs to just below the second green light coming on & be
_gentle_ applying extra power. I can usually keep with or even outdrag
the CCs now (I was _hopeless_ when I tried using the clutch option
<g>).

I always crash into them <G>

Slow-mo time does come into it if you are in the middle of lots of
cars & maybe some brake smoke, but it affects the whole game - not
just _your_ car! The slow-mo is due to the number of cars being drawn.
If you are away from them, there shouldn't be any slow-mo. They will
still drive through that chicane really slowly, but that's just the AI
(try it in VGA with few graphics - the cars will still bunch & crawl
through the chicane, but very smoothly!).

The AI in ICR2 always slows down _way_ too much at the start,
especially for the first corner on road races, even more than in GP2.
The main difference seems to be that the ICR2 cars then manage to
actually get through the corner a lot quicker..

I still think the GP2 "compress" is either a mis-report or over
stated. As I said before though - I still haven't measured it so its
just my opinion.

Richard

: I still think the GP2 "compress" is either a mis-report or over
: stated. As I said before though - I still haven't measured it so its
: just my opinion.
: Richard

Sorry to repeat myself:
The maximum time compression effect I have measured is 5% with
50% < occupancy < 75%. It's little, but significant.
The measurement was made on a DX2/66, and it is possible that the
compression effect is related to running on a slower machine than intended.
Please, somebody with proper hardware, go settle this question - it's
really not that difficult to measure to an accuracy of better than 1%.

Have fun,
Anton Norup Sorensen

How exactly do you make the measurements?  Thanks.

Sorry for doubting you earlier! I have since made some measurements of
my own and recorded between a 3.2% and 3.9% compression. This was on a
P133 with processor occupancy as low as 25% for one test!

My theory is that the time compression is designed to allow for the
fact that if you run the game at the estimated fps you will have some
bits at > 100% PO which will slow the game down. By being slightly
faster when < 100% it might even things out.

Doesn't help for consistent driving though :(

Cheers,
Richard