TGM3 Input/display LAG

Thread in 'Research & Development' started by K, 14 Jul 2011.

  1. Muf


  2. I tried to find that in about everything possible (game, tgm3, typex_loader, typex_io, etc.), but I can't. Is there another way ? subst [noparse]D:[/noparse] doesn't work.
  3. Muf


  4. Ah, so that's what you're talking about.
    Yeah, I found "sramdata", I modified it to [noparse]"E:\Tgm3"[/noparse] and now it works ! Thanks a lot mufunyo !

    Now I might post records on this forum. For instance, my best records in Tex were S6 in Special Ti and 2xx on Sudden Ti.

    (edit : fixed a typo : was -> were*)
  5. Edo

    Edo a.k.a. FSY

    By the way, to prevent the [noparse]:\[/noparse] being parsed as a :\ smiley, you can wrap it in [noparse[noparse]][[/noparse]/noparse] tags. Or, if you click "Go Advanced" instead of "Post Quick Reply", there's a check box to "Disable smilies in text".
  6. [noparse]:[b][/b]\[/noparse] is fewer characters. :p
  7. i think it is not just input delay, maybe it's something designed inside the game.

    i tried to TAS play, in 30fps, (using some BBS reader, see this )
    and i felt that when i want to move/rotate a tetrimino TWICE in one direction, if i repeat pressing the button too fast (but of course much less then 15TPS), it will kick in DAS and fly away / rotate only once.
    seems if the delay between two keypress event of one key is too small (smaller than (2 or more) frames), they will be regarded as only one.

    :o = 囧
  8. Muf


    The game polls for input once per frame.
  9. Maybe because it takes two frames to make a button press, not one?

    Game polls for input on frame 1. Button is pressed.
    You release the button, but game hasn't polled for input again.
    Game polls for input on frame 2. At this point, you've pressed it again.

    But the game thinks the button is held down, because it's down when it polled for input this frame, and it was down when it polled for input last frame.

    In short, the game needs to poll a release of the button, or it will think you've been holding it down the whole time.

    It's especially bad with playing at 1 FPS. You have to press the button for a full second until rotation, then release the button for a second, then hold it down for a full second to do a double rotate using only one of the rotate buttons.
    Last edited: 24 Jul 2011
  10. i have mentioned the speed i play..
    i played at 30 fps ....
    and i believe that my fingers can never get the speed beyond 15taps/s ...

    faster than 4taps/s, as i feel, something bad happens.

    i think after pressing a button, it waits for some frames to "comfirm" if you press a rotation key, in order to perform a (wtf move, can't remember the name, u know that.. rotation->move->fall).
  11. Why are you playing at 30fps? What are you using to play at 30fps and how confident are you that it's not screwing stuff around? How sure are you that the game is running at a consistent 30fps? How sure are you that whatever input hack you're using isn't screwing around how your inputs get sent? (These are rhetorical questions - I don't actually care what the answers are)

    You're using crazy hacks and tools to try and make judgements about how the game responds to inputs. That's a totally dumb way to go about doing things.

    And even if what you say is true, it's a completely irrelevant finding when the game is played at 60fps, because I strongly doubt that the main reason people feel lag is because they're constantly repeating button presses faster than 10 a second.
  12. Synchro. They're one-frame maneuvers unless you're DASing into a wall (i.e. an AutoSynchro). It doesn't leave your rotation input out there to make the timing any more lenient than a single frame window.
  13.'s because it runs without frame rate limit on some pc, and someone discovered a tool(mentioned above too..)to limit the frame rate to 60fps. sure it can be 30 fps. and i use fraps, it shows the frame rate.:sneaky:
  14. Zaphod77

    Zaphod77 Resident Misinformer

    The issues with TASing are most definitely related to the input hack.
  15. K


    since 2 weeks i got back the 240 fps camera wich allowed me to perform more test on TGM3 (different PC + supergun + sega I/O + CRT monitor), TGM2, TGM1, texmaster and shmupmame..

    about how i counted the frame : i validate when the button is pressed when the electron beam of the CRT reach middle of the screen. people might use a different way but since i keep it the same for the test i think it is reliable.
    all game using the USB arcade stick don't have lighting button so it's a smashing button approximation of the contact momentum.

    i will not put online all the videos but the most interesting to us.

    --- TGM1 original board = 3 frames delay
    --- TAP original board = 2 frames delay
    --- shmupmame TAP = 3 frames delay
    the best result i get is to keep windowed (if maximised the screen rez must not change). the frame delay is inconstant but quite reliable.
    --- shmupmame TGM = 4 frames delay
    same as shmupmame for TAP.. it is strange to see that the TLS update the frame BEFORE the active tetramino (the same appear in shmupmame TAP) !
    --- texmaster = 2-3 frames delay
    i'm not exactly sure about it but it's quite consistent

    --- TGM3
    so come the real interest of those test. There were only one test about the typeX_loader.exe and it was so laggy that i was clearly useless trying to find out the best configuration.
    i abandonned all hope about win7 (and vista?) for playing TGM3. Muf and other confirmed me that this OS add extra frame(s) delay to openGL and the best i could get was alway 5 frames delays . So i got back to XP. i've got an ATI graphic card and tried to lower as much as i could the graphic fancy 3D options. The catalyst control center is poor in 3D options so i used the ATI Tray Tool from for more optimisations.
    whatever i do, i always get the best result by keeping the game windowed instead of fullscreen, with a screen resolution of 1024x768 (60hertz). The game run with delay of 'only' 3 frames wich is better than muf original typeX... but i still don't get it all the time : for instance it's better to reboot the sega I/O each time the game is started again, and not trying to move the windows with the mouse (pause)
    and you can find the raw videos at
  16. Amnesia

    Amnesia Piece of Cake

    I am really really surprised here.
    It is not that I dont believe you but...:|
    shmupmame doesnt work that well for me.
  17. It really depends on your computer.
    I play on a laptop at home which is rather powerful but I "feel" much more lag that on my good old pc at my parent's house (which has quiet the same specs than my laptop). How do I feel the lag ? I play 40 credits of death in a row and check my results, the average is FAR better on the pc than on the laptop.
  18. The lag in that case probably comes from the LCD screen in the laptop.
  19. K


    Oh oh oh...
    More updates to come...
  20. K


    Last edited: 29 Sep 2011

Share This Page