I have disqualified a lot of recordings by MvdV. Reason: Slowdown. I think the following says it all: He had a 219,000 on Pnickies on regular MARP, but under controlled circumstances (tg3 version of MAME), he got only 131,900. And just look at the enormous difference in how fast he presses keys...

mv_pnickj_219000_m36b3.inp (regular MARP)
Bit  75, average length (in number of frames): 3.39
Bit  77, average length (in number of frames): 6.27
Bit  78, average length (in number of frames): 2.94
Bit  79, average length (in number of frames): 2.83



mv_t4-pnickj_131900_win35tg3.inp (correct speed as verified using InpSpeed)
Bit 187, average length (in number of frames): 7.04
Bit 189, average length (in number of frames): 8.43
Bit 190, average length (in number of frames): 5.43
Bit 191, average length (in number of frames): 5.31

About twice as slow all of a sudden? 'Nuff said.

Cheers, Ben Jos.

-- Anonymous, July 19, 2000

Answers

I was very suspect when mvdv had the 64 in the "incredible" miexchng recording he submited. I'm affraid to look into it further but Hisa Chan says he did not use autofire, but the "incredible" dspirit recording he submited shows extrememly fast pressing compared to other shooters he's uploaded... easily pushing 3 frame averages for incredible durations in dspirit. dspirit is a game that plays slow by nature (but usually can get 100% most of the time on my 350 P2) so it could be hisa isn't monitoring the framerate on a slower computer.

And what irks me the most is that having a tgmame leadboard or some exodus away from an insecure mame recordings would stop this from happening and we could be more certain to disqualify scores when we see some people uploading lower scores in tgmame but their non tgmame scores are lacking, which ben jos has done here. having that extra qualification of comparing a tgrecording with a nontgrecording, makes the uncertainty of the dq-ing a upload souly on framerate averages much more comfortably done.

oh another thing ben jos? Have you tested the InpSpeed with a m35tg3 recording that you alt-tab back to windows (to pause) to see what frame rate that frame that was "paused" gives? I'm curious if you can use m35tg3 to pause cheat with when using windows. I think i'm going to try this when i get home and see what frame rates the paused frames get. And i'm affraid the frame rate that gives will not be a very large value (which could be hidden by averaging) or perhaps overflow/wrap around into a good one depending on how long the pause was made. Something that should be engineered into the new tgmame not to wrap the framerate time value if it is very large...

-- Anonymous, July 19, 2000

cool, my suspects weren't waranted even in tgmame alt-tab pausing can be caught with the new inpSpeed C:\MAME>analinp -S -k inp\lbowling.inp
Analinp Version 2o

1 slow frame (average speed: 0%)
11 good frames (average speed: 100%)
12 slow frames (average speed: 88%)
1296 good frames (average speed: 100%)
12 slow frames (average speed: 2%)
204 good frames (average speed: 100%)
12 slow frames (average speed: 1%)
1404 good frames (average speed: 100%)
12 slow frames (average speed: 1%)
49 good frames (average speed: 101%)


-------- InpSpeed Report --------
Good frames : 2964 (average speed: 100%)
Slow frames : 49 (average speed: 23%)
Fast frames : 0 (average speed: 0%)

Total frames: 3013 (average speed: 99%)
-------- End Of InpSpeed Report --------

I paused three times, for variying durations, luckily the time isn't recorded but the frame rate is which just goes down to zero and doesn't wrap... The 88an

-- Anonymous, July 20, 2000

I DQed a mvdv gururin that showed 2 frame averages and incredibly quick decisions/moves made. However i confirmed a mvdv gururin which had a clean inp file, FEI. i believe the confirmed gururin of mvdv to be ok, but someone else might want to check it out.

-- Anonymous, July 27, 2000