Official shmupmame Super Turbo thread


Yea, mouse rate only checks the polling rate for the mouse. Just toggle between the different polling rates and test them out via your joystick…it’s pretty noticeable between each setting.

Testing Update: Disabling Bi-linear filtering on Direct 3D also seems to reduce the delay slightly.

Ah…someone has been on the forums I see! I’ve been following the developments of groovyMame for the past few months. I definitely caught the post mentioning the “-frame_delay” option and it sounds very intriguing. I haven’t spent too much time with the emulator, so I don’t really know the optimal settings…I think it works best with Radeon graphic cards using the CRT_EMU and/or SwitchRes programs along with WinXP I believe. Hopefully the community can delve into this other Emu and compare results.


Ha! Yep! I’ve been fooling around with it for a while and its really good, imo. I set the framdelay to 1 and the apparent (subjective mind you) input lag is very low, especially in combination with 1000hz setting on my stick. Been testing with a lot of very input lag sensitive stuff like super turbo and many shmups.

Its great, too, because unlike shmupmame, you can access the entire mame library. The ati support is much better than its nvidia support, right now, but I know its nvidia support is being worked on.



papasi, when you did you “regular” MAME measurements, can you tell us what version of MAME and what video settings you used, including whether v-sync, triple buffering, etc. were on / off?



i downloaded the latest mame 0149, everything is default.


Thanks. I believe stock, MAME does not use v-sync or triple buffering but DOES use d3d / post-processing filter instead of directdraw / no filter. I think this (d3d + filter) is laggier than directdraw / no filter.



Yeah, 360 controllers use a “Vendor Specific” device class (FFh). Sounds like that program will handle any HID device (class 03h). Unfortunately, even if the author of that program wanted to make an exception in order to support 360 controllers by allowing vendor specific devices, there’s really no general way to be guaranteed that it’s a 360 controller and not some other crazy vendor specific device. With UD-CPS2, I am able to make the assumption that people aren’t plugging in a USB coffee pot, so when I see a device class of FFh, I assume it’s a 360 controller.


Coffee pot Latte setting Pro at ST.


Training mode is buggy as hell. XD I went to jump in combo with dictator and the dummy character flew off without a single hit to the floor a couple times. Reminds a lot of CCCv2 ST. Weird/funny stuff.

The cheats need a major update. Other than that great stuff.


If that bothers you, don’t use the “training mode” cheat, instead, use the “infinite p2 life” cheat and bind the “toggle cheat” shortcut (default is F6) to your 7th or 8th buttons on your joystick.

Double tap the toggle cheat button when the dummy’s health is down to 50%.


I have to agree with James chen on your sig. There is something special with ST on a cab. Nostalgia factor.
I was drinking my ass off at NL just past friday with damdai and a handful of players. Fun times every friday.
I still need to master sanwas though.


Interesting GroovyMAME testing results!!:,133194.0.html

Scroll down to Calamity’s post towards the bottom.



Dude… I can’t thank you enough. I noticed I was having a weird time playing in Linux, and this link has confirmed why. Im also feeling some delay with ZSNES, which also uses SDL, IIRC. I’m currently playing Super Metroid a few times so as to learn the maps and eventually try some speed runs, but the commands are a bit strict*, Square gate screws my precision and there’s input latency on top of all that.

*wall jumps are very important in that game for speed, and if you ever reach down or up microswitches while doing them, spin jumps are disabled, wall jumps become impossible, and you fall all the way back to the bottom.

Edit: if that (currently) last post is right, I will also be able to increase the USB poll rate in Windows XP, which I thought was impossible without major trouble, i.e., learning win32 and doing it myself.


Absolutely! Calamity’s testing shows .65 frames of incremental lag over the cps2 hardware. That is pretty phenomenal! I’ve done the usb polling rate tweak and it seems to further help. At that point we are likely hovering RIGHT ON TOP of the hardware.



sorry i haven’t followed groovymame’s development.

so it has .65 frames at best? that’s not better than shmupmame is it?

i didn’t go into the details of how shmupmame reduce the input lag (as well as groovymame).
can groovy or shmup take advantage of each other’s trick and make it even better?


regarding input lag. will it be reduced if I use a non-USB keyboard?
Eg playing the game on laptop.


Since the original developer of shmupmame is not supporting/replying on his forum, I couldn’t make any contribution to make it easier for ST tournament/practice.

OTOH the groovymame dev did the opposite, I have now attached a modified version of groovymame that does

  1. fast buttom mapping for P1 / P2
  2. fast save/load state – no prompting for slot numbers and you can easily map the extra 2 buttons on your stick to do save/load. Just turn on cheat > training mode + infinite time.


FYI, piFBA was released for the Raspberry Pi. CPS2 now runs at 60FPS, but I am experiencing significant input delay. If any Linux gurus with a Pi want to download the src and mess around, please do so. I think this could be an excellent, portable, low cost alternative.


this has potential to be huge for 2d fighters because those raspberry pi things are only like 35 bucks a pop. Maybe rufus can take a look into this.


Mini ST as the future?


I think the biggest problem for ST has always been the lagless display, now that we have solved the custom input controller issue (ud-cps2).