Figured I'd do a little test to check for any timer discrepancies between regional variants of the game since I know it creates timing issues for other Speedrunning communities on N64.
This test is not intended to determine if a version is faster than the other, but I may try some PAL runs in the future to determine that.
A few notes about this test. Both runs have the exact same settings: Music Off / Voice Off / SFX On / Stereo off
Both runs use the same car (#1 as I couldn't be bothered to unlock everything on PAL again)
Both times are synched to when the Green Light comes on and the timer starts. (The Red light appears on the same frame as well)
Not sure why the windows on each console are different, maybe something to do with running the PAL console through S-Video (or maybe my RetroTINK is being weird)
Observations: The PAL timer seems to run through milliseconds differently to NTSC, and didn't appear to ever stop exactly on a new second. In fact it skips an entire tenth at the top of the second (e.g. The timer jumps from 0:59.99 to 1:00.09). You can actually see this at regular speed, the timer seems to hitch at the top of each second.
The differences between both versions is practically nothing, so clearly the next step is to do some PAL runs to check if any courses are slower/faster.
If there's any way I could improve this test let me know.
Really interesting that the PAL version doesn't properly update the last digit in the timer when going to the next second. It does seem like that time is kept properly in the background though. However, PAL still being about .07 behind after a minute seems more significant because it appears to be accumulating. Curious if it gets really bad especially over the longer tracks.
Also, if you complete a lap on the PAL version in that moment that the second ticks over (the 0:59.99 to 1:00.09 thing), does the results screen show the correct number? Do you have any records that end with .00-.08?
Testing the clock to 4 minutes (since the longest tracks only go up to 3:30ish) would be a good plan.
Testing for lap times between .00-.08 would also be a good idea to test another curiosity that I've had. I've noticed on a few record runs (both mine and others) that the math doesn't always add up properly (e.g. The combined times of all three laps doesn't match the final time) and the discrepancy can sometimes be up to a tenth of a second.
I have noticed the lap total to final time discrepancy, but I thought that was due to the timer truncating the 3rd millisecond digit. Really every time is like 1:00.00X. If each lap ended with a .XX9, then we could see up to 27 milliseconds of difference from the total time. But that doesn't account for a full tenth of a second, so maybe something else is going on.
Ok, so a little more insight into how the clock works in real time and what times we see out of it.
On NTSC, the lap timer advances frame-by-frame in a repeating pattern of 3-3-4 milliseconds (e.g 00:00.03 > 00:00.06 > 00:00.10 > 00:00.13 and so forth).
On PAL the first frame of the minute is 9 milliseconds, then advances by 2 to 11 milliseconds, followed by instances of 4 milliseconds every frame until .99, where is skips to .09 again (e.g 00:00.09 > 00:00.11 > 00:00.15 > 00:00.19 > 00:00.23 etc.)
From console lap times that have been submitted, we can see that lap records do appear to be calculated to the millisecond, otherwise all NTSC lap times (which is the region I assume most emulator users use) would end with either a .X3, .X6 or .X0
From that data we can assume that PAL lap times can end with a number between .99 and .09, but I'll test to make sure
One comment I'll add here is that according to several sources online this game runs @ 60fps on NTSC and 50fps for PAL. It does look like the timer is roughly adjusted to account for the framerate difference.
Based on my first-hand knowledge/previous testing of other n64 games like Perfect Dark and Diddy Kong Racing with similar framerate problems I would suspect that NTSC is faster for courses that maintain a high framerate where PAL could be faster on levels with slowdown problems(Midnyte Rumble/Alpine Ridge likely). Clearly, it's not like Mario Kart 64 where the timer isn't adjusted but also it does have lots of frame-rate problems on console(maybe even emu this game even struggles on that a bit looks like). A comparison like I did on alpine with PAL/NTSC should indicate clearly if that's the case maybe I'll break out the PAL console(though I no longer have a TV that supports it I can still run it through capture card).
The comparison to Mario Kart's timer was the main reason I decided to check it, in case timer adjustments needed to me made for PAL runs, fortunately it doesn't appear to be the case.
The difference in framerate is honestly why I never thought to run the game using my PAL cart, despite having it since I was a kid.
There is a distinct opportunity that PAL is just faster overall 60fps is a mistake for this engine/game should have been 30 it struggles a lot there's good reason most serious competition is done on emulator.
I figured out how to read the save file's hexcode and deduced the addresses of the time trial records. Running at 60fps on emu showed that the complete milliseconds only end with a 3, 7, or 0 which is expected. So if you wanted some hyper accurate numbers during your comparison, it's not too difficult to get. Obviously easier to get to the emulator save file, but if you have some magical way to get a console save file, it should still work.
