Thursday, February 21, 2013

It's done!

I could put in many, ways, but the simpliest one is to see the video below:

The Isos should be floating around before the end of the month.

Tuesday, February 19, 2013

Hacking from Dammam

Unfortunately I'm on a business trip and couldn't bring with me the real thing.

I identified together with Joerg Hartenberger, a new set of interrupts not emulated by MAME that happen in the real thing. This is most probably where the error is coming from.

Bottom line, as soon as  I can grab my setup back home, I'll test it on the real thing and I'm quite confident it's going to work. If that's the case, before the end of the month all tables for all games can be released...

Stay tuned.

Sunday, February 10, 2013

The Hartenberger effect

Looking at the title, you may thing I'm talking about a film about secret weapons in the second world war, but no.

I've been regularly discussing with Joerg Hartenberger, a Mame expert and creator of the site: with lots of interesting hacks that has been very supportive and has given me many ideas, so I wanted to make sure his support was recognised.

The status of the tables for the tool is the following:

As I mentioned before, I'm able to create an ISO that will change any system with battery free SFIII-2 to any other game and that works great on MAME....but not on the real thing.

In order to locate where the problem was I've been working the last weeks in an interesting approach, which is the following:

I create a hacked version of a game and after updating the SIMMs it will run code from the game ONLY until instruction X where I force it to JUMP to the BIOS to show something on the screen (the Game Update menu for instance). If we can see the BIOS code, it means that my hacked version was able to run code until adress X, so I will go further some more thousands of instruccions and check again with a new CD if the hacked version is able to arrive now to address Y. If it doesn't then I have to look for a new address Z somewhere between X and Y.

This might sound like something that can be sorted out with 2-3 tries, but the reality is that after 20 tries (and 20 CDs burnt) I have managed to narrow it down to only 4000 instructions executed.

This is now much easier to read and handle (vs a 500Mb file created with an execution dump). You can get an idea of the process by starting mame in debugging mode ( -debug )and type in any moment the command "trace", the game will run about 20-30 times slower than usual and will generate a file of hundreds of Mb.

I'm analysing the code and I hope to have some conclusions soon, but as I travel very frequently it might still take some weeks as I can't take my MAK and CPS3 real board with me :((.

Stay tuned for more :)