TGM1 Gm requirements

[rednefed]

Was playing a few games of TGM1 for fun, trying to get a very high score. Yeah, I had a game with 228000 points but that's not what was interesting.
I had S2 or S3 at level 300, and was very close to S6 at 500, finished three times with scores all over 170000 and times under 13:30 (but over 13:00), and still was not awarded GM. Just to make sure the game wasn't broken I played a game regular-style, and comfortably finished in 11:50-odd with the Gm grade.

Does this mean TGM1 has a tetris requirement for Gm? I am absolutely sure I made every checkpoint (1 @ 300, S4 @ 500, S9 126000 and <13:30 @ 999). Or is there a time requirement earlier in the game that you can silently fail?

My play strategy was to leave column 9 five blocks high, get a tetris, and combo off of it for score. However, because I wanted to avoid holes and overhangs at all costs, I skimmed a lot and didn't actually land that many tetrises, especially in 20G. I would guess that I made something like 8-10 over the course of the game.

Please investigate: whether a game that otherwise fulfills all the requirements for GM, without making many tetrises, can actually be awarded the grade. I have a niggling feeling that it's time-related (over 13 minutes but below 13:30) and not at all related to the number of tetrises made, but we've been surprised before.

[TWF]

This is very interesting because I was planning on trying to do a no-tetris Gm play. If someone made a quick TAS of that it would be interesting to see.

Do you have a screenshot of the game you didn't get Gm in?

[colour_thief]

Forget TAS'ing, I'll give it a try when I get home. Bet I can pull it off first try.

[rednefed]

Played a game trying to do the same strategy. Made about 6 tetrises during the whole thing and got GM.
300 @ 4:14:65 (S3, 33132)
500 @ 6:54:33 (S4, 51284)
999 @ 13:14:16 (GM)

Edit: It might be < 7 minutes at level 500. I played intentionally slow 0-500 to get over 7 minutes, played 500-800 at my normal speed, and slowed down 900 to get over 13 minutes.
These were from the same game.
http://defender.arcimpulse.com/TC/tgm01.jpg
At level 300: S2, 4:30:58

http://defender.arcimpulse.com/TC/tgm02.jpg
At level 500: S5, 7:11:41

http://defender.arcimpulse.com/TC/tgm03.jpg
Sometime after I hit S9: 917 @ 11:55:66

http://defender.arcimpulse.com/TC/tgm00.jpg
At level 999: 13:01:33

[colour_thief]

I think you might be on to something... I played a game with 1 tetris (oops) with S9 and 999 at ~12:30, but the first half was over 7 minutes.

[jujube]

i can't believe nobody found this before. very cool.

[colour_thief]

For what it's worth, I just got Gm without any tetrises in 11:17:65. No serious comboing either.

[tepples]

For what it's worth, I just got Gm without any tetrises
Not even the one on the title screen? So is "Triple the Grand Master" anything like Double the Killer (http://en.wikipedia.org/wiki/Windows_Speech_Recognition#Notable_incident) (delete select all)? http://www.tetrisconcept.net/forum/images/smilies/icon_wink.gif

But seriously, good job on figuring that out. It makes it even more likely that we're likely dealing with a hidden torikan. But we'll probably need TGM1 GMs with especially fast 20G in order to narrow down the exact thresholds. And that'd probably include you, c_t, as you're GM in T.A. Death.

[Ai]

So most likely one must complete the first half in under 7 minutes. It's surprising this wasn't noticed before, though most people reaching the end with S9 are fast enough to overcome this obstacle. Nice find rednefed!

Does anyone remember this?

http://img393.imageshack.us/img393/4007/tgm1kp4.png

How is this not GM?
viewtopic.php?p=21414#p21414 (http://www.tetrisconcept.net/forum/viewtopic.php?p=21414#p21414)

Maybe he wasn't lying after all. At that time...

rofl!!! ^^

EDIT:

Though...

Wait, what the-- I checked and it showed up in the rankings as GM. WTH!!!!!!!!!!!!!!???????????
Does it show as GM in the Grandmaster Ranking or Todays Ranking? I doubt it...

[rednefed]

Nope, shows as S9. Ghetto's a liar, whodathunkit? http://www.tetrisconcept.net/forum/images/smilies/icon_exclaim.gif

[Amnesia]

Shall I add this 7min Torikan as an official condition ? Or do you prefer continu some tests to really be sure ?

[Ai]

Shall I add this 7min Torikan as an official condition ? Or do you prefer continu some tests to really be sure ?
As far as I know we're not certain that it's exactly 7 minutes, but I think it's safe to point out to the players that they should aim to finish the first part in under 7 minutes.

Nope, shows as S9. Ghetto's a liar, whodathunkit? http://www.tetrisconcept.net/forum/images/smilies/icon_exclaim.gif
Whatever shred of credibility Ghetto still had is now lost. Ah Ghetto! Why? ^^

[colour_thief]

Shall I add this 7min Torikan as an official condition ? Or do you prefer continu some tests to really be sure ?
As far as I know we're not certain that it's exactly 7 minutes, but I think it's safe to point out to the players that they should aim to finish the first part in under 7 minutes.

Well for what it's worth I was convinced enough to edit our wiki. We have 7:0x failing and 6:5x passing. We won't get more accurate without disassembly and the series has a history of choosing round numbers so Occam's razor points to it being 7:00.

[jujube]

either way, the article is more accurate now than it was before.

[tepples]

Conspicuous by its absence is a time in the 300 row. So I guess the next step for research is to improve your 300-500 time attack, so that we can see if there's another torikan at 300 like there is at 500.

[DeHackEd]

S3 at 300
S5 and ~7:20 at Lv500
161,270 in 12:42.21 at Lv999

Try again to be a grand master.

[rednefed]

Conspicuous by its absence is a time in the 300 row. So I guess the next step for research is to improve your 300-500 time attack, so that we can see if there's another torikan at 300 like there is at 500.
I just tried for 5 minutes at 300 but, despite my best play, had no chance of making 500 in under 7 minutes. Someone with a less conservative 2-5G could give that a try.
Maybe 4:30.

[Quad]

Speaking of disassembly, how does that work with arcade ROMs? Is there any good on-line reference with opcodes and specs for how the ROMs actually work so that a clueless person like me could give it a try? I'm not really interested in actually disassembling a ROM but I'd like to know how it works, I think disassembly in general is a fascinating subject!

Is disassembly how you guys collected some of the more obscure and hard-to-verify information on the Wiki?

[tepples]

Speaking of disassembly, how does that work with arcade ROMs? Is there any good on-line reference with opcodes and specs for how the ROMs actually work so that a clueless person like me could give it a try?
Games that run in ePSXe or Zinc use the MIPS R3000 CPU. Arcade games are likely to have the code encrypted in the ROM, but if your emulator can decrypt the code to run it, it can probably dump the decrypted code to a file so that you can run it through a MIPS disassember. Might the Binutils in devkitPSP (http://www.devkitpro.org/default.htm) have a MIPS disassembler?

Another pain in the behind, if you're planning on doing things the legit "I wanna sell something" way, is getting the data from the arcade PCB to the PC.

[colour_thief]

Yeah a lot of stuff was found through disassembly. But there are also exceptions like when I reverse engineered Ti's grading with repeated tests on my actual machine. I didn't get precise cool/regret requirements or credit roll grading, but I'm damn proud of what I did find. Especially that the comboing bug was fixed in the TAP portion of the grading.

You don't just open up the roms in a hex editor and start disassembling. I mean, there's no reason you couldn't, but there are much, much better tools for it. When people disassemble code for these games they use MAME's debugger. You don't really need to know how the ROMs are organised. Typically the process is using debugger commands to search the (emulated system's) memory for particular values, like say the level, while the game is actually running. Repeated searches as the value changes narrows your results by eliminating coincidences. Next you can use debugger commands to pause the execution when that area of memory is read and/or written to. All the sudden BAM you're in a function that actually does something tangible. As far as making sense of the function, opcodes can be looked up using official documentation freely available online, or you could even use MAME's source as a reference.

Also tepples are you sure that "encrypted" is the right word? I mean a few games do use encryption as a security measure but often times the code is sitting on the ROMs chips in direct machine-readable form.

[DeHackEd]

My way is be something as follows:

First determine where the game saves your Gm qualification data. You should be able to build a cheat that guarantees a Gm result even if you fail the 300 or 500 criteria.

Next, modify the emulator (if it does not already support it) to do running disassembly. Snes9x does this if you set the right fields. Basically in the CPU emulation code, print out every instruction as you execute it. Memory addresses and CPU registers a bonus.

Finally, let the game run normally but set watch points on the Gm qualification fields. The emulator should freeze if the data is read from or written to. When the game freezes, the running disassembly should show the instructions leading up to the freeze and they'll be exactly the game noting data or just reading it (999?). If necessary, instruction-step and watch what it does.

I used this strategy to figure out the random encounter formula for Final Fantasy 6. Maybe I'll see what I can do tomorrow.

[tepples]

Also tepples are you sure that "encrypted" is the right word? I mean a few games do use encryption as a security measure but often times the code is sitting on the ROMs chips in direct machine-readable form.
I had assumed that measures like those in the CPS-2 had become widespread.

[Amnesia]

viewtopic.php?f=2&t=539&p=8817#p8817 (http://www.tetrisconcept.net/forum/viewtopic.php?f=2&t=539&p=8817#p8817)

I fixed everything, even my personnal recommandations for the section time steps, because I always suggered to young padawans to be at 500 @7:05 http://www.tetrisconcept.net/forum/images/smilies/icon_eek.gif

I would bet that it is 7:00 like everybody, but why not 7:02 ??
Nobody can explain why SHIRASE 500 @ 2:28 and not 2:25 or 2:30.. http://www.tetrisconcept.net/forum/images/smilies/icon_rolleyes.gif

[TWF]

http://www.tetrisconcept.com/forum/viewtopic.php?f=2&t=539&p=8817#p8817

I fixed everything, even my personnal recommandations for the section time steps, because I always suggered to young padawans to be at 500 @7:05 http://www.tetrisconcept.net/forum/images/smilies/icon_eek.gif

I would bet that it is 7:00 like everybody, but why not 7:02 ??
Nobody can explain why SHIRASE 500 @ 2:28 and not 2:25 or 2:30.. http://www.tetrisconcept.net/forum/images/smilies/icon_rolleyes.gif
He has a point. Someone will need to try breaking 500 in less than a second over 7:00... or disassembly could find the value too.

[Zircean]

EDIT: Edo and Billmaan jointly have found that the 500 requirement is 7:30.00. Also, Edo has confirmed that there is, in fact, a 300 time requirement. It is 4:15.00.

This has been confirmed by looking at the memory. At this location is 3BC4 (4:15 in hex) and 6978 (7:30). An interesting note is that 777 had made these observations before (which can be found here (http://ameblo.jp/eiji777xyz1105/theme-10014501653.html)).

[rednefed]

Heh, looks like I actually failed the 4:15 torikan, not the 7:30 one when I started this topic. Can't disagree with disassembly.