TGM and TAP framerates

Thread in 'Research & Development' started by Muf, 10 Jul 2011.

  1. Muf


    Since I've had my Datapath VGA capture card, I've had some insight into the actual video modes for TGM and TAP. And they're not as straightforward as you might think:

    TGM1: H 15.738KHz, V 59.840Hz, 320 x 240 progressive
    TAP: H 16.165KHz, V 61.680Hz, 320 x 240 progressive

    What this means, is that all current MAME versions run TAP at an incorrect framerate (60fps instead of 61.68fps). For TGM1, the difference is negligible.

    If you wonder how this could slip by unnoticed for all these years; the answer is that the programmers assumed 60.0fps when writing the game timer routines. So on MAME, the game timers are correct: 8 minutes is actually 8 minutes. On the actual PCB, 8 minutes on the timer elapse in 7 minutes and 46.7 seconds of real-world time. This is a quite significant difference, and it's just been confirmed with stop-watch timing by colour_thief and DeHackEd.

    I've made an account at, but I can't file a bug until having waited 24 hours for approval.

    Consequences: all TAP MAME records are technically invalid due to having been played with 2.75% slowdown. Even when the video modes get fixed in MAME, it will be mandatory to play without "syncrefresh", unless you have a custom video mode (for instance with Soft15KHz).

    When the bug gets fixed, a new version of shmupmametgm will be compiled, and all old records marked with "speed bug" in the rankings.
    Last edited: 28 Jan 2015
  2. K


    until we discover another new hidden shit....
  3. Shit hell.
  4. So this means that, in one sense at least, people using MAME are playing the game the way it was intended to be played.

    Wouldn't it be better to just split the records threads into TAP-PCB and TAP-MAME?
  5. Muf


    With the current amount of PCB players/owners? No.
  6. So clones are even more inaccurate than we thought...
    Does this apply to TGM2 or just TAP?
    This should be added to the wiki. Is anyone allowed to edit or just a few people?
  7. Fucking framerates, how do they work?
  8. Anyone can edit the wiki.
  9. mat


  10. Do the various versions of LTAP run at 60 FPS or at 61.7?
  11. Probably 60. LTAP is accurate at the software level and was not informed by any hardware behavior.
  12. I remember a long time ago that Mat or someone had mentioned that the "Tetris Japan Finals" video was sped up, since the timer didn't match the video length. Now we have an explanation for that. Interesting.
  13. !!! I remember that too! I guess it's just the native speed of the game. Damn.

    Now this means sub-8 is that much further away :'(
  14. mat


    yeah, if i were smarter i would have looked at more than just that video and figured this out at the time. or maybe i did look at other videos and just assumed it had something to do with the way arika was capturing? i don't remember--regardless, this is silly.
  15. Zaphod77

    Zaphod77 Resident Misinformer


    Color NTSC field rate is 60/1.01 or about 59.94 fps. only black and white was exactly 60 FPS.

    So even accounting for that, TGM is ever so slightly slower, unless 59.84 was a typo. Considering that playstations tend to output NTSC, and tgm1 is "playstation hardware", i'm suspecting a typo.

    That doesn't explain the Psikyo hardware issue, though. Strange that it would be fast.
  16. Muf


    It's not a typo. And game consoles are known to "play around in the margins" in terms of video standards; 240p is not part of the NTSC standard, either. -0.1Hz of timing difference won't throw off any TV, and it's very possible that consumer Playstations use that same video mode. Fact is, most TVs will happily accept strange video modes, and that fact is exploited by several if not all consoles produced in the eighties and nineties.
  17. Zaphod77

    Zaphod77 Resident Misinformer

    I didn't claim that it definitely was a typo, but that it appeared it could be one. :)

    and all this trickery sucks when you don't have a real crt TV around. with the NES it's particularly jarring, because instead of dot crawl you get stationary dots on a progressive display. (and I had an old composite computer monitor that did exactly the same thing. he dots didnt' crawl, and the NES looked like crap on it with diagonal lines everywhere)
  18. Alright, finally I'm caving in. I'm buying a TAP PCB soon, along with TWF and Samael and maybe other Seattle people if they're interested in chipping in. I don't want to redo my scores in a fixed version of MAME, so I'm finally moving to the real thing. Guess this just means I'll really have to keep playing. :(
  19. tepples

    tepples Lockjaw developer

    Checkerboards between two different hues or between black/gray/white and a color would produce diagonal lines on NTSC due to the shortcuts that the NES PPU uses to generate its composite output signal. Nestopia is capable of simulating NTSC artifacts.

    As for frame rate, the picture on an NTSC or RGB NES is at 60.0988 Hz (see full derivation), which is a lot closer to 60 Hz than TAP is. Accurate NES emulators already run at 60.0988 Hz like the console, and PC-based emulators are laggy anyway, so I guess LJ65 records wouldn't need to be invalidated.

    I'd assume TGM2 uses the same video mode as TAP. Wasn't it just a software upgrade from TGM2 to TAP?
  20. Zaphod77

    Zaphod77 Resident Misinformer

    The NES phase shifts the colorburst every frame in order to make the picture more stable and get better color reproduction without artifacts. This causes dot crawl on a still picture, normally.

    But on a progressive display, this dot crawl is nonexistent. I can see the diagonal lines on the flagpole on super mario bros. If you have a ps1 and metal gear solid, look at the soliton radar's vertical lines to see what i'm talking about. The dot crawl gets frozen in time on a progressive display.

Share This Page