Both
of my machines suffer from the fact that the steering response is severely
non-linear - in a very bad way. As you may know, both ICR2 and GP2 have
nonlinear steering options, designed (I assume) to allow more precise on-center
control with joysticks and other controllers with limited travel. If graphed,
the nonlinearity would look something like Figure 1.
In other words, the more you turn the wheel - or joystick - the more steering input you get. Smoothly, though, and in GP2 you have fairly good control of the shape of this curve through the control settings page. ICR2 just gives you a default curve, but it's an excellent one.
The
folks at Terminal
Reality (who developed CPR for Microsoft) seem to have stumbled a bit
in their implementation of nonlinear steering for CPR. It's nonlinear, but
in a very undesirable way. A graph of CPR's steering response might look
like Figure 2.
The result is that the car is quite unstable on the straight. There is a dead zone in the middle, where moving the wheel does nothing. As you gently roll in further steering input, at a certain point in the controller deflection, a sudden, large steering response occurs. The car shoots for the wall. If you react in time, and put in an equal and opposite response, it shoots for the other wall. You find yourself proceeding down the track in a series of increasingly divergent swerves until you smack the wall very hard.
I recalibrated my Thrustmaster T2 several times, but this made no difference. I tried adjusting the controller settings for minimum and then maximum nonlinearity, but found that this only seemed to affect the width of the dead zone, not the amplitude of the steps at its edges. The car was still undrivable on the straights.
Just to make sure the problem wasn't dirt in the T2's steering pot, I quit the game, dropped down into DOS, and fired up good ol' ICR2, er, Papyrus CART Racing, to see how its response was at the same track.
Ahhhhhh!! Wonderful! Smooth, linear, predictable response. I can plant the car anywhere I want. With my favorite Jake Myers setup, I was quickly down into the 53's at Long Beach, as is usual for me in ICR2.
So maybe it's a setup issue, I says to meself. Back in CPR, I take a look at the supplied sample setup for Long Beach, which is what I've been running. Toe-in is the obvious first place to look; too much toe-out at the front, or any at all at the rear, and you'll get a car that's unstable on the straights. It has some toe-out in the front. It won't let me do any toe-in so I set it for zero toe-out. I set the rear for maximum toe-in at the rear, and head back out to the track. Hmmm. A little better.
I also notice that the car is soft as a marshmallow; any steering input is amplified as the car wallows from side to side, which in turn amplifies the dartiness. So I crank up the spring rates. Better, but still very darty on the straights. I try every imaginable combination of controller settings: minimum and maximum nonlinearity, min and max sensitivity change with speed; I even try cutting down on steering lock. I arrive at a combination that gives me something reasonably drivable on the straights. By now I am down in the 56-second range, compared with my wildly erratic best of 1:03 with the stock setup.
Eventually, with tire pressure, spring, shock, roll bar, and third spring changes, I can run fairly consistently in the 51 second range at Long Beach, but the car is still much harder to drive on the straights than in the corners. It requires intense concentration to keep off the wall in the fast, sweeping, turn 5, which should be a no-brainer, flat out, flick left-flick right. It's impossible to position the car precisely for the entry of corners, so I must leave plenty of room at the outside of the track to allow for the steering slop-and-twitch.
John Wallace suggested: "One thing you can do is try changing your control type in the setup options from "wheel" to "joystick" (although still use your wheel).
I tried this, and it did help some, but the problem was still not cured.
Many people apparently do not experience this problem, at least to the degree that some of us do. My thinking now is that there is some subroutine in the controller interface that decides, based on some criteria unknown to me, that the joystick/controller needs to have a dead zone of a certain size. It works fine under some conditions, but on certain machines, its decision-making process results in too large a dead zone, with edges that are much too steep.
Note that it's what's happening at the edges of the dead zone that is the problem. After some additional testing, I've found that ICR2 and SODA both have dead zones of similar size to what I experience in CPR. However, the transition from the dead zone in ICR2 and SODA to the beginning of response is very smooth and gentle, while in CPR, no matter what I do, there remains a very sharp, large reaction to a very small steering input at the end of the dead zone.
Note: Microsoft has said that they will soon release a patch which will allow the user to adjust the size of the dead zone in the middle of the steering. This will hopefully improve the steering in CPR. Needless to say, I am eagerly awaiting this patch!
I experienced another problem, peculiar to my primary machine. Almost every time I entered race mode, I found a frame rate of an impossibly slow rate, maybe 5 to 10 fps. Maddeningly, occasionally I would experience a much more acceptable 15 to 25 fps, for no apparent reason.
Finally, I found the reason. While the search was a nightmare, the fix was trivial. With appropriate display options deselected, I now get a consistent 15-25 fps on my P-166 with Screamin' 3D video card.
What was the problem? CPR has an arcade-game like feature (presumably to lend more excitement). If you have the CPR CD-ROM in your CD-ROM drive, CPR plays music from several audio tracks on the game CD while you are racing.
Unfortunately, this feature entails considerable overhead, reducing the frame rate drastically on some machines, including mine. Even more unfortunately, on my system, when you try to play a music CD, no sound comes out of the speakers. [And yes, the wire from the CD-ROM drive to the sound card is connected properly, and no, I didn't have the CD player muted in the volume control.] I had no way of knowing that the audio play was even going on, and could not have guessed that its overhead was impacting my frame rate so dramatically. The other clue that might have tipped me off as to what was going on was the flickering of the CD-ROM LED, but this was not visible to me because my computer sits on the floor beside the desk, out of sight when I am racing.
The phenomenon of audio CD play not being brought out through the speakers by the sound card seems to be fairly widespread on Windows 95 machines. I suspect that many of the people on rec.autos.simulations who have experienced abominable frame rate may be suffering from the same problem.
If you have poor frame rate, go to the Sound options screen and turn off the Play CD Tracks option. If you're lucky, you may see as dramatic an improvement in frame rate as I did.
If I were in charge of development of a racing sim, and someone insisted that the game default to a mode in which music is played through the speakers while racing is going on, I would in turn insist that the installation routine include a check to verify if the sound is indeed audible to the user. I'd play a bit of an audio track, and ask the user if she could hear music from the speakers. If the answer was yes, I'd default to having the music play option on, present a warning that the audio could significantly impact the frame rate, and explain how to turn it off. If the user answered no, she couldn't hear music, I'd default to music off and explain to the user that some difficulty in the configuration of the machine prevents music play, but that this would not affect game play.
When I was unable to resolve all the problems on my primary machine, I tried my other machine, a P-133 with Intense 3D video card and Thrustmaster GP1 with CH pedals.
On this machine, the frame rate was better, but another problem surfaced: the game was totally unresponsive to requests to increase the paint-ahead distance. I increased the value, the frame rate degraded, but the paint-ahead distance didn't: it remained impossibly short. The sim was totally undrivable.
John Wallace remarks: No, you must have imagined that. Microsoft assured Marc Nelson and others that the software didn't adjust the draw ahead distance independently of user settings. "But we've seen it" we cried! "We set up two systems side by side, and at the exact same point on the track with the exact same settings, the P200 was showing the Firestone bridge and the P133 wasn't - it didn't appear until a couple of seconds later". "The software doesn't independently change the...." "WE'VE SEEN IT!!!!!" - "The software doesn't.." Ah, forget it.
After a lot more additional testing, I have found a resolution to this problem, at least on my computers. It seems that early versions of the RRedline drivers (the native API for the Rendition chipset) have a problem with clipping distance, displaying only those 3D objects which are relatively nearby. When I switched to a version of the Rendition drivers without RRedline support, the problem disappeared. But I need RRedline support to play SODA.
I installed the most recent version of the Rendition drivers, the 2.1 beta reference drivers from the Rendition Web site, and the problem was solved. I can now play SODA, and CPR now paints ahead properly. In addition, several other bizarre problems with replay and Pi analysis also were resolved.
My experiments in looking for a solution to my frame rate problem led to my worst disaster. After many hours of effort failed to produce a reliable work-around, I decided to go back to my second machine. But when I tried to boot this machine, it refused to boot into anything but Safe Mode. In a Normal Mode boot, it would get past the blank screen (where I believe Windows is initializing the Rendition card) and then pause, and then reboot itself again, this time into Safe Mode.
I tried switching drivers, and screen resolutions, but the only way I could recover was to restore a previous configuration, which I'd fortunately saved with ConfigSafe prior to installing CPR.
After the incredible hoopla that appeared in rec.autos.simulation around this issue after I published this review, I tried installing CPR on this machine again. This time, I let CPR install MS IE 3.02, an option I'd declined the first time around because I don't use this machine for surfing. Upon rebooting, I got a Windows Protection Error, but after another trip through Safe Mode (where, as before, the Device Manager reported that all was well) I was able to boot normally and play the game. At this time, it appears this machine is still working normally.
Now that I've had some time to give some more thought to this issue, I regard most OS corruption problems as a Windows 95 issue, and probably not an application issue in this case. The combination of the easily-corruptible yet vital registry, old .ini files, and the habit of all kinds of applications and system software installations over-writing existing dll's simply invites trouble and creates a maintenance nightmare. As an old Unix junkie, I feel that applications should never update system files and shared subroutine libraries, and system patches should have rollout capability and good logging. There's a good discussion of this at the Arachnoid Web site.
Adjustment Blindness. Certain adjustments, such as the aforementioned popup control, as well as brake bias from the cockpit, seem to have no effect whatsoever. I can set the brake bias slider to min or to max, and the car still locks up the fronts and slides for the wall at the slightest touch of the brakes. John Wallace notes: "Yep, both I and Doug Arnao found this. I have my doubts about several other of the setup options having an effect."
Since this initial testing, I've found that the setup files are text, so they can be edited with Wordpad. The problem with the brake adjustment is that the slider in the CPR garage will not allow me to get enough rearward bias. With Wordpad, I can set the bias to anything. A bias of 0.01 is a trip! You can lock the rears at 160 mph, going into turn one at Long Beach, and you go on a wild ride! With the bias set at 0.35, I find pretty good balance in my Long Beach setup; I'm able to use the brakes to get the tail out a little and point the car into the apex as I'm turning in.
Unfortunately, if I change anything in the garage on the page containing the brake adjustment (such as fuel level when I am going to start the race) the brake adjustment reverts to an undrivable value within the range permitted by the garage. The result is a cumbersome fumbling back and forth between Wordpad and CPR whenever I want to make certain adjustments. Grrr.
Steering Lock. Another garage adjustment that lacks sufficient range is the steering lock setting. Particularly when the Nonlinear Steering and Speed Sensitivity sliders are at the right-hand end of the sliders to cope with the steering dead zone blips, there is inadequate lock even at maximum to get around tight corners, such as the hairpin at Long Beach.
Note: This is being addressed in the upcoming patch.
Wall Glue. Longtime racer and racing and computer magazine editor Steve Smith has published a very informative new book, Winning: CART Racing Strategy Guide, focused on racing sims. In this book, which is included with the re-release of ICR2 as Papyrus CART Racing, Steve talks of the phenomenon in CART Racing where, if you hit the wall just right, the car gets stuck against it and won't get unstuck until you come to a complete stop. He says this phenomenon is unique to Papyrus CART Racing.
Not any more, Steve. The MS-CPR Wall Glue makes the Papy CR Wall Glue look like dried mucilage. Worse, the harder you press the throttle, the worse it sticks. I've even spun the car after it was stuck in the wall by leaning on the throttle in frustration.
Indycar Pinball. In Professional Mode, the highest level of difficulty, if you touch the wall even very slightly at speed, the car rebounds off like it was shot from a gun, and goes into a wild spin. This doesn't happen in Intermediate Mode, so I use the latter mode since I find playing billiards with my Indycar to be a somewhat irritating past-time. Also, Intermediate Mode allows me to turn on auto shifting, which I find necessary anyway because if I take my hand off the wheel to shift, the bizarre steering behavior ensures I will hit the wall.
Desktop Icon Scrambler. Every time I exit the game on either of my machines, I find that the desktop icons have been magically rearranged to fit in a 640x480 area. I am one of those foolish people who takes advantage of Win95's capability to place icons where I want them, so I find this immensely annoying. Since I have my screen set up for 1180x864, half of the icons wind up on top of other icons, hopelessly scrambled.
Really, Microsoft! What is going on here?
Race Results. At the end of the race, even though I finished fourth, the results reported my race speed as being faster than the winner's. What?
Tire Temps. John Wallace: "How the heck can you adjust pressure, toe-in, toe-out or camber when there's no tire temps? I'd guess that any engineer or driver would have picked up on that in a second, which pours water on the theory that [CPR] was made with all of that technical assistance." Anyone who's read Steve Smith's extensive discussion of this topic in Winning is likely to concur. I consider this a major omission from what is claimed to be a serious racing sim.
Autopilot. Why does the damn game have to drive the car out of the pits?? Am I chopped liver? Can't it let me decide when I want to pull out, and let me drive it down the pit lane myself, as if I were in a real car? This is supposed to be a simulation, right? Right??
To add insult to injury, when I finally decided to race, I found that the Pace Lap Genie drives my car for me till the green flag drops. Pah!
Note: Microsoft has stated that the upcoming patch will make the Autopilot a user option. Great!
"Good" Default Setups. In a FAQ on their Web site, Terminal Reality says that "unlike other sims, CPR comes with good setups." Great! Always frustrated at having to develop decent setups before I could race in other sims, I was excited about this, and pleased with Terminal Reality's good judgment in taking this route.
Hah.
I'm no setup wizard, but every one of the first dozen or so setup changes I made to the sample Long Beach setup resulted in a significant improvement. Similar tweaks to the sample setups were necessary to arrive at a drivable setup for all the other tracks I tried.
Good setups, huh? I'll bet my Pentium that neither Bobby Rahal, Alex Zanardi, nor Mark Blundell had anything to do with the sample setups shipped with the game.
Note: There are some much better setups available at the Apex Web site. Hopefully as time goes on, the hordes of dedicated sim racers on the Internet will make more and better setups available, at least to those who are willing to look for them.
Racing. At one point, I tweaked my setup to the point where I could do 52's at Long Beach. It was great! My confidence bolstered, I decided to try a race against the AI.
What a joke! The same AI cars that qualified within a second or so of my time were 80 mph or so slower on the straights. They were so lost in the darkness of the asphalt and shadows on the walls that they were invisible until it was too late to avoid them. If they were behind me, and I wasn't up to their speed, they rammed me, even if I was on the straight! So much for Microsoft's excuse that they hit you because they can't see you in time.
Even with a collision at the start that left me last, several spins after collisions with AI cars, and a trip down the cul-de-sac at the end of the pit lane, courtesy of the Nonlinear Steering Bug, I still finished fourth out of 15. Huh?
It was one of the least satisfying sim racing experiences I ever had.
Note: Microsoft promises much improved AI with the upcoming patch.
Other Goofs. The list of minor but annoying design blunders and omissions goes on. I couldn't find any way to print setups. You can't reduce roll stiffness in the cockpit; you have to cycle through the settings to get a softer setting. There's no sound in replays. You can't go into the garage during a practice or qualifying session without ending the session (so how can you sort out the chassis?)
The pit board doesn't give opponents' times in qualifying - only their position relative to me on the track(!) Why would I want to know that?? Maybe so I will know when I am about to run over one of them?
Argh!
In response to my cries
of dismay as I struggled to get CPR to work well enough to play it, John
Wallace commented: "MS
need to face it, Win95 just isn't a games platform. Too many other things
messing around with the system to have any clue what it's going to be like
on any given computer. That is especially true with D3D."
And yet...SODA works quite well. Frame rate isn't that great but if you
turn off some of the graphics options it's quite acceptable. The scenery
isn't pretty, but the game works well.
In fairness, I must note that while SODA worked well on four machines with Rendition cards, it was virtually unplayable on P-150 that lacked a 3D video card. I couldn't even get it to run on a P-133 laptop.
Perhaps the pattern that is emerging here is that games run best on their targeted platforms, despite Microsoft's goal of having DirectX isolate the developer and user from needing to worry about specific 3D video cards.
Hopefully as the technology matures, this goal will become less elusive.
The conclusion of my CPR review is in the next page.