Garrett Holland's Feed

Garrett Holland
01-20-2021 at 11:59 PM
17 Comments
Rate this Entry

Atari 2600 Pole Position - TAS and analysis - Fastest time possible. ~212¼ seconds.

Official OEM ROMs and cartridges, both NTSC and PAL:

Fastest Qualifying Lap Time possible: 56.73s
Fastest Grand Prix Lap Time possible: 52.28s
Fastest Full Race time possible: ~212.28s
Maximum seconds left at the end of the race: 15s
Scores above 61,000 are not possible.

All of the above are possible to beat on a hacked ROM, hack cart, or nerfed console.

Comments
  1. Scott S.'s Avatar

    I should note driving perfect laps gets you a faster time but a lower score (60,910 max), but driving imperfect laps (going around cars off-road in the corner and occasionally skidding) allows you to pass 1 or 2 more cars. How much this quirk can be exploited, I don't know (yet). There might be an issue with emulators, as noted in this thread: https://atariage.com/forums/topic/166438-the-official-2600-easter-egg-and-programming-mysteries-thread/page/4/?tab=comments#comment-3716975

    Here's a post from Jeff Adkins (AKA Northcoastgamer) where he talks about driving imperfect laps to get a higher score: https://atariage.com/forums/topic/267079-season-6-the-new-hsc-week-23-pole-position/page/3/#comments

    I was seeing all these other folks getting 60,910, yet I was running nearly flawless laps and the best I was getting was 60,860. This was aggravating the **** outta me. So, I watched a Youtube video of someone scoring 60,910. He/she was going around cars off road and skidding here and there, no where near perfect. So I threw the perfect approach out the window and employed the slower sloppy method and was finally able to score 60,910. Upon completing the final lap, I also noticed that the next car was pretty close to being passed as well. So I slowed it down a bit more and that's how I scored 60,960. It's crazy how this slower method works to score higher, but it does.


    There's also a hacked version of the game that's been floating around that allows for higher scores, so emulator submissions need to be carefully verified.

    ThanksGarrett Holland thanked this post
    LikesGarrett Holland liked this post
    Updated 01-21-2021 at 05:00 AM by Scott S.
  2. Garrett Holland's Avatar

    There's about 3/4ths of a second to play with, and is perceivable to slow down an aggregate of that duration during the grand prix that enables another car to appear and pass for the extra 50 pts, which is what Greg demonstrated in his perfect 212-second 60,960 run. I do have a theory regarding that "extra car". There is a bit of RNG involved in *when* the first car appears. The sooner it appears, which is different every time, the more likely the player will get the extra car at the end of the race.

    My motivation for creating this analysis was to either validate or debunk a claim of either a 62K+ score or a sub-212 run on original hardware or emulation with the official release ROM, and specifically, the recent video by the dude who ran a 199-second, 63K+ game, and passed less cars than Greg did. A whopping 13-second difference Seemed fishy, especially since that player ran off the road and slowed down a couple of time and couldn't imagine how Greg could possibly have lost 13 seconds somewhere in his run. The quick answer: Greg didn't lose any time. He simply used the correct OEM version, whereas the 63K player didn't.

    Ill be juxtaposing Greg's video and Mr. 63K's video by the weekend. I predict that I'll discover the 63K player used a hack version that increases the max speed from A3 (163 units) to something around AF (175 units).

    LikesRaGeNyC, GregDeg liked this post
    Updated 01-21-2021 at 06:25 AM by Garrett Holland
  3. Scott S.'s Avatar

    I've offered to dump that YouTuber's cart to see if it's actually different, so hopefully he'll take me up on that.

    The other possibility is there's a strange glitch in the program that's allowing for these extra cars/higher scores.

    LikesGarrett Holland liked this post
  4. Scott S.'s Avatar

    Here's the thread on Atariage where Kurt Woloch looked at the code to see how the scoring works: https://atariage.com/forums/topic/225033-fixing-pole-position-and-enhanced-source-code/

    I didn't look into the details of scoring, just noted that it's tightly coupled to the car's position on the road... the tens digit gets advanced smoothly, and each time it fully reaches a number, the variable CARY, giving the LSB of the position of the car on the road, advances by one, so there's 1024 possible positions. The first lap gives 10,000 points, and the following ones give 10240 points each if I'm correct. The question would be how much the maximum time remaining would be, and the maximum cars passed. I don't know at which rate the cars can be passed and what lap time the maximum speed would give. Then you'd have to factor in some losses in the first lap of the race by accelerating to full speed, which is further complicated by the fact that there's two gears, so you'd have to find out the optimum point for shifting into 2nd gear as well. Once you've done all this, it should be possible, I think. In my tests, it was certainly possible to go through the laps without ever slowing down (which would be impossible in reality! Keep in mind that the track is based on the then-current version of Fuji Speedway, but greatly simplified to fit into the cart).

    This is the thread that was linked to from my high score page where "MayDay" talks about getting 63,210 and 63,260. Near the end of the thread, he talks about the extra car anomalies.


    ThanksGarrett Holland, Luigi Ruffolo thanked this post
  5. Scott S.'s Avatar

    So the person who posted this YouTube video () posted a video of his cartridge. It's a later re-release by Atari Corp and would have come out between 1986-1988 (impossible to know for sure w/o the original box). The other issue could be related to which variant TIA chip is in his system. I asked him if he has another system to try. He also mentioned he has another cartridge coming, possible an original 1983 release (he noted the label was different, which it was).

    ThanksGarrett Holland, Luigi Ruffolo thanked this post
    Updated 01-21-2021 at 04:28 PM by Scott S.
  6. Garrett Holland's Avatar

    In the video, I demonstrate that going maximum speed the entire run other than the acceleration phase in Lap 1, eliminating the threat of collision from the enemy cars, and ensuring maximum speed even when going off road or skidding, results in a remaining time left at about 15 seconds. If the timing of the changing of gears from LO to HI is a factor, it does not explain his sub-49-second laps 3 and 4, (and he even ran off the road more than once in lap 3 losing a bunch of speed) or his 199-second race time. I cant see a loss of 13 seconds manifested by one poorly timed gear shift at the start of lap 1 on the grand prix phase. Laps 2, 3, and 4 of the grand prix phase in the TAS start and end at full speed, remaining on full speed on 100% of their respective frames returning a 52.28-second result, so shifting would not be a factor at least for those three laps.

    ThanksLuigi Ruffolo thanked this post
    LikesGregDeg liked this post
    Updated 01-21-2021 at 09:47 PM by Garrett Holland
  7. Scott S.'s Avatar

    The cartridge that guy has is the 1987 variant:

    http://www.atarimania.com/game-atari-2600-vcs-pole-position_19383.html

    It looks identical to the 1987 NTSC release:

    http://www.atarimania.com/game-atari-2600-vcs-pole-position_18702.html

    Greg mentioned:

    He ended up finishing with 28 seconds in reserve (13 more than usual) which gave him an extra 2,600 points. If it wasn’t for the extra time his final score would have been 60,660.

    The only way to know for sure (esp if the TV can handle both) is to dump the cart and run it in an emulator. If he's not willing to send me his cartridge, I'll try to find both 1987 versions and go from there.

    ThanksGarrett Holland thanked this post
    LikesGregDeg, Garrett Holland liked this post
  8. Garrett Holland's Avatar

    That will be interesting to see if Atari changed their own ROM and released it. They did that with Stargate/Defender II and Atari Video Cube/Rubik's Cube (albeit just graphical changes), so it wouldn't be a first. But changing the gameplay (such as max velocity) would be surprising. Scott, you know waaaaay more than I do regarding this ... have they done that with any other game that you're aware of?

    Updated 01-22-2021 at 09:01 AM by Garrett Holland
  9. Scott S.'s Avatar

    Several games were released under different names, but only a few made noticeable changes. Usually if a game was re-released (by the same company), it was to fix something. I'll put together a list and post it. I can name a few examples. Burgertime and Decathlon were 'revved' to fix bugs that affected gameplay. There's 2 different versions of Superman; the first version has some tricks that affect your timer. Taz was changed to Asterix for release overseas. Same game but different graphics. Most of the adult games were released under 2 different names (and with different graphics) to cater to either male or female players.

    ThanksGarrett Holland thanked this post
  10. nads's Avatar

    Quote Originally Posted by Garrett Holland

    There's about 3/4ths of a second to play with, and is perceivable to slow down an aggregate of that duration during the grand prix that enables another car to appear and pass for the extra 50 pts, which is what Greg demonstrated in his perfect 212-second 60,960 run. I do have a theory regarding that "extra car". There is a bit of RNG involved in *when* the first car appears. The sooner it appears, which is different every time, the more likely the player will get the extra car at the end of the race.

    My motivation for creating this analysis was to either validate or debunk a claim of either a 62K+ score or a sub-212 run on original hardware or emulation with the official release ROM, and specifically, the recent video by the dude who ran a 199-second, 63K+ game, and passed less cars than Greg did. A whopping 13-second difference Seemed fishy, especially since that player ran off the road and slowed down a couple of time and couldn't imagine how Greg could possibly have lost 13 seconds somewhere in his run. The quick answer: Greg didn't lose any time. He simply used the correct OEM version, whereas the 63K player didn't.

    Ill be juxtaposing Greg's video and Mr. 63K's video by the weekend. I predict that I'll discover the 63K player used a hack version that increases the max speed from A3 (163 units) to something around AF (175 units).

    Thats the interesting part about the cars and scoring.

    Greg passed 58 cars which is the same as my emu run and same for a lot of other 60910 games. The scoring is weird for the game as well, as mentioned if you finish with 15 ticks left @ 200 points each Thats 3000 add that to 54960 and you get 57960 now add the cars passed , 58 * 50 = 2900 now the total is 60860 , where did the extra 100 points come from?

    The point values come from the Game Manual.

    Barthax’s PAL run yielded 60910 but he only passed 57 cars, again where did that extra 100 come from?

    The other thing I noticed with the race time, not sure if it has already been mentioned but it’s not the sum of the race laps. The total time shown At the end will always use the last 2 digits of the last lap ie you finish with 52.38, the total time displayed will be XXX.38. For an old game it does seem to hold some mystery!

    LikesGregDeg, Garrett Holland liked this post
    Updated 01-22-2021 at 10:56 PM by nads
  11. Garrett Holland's Avatar

    Update: Ive downloaded every available version ROM from AtariMania. According to their checksums, all of them are one of 4 known versions: Official NTSC (MD5: ...1fe), Official PAL, CCE hack, and "Tron" hack. I tested all 4. Fastest lap possible (100% velocity every frame of lap): 52.28 seconds on each of the four, including PAL.

    Likesnads liked this post
  12. Garrett Holland's Avatar

    Quote Originally Posted by nads


    Thats the interesting part about the cars and scoring.

    Greg passed 58 cars which is the same as my emu run and same for a lot of other 60910 games. The scoring is weird for the game as well, as mentioned if you finish with 15 ticks left @ 50 points each Thats 3000 add that to 54960 and you get 57960 now add the cars passed , 58 * 50 = 2900 now the total is 60860 , where did the extra 100 points come from?

    The point values come from the Game Manual.

    Barthax’s PAL run yielded 60910 but he only passed 57 cars, again where did that extra 100 come from?

    That is odd, indeed! I haven't gotten around to studying enemy car behavior yet (I definitely will, though), just fastest lap time and the 13-second loss of time by literally everyone on the planet who has released a video who remained at 99-100% speed the entire race ... except one single person who managed to not only not lose 13 seconds, but did so while frequently skidding and running off-road multiple times every lap, one of those times dropping their velocity to almost 50%, and STILL posting a 48-second time on that same lap. As soon as I can wrap my head around that problem, Ill look into the passing cars and the phantom 100 points.

    Quote Originally Posted by nads
    The other thing I noticed with the race time, not sure if it has already been mentioned but it’s not the sum of the race laps. The total time shown At the end will always use the last 2 digits of the last lap ie you finish with 52.38, the total time displayed will be XXX.38. For an old game it does seem to hold some mystery!

    Ohhh ... this is fascinating! Great observation! Another interesting observation regarding that: If the player got a 52.28 on lap 4, 52.xx on laps 2 and 3, and a 56.xx on lap 1, the race time as shown on the in-game scoreboard will be 212.28, no matter what the .xx was. So it appears as though in order to get the last two digits to be less than .28, the only possible way would be to get between 53.00 and 53.27 lap 4, but then that will add a second to the total lap seconds making it 213.27, so it looks like 212.28 would still be the fastest race time possible either way.

    LikesGregDeg, nads liked this post
    Updated 01-22-2021 at 08:38 PM by Garrett Holland
  13. redelf's Avatar

    If I remember right the game adds your qualifying, 2nd, 3rd, and 4th lap whole seconds and then adding the last 2 digits from lap 4 to get your total time.

    So really you have just under 3/4 of a second for each lap 2-4 to slow down and pass more cars.

    Thanksnads, Garrett Holland thanked this post
    Likesnads, Garrett Holland liked this post
  14. Scott S.'s Avatar

    I asked Gordon Taylor to look at the recent YouTube video by "LightBlue2222". By going frame-by-frame at the glitch parts, he noticed each time the video glitches, the title screen is actually flashing for a split second! He feels either the cartridge or the video has been hacked or altered, but I'm still leaning on either a particular issue with the 1987 cart releases that may affect certain VCS models (specifically 1 or more variations of the TIA chip). The YouTube user just purchased a copy of the original 1983 version and posted a video of it, along with some gameplay:

    He only played one game, but the score was more of what would be expected.

    Thanksnads, Garrett Holland thanked this post
    Likesnads, Garrett Holland liked this post
  15. Scott S.'s Avatar

    The user on Atariage, Mayday, who got the +62k score back in 2014 used an emulator.

    Thanksnads thanked this post
    Likesnads, Garrett Holland liked this post
  16. Snowflake's Avatar

    Quote Originally Posted by Scott S.

    The user on Atariage, Mayday, who got the +62k score back in 2014 used an emulator.

    this is wheer i realize i'm old. i could've sworn he said he got the score in the 80s, which would make it hard to be on emulation.

    rereading he said played as a kid, not played in the 80s

    of course to me, playing as a kid means in the 80s, dang it, never occured to me young people playing the game might have played it "as a kid" way more recently

    ah reminders of old age

    LikesGarrett Holland liked this post
  17. Scott S.'s Avatar

    I recently asked "MayDay" about it and he confirmed he used the Stella emulator. He wasn't sure where he got the image from - either Atariage or EmuParadise. I have 6 Pole Position carts at the moment, and some of them are the 1987 version. I'll dump them as soon as I can and let everyone know the results.

    LikesGarrett Holland, GregDeg liked this post
Join us