Hi guys, I thought it would be worth going through each of the settings
options offered by RT and see what gave the best frame rates with the best
quality graphics (at least for one system). While not all of you can
directly compare, it should give some in the sim community a decent starting
point. It also raises some interesting issues with regard to the graphics
engine RT uses.
Here's the setup information for anyone with a similar system that wants to
run this and see how close it comes to these results.
System Specs:
CPU: 1.4 gig Athlon (not oc'd)
Mobo: MSI K7 Master (vers 1.0 BIOS) FSB=133, CM=10.5, AGP=4x
RAM: 512 MB, PC2100
Video: Asus V8200 Deluxe (GF3) - not oc'd
Video Drivers: 21.85 NV Ref (on Riva Turner)
Sound: SBL 5.1 Platinum
OpsSys: Win98SE (current updates, with ACPI turned off)
DirX: 8.1
Monitor: 21" NEC FE2100 (1600x1200x32 desktop @ 85hz refresh)
Controls: BRD F1 Sim Pro, with clutch (analogue gameport: 4 axis, 6 button)
Note: BIOS and video card settings for D3D are as per the "Video Tweak
Guide" on Powersims website.
**RT Video Settings:
I just sort of guessed at what would be the best settings to start with,
knowing my system and how it runs N4 and GPL. This wasn't so important
because I was going to test all the reasonable combo's. - so just needed to
pick something as a "mark" point. This is what I used to start (sound was
set as Creative EAX @ surround):
Display: 1600x1200x32
Triple Buffer: checked
Texture Quality: High (32bit)
Filtering: Linear
Mipmapping: Linear
Multi-sample: None
To measure the in-game frame rate I used Fraps with the min-max log. The
"start" FPS below is initial frame rate when the stage starts. Then I
activated the Fraps min/max log at the start of the stage, made the run,
turned the min/max off and looked at the log file to report the min FPS and
max FPS recorded during the run. If I made any really bad driving mistakes
(like ending up high in some tree a couple hundred yards off the
course...lol...) I exited the run and started over. The reason is because
if you stuffed yourself into a tree somewhere (or rock or what ever), the
damage and the "consistent texture of the tree" would cause the FPS to jump
(or drop) to abnormal levels, so I didn't want to count those. Besides,
this was a good way to get several dozen practice runs in! I also made
double runs at first, one with "automatic" and one with "manual" trans,
thinking that might make a frame rate difference, but it did not (at least
not on this system), so I stopped doing that after a few runs. Any settings
I changed that had better results than in #1 below, was left in that setting
for the rest of the tests.
***Settings Parameters*** **Start** **Min**
**Max** **Graphics Quality**
1. All set as above. 45 FPS
47 FPS 57 FPS 9.2/10 (very good)
2. Triple Buffer: Off 45
47 62 9.5/10 (better)
3. Filtering: Anistropic 21
20 23 9.0/10 (not drivable - but
pretty)
4. Mipmapping: None 42 43
52 7.0/10 (too much flash)
5. Filter/Mip: Nearest 49
51 64 8.5/10 (some flashing)
Okay, so I went back to the original settings in number #1, except left the
Triple Buffer unchecked (that was the best FPS and graphic quality of the
above set. Then I tried:
6. 1600x1200x16bit 49
51 59 9.7/10 (best color)
7. 1920x1440x16bit 41
47 58 10.0/10 (awesome)
8. 1280x920x16bit 49
53 114 9.5/10 (lwr detail)
9. 1024x768x16bit 49
57 117 9.0/10 (lwr detail)
So that's it. For me, running at 1900x1440x16bit, triple buffer=off and all
the other initial settings at the top, was the best. The graphics at these
settings are breathtaking. Its a "Kodak" moment for the entire stage run.
I also could have sworn that the FPS was going to be higher than it read for
this setting, because it just looked smoother and faster. But, that could
be because of the high detail and awesome/realistic coloring of the
landscape, trees, sky, everything......that I could see further down the
track and that made it seem like the FPS was very high?
**Some Weirdness Behavior**
During this benchtesting, there were a few weird things that happened....I'm
just reporting these in case the guys at JW don't know about them yet, or
for grins.
1. The first time I selected the 1920 res, the music wouldn't play in
sync with the UI screens (it was also playing very fast, maybe 30% faster
than normal), and it didn't turn off when I went into the stage to drive. I
drove about 1/3rd of the stage (UI screen music still playing) and exited.
The music was still playing. I exited to my desktop, re-entered the game -
same thing. Then, I rebooted - had no problems with this after that.
2. The desktop icon shortcut takes forever to bring up the RT program.
Sometimes as long as 30 seconds. This was not an issue of the res.
settings, and it was consistent throughout the testing.
3. When exiting the game back to the desktop, sometimes the video setup
screen would still be up, sometimes it would pop up (like it returned) and
sometimes it wasn't there at all. After a while, I started checking in
"control-alt-delete" and found that the RT demo was **always** still
running....and this didn't matter if the video setup screen was up or not.
I had to end-task from that point on to get the game to release from memory.
I wished I had discovered this before I had the problems in #1 above,
because I bet you that what happened was I actually had two different
versions of RT in RAM at two different res settings, and that is what caused
the UI screen music to lock. I couldn't reproduce the problem, so that's
only speculation.
I had no other problems with this demo, except trying to re-learn how to
drive a 100 hp, mini-Cooper with front wheel drive as fast as possible
without dying! I eventually was able to tap some of those driving skills I
learned back in the late 1980's when I drove in a few IMSA Firehawk races
for TC Kline, in a 16V GTI (VW) - Sebring in the rain was just a thrill -
not! And, I also ran one of those in 1990 in SCCA, SSB class (for one
season). Front wheel drive, low-horsepower race cars are the most
frustrating race cars on the planet. They also take the most driver skill,
concentration and keen awareness of the car's attitude....than any race car
I've ever driven. You make any mistakes, that's it - you can not make up
for anything with horsepower because you don't have any. The guys who are
really fast in these types of cars (Randy Pobst, Peter Cunningham, Tommy
Archer and so forth) say that in order to be consistently fast in these
cars, you have to drive the car (mentally) about 30-60 feet in front of you
(mentally, you are always 30 to 50 feet further down the track than you
actually are at any moment). It took me a while before I remembered that.
This sim is awesome. It may well gain "shrine status" next to my bookshelf
copy of GPL. Hey, this is kind of the GPL of rally sims, is it not? The
only two things I thought it was lacking in was "no wheel spin" - that's
painfully missing. And, no clutch - which would be a mistake to leave out,
IMHO. In the more powerful rally cars coming in the final release, the
lack of wheelspin is going to be a major "missing" element of an otherwise,
very good physics model.
For those of you struggling with running this car fast (getting under
2:50's), here's a tip that helped me pick up a lot of time at first - while
learning how the car handles and what it wants from me as far as driver
input:
Map the handbrake to an easy-to-reach wheel button (I used the right button
at my thumb) and run for a while in "automatic" instead of "manual"
transmission. This does a couple of things for you. It gives you one less
thing to deal with (gear selection, especially when to downshift to avoid
lugging the engine) and it let's you deal with what the car wants in terms
of throttle and steering input. When entering a #3 or higher turn, keep
your foot on the throttle (or lift maybe 10% max) but right foot brake just
a tap, to help point the nose of the car into the turn. When you get the
hang of this, you can also use the handbrake instead of the brake tap, its
even more subtle and slows you down just a tad less. After I got done
destroying a dozen cars or so (and listening to my co-driver's bitching,
whining, crying and chirping - of course, I countered with suggestions that
he call the turns out a tad faster so I could actually do something about
them instead of just "going along for the ride into the woods" along with
him) - I began running completed stage times around the 2:55 range....and
used the "automatic" transmission all the way down to a 2:44. I went back
to "manual" and had a much better understanding of what the race car wanted
from me. I got down to a 2:41.22 - so far. I've probably made around 60
to 80 runs so far. Good luck.
Tom
PS: I saw somebody make a post somewhere that they did a 2:22 or something
like that? Not on this planet they didn't do that! When are the "aliens"
going to figure out that nobody cares how fast they are when they cheat?