A new die is up at http://cs.sipr0n.org ! This is another member of the TGP series on the SEGA Model 1 hardware. There will be a few more of these, so don't give up, as the monkeys are already proving to be worth their weight in bananas! Check out some of the things you've accomplished:
Each of these dies had been previously converted to bits by a combination of an automatic algorithm and hand-validation. Take a look at what the typing monkey project helped us accomplish:
The first die typed, sega_315-5571_xpol, confirmed the previous effort was 100% correct. This may seem like a bummer, but we now have 6 separate groups of people confirming the dump is correct. This is *very good* for those working with this data, as confidence like this is essential to moving forward.
The second die typed by the monkeys, the sega_315-5572_xpol, fixed 8 incorrect bits in the previous typing! Huge.
The third die monkey-typed, the sega_315-5573_xpol fixed 10 incorrect bits in the previous typing! Huger!
The fourth die, the sega_315-5677_xpol, fixed 7 bits(!), and confirmed that there are ~4 bits that are too difficult to see. This greatly helps the person taking the photos of the dies narrow in on where to take better pictures and fix the dump for good.
So there you have it. The monkeys have typed their way into the hearts of dumpers everywhere, and are already making a difference. The TGP emulation work in MAME will be slow and arduous, but it can move forward with confidence, knowing that the dumps very high quality.
> "it was the best of times it was the blurst of times" (ie: Hopefully what a thousand > monkeys can accomplish without any typos) > > https://www.youtube.com/watch?v=no_elVGGgW8 > > > > as the monkeys are already proving to be worth their weight in bananas! > > Nice to see results coming along.
> The fourth die, the sega_315-5677_xpol, fixed 7 bits(!), and confirmed that there are > ~4 bits that are too difficult to see. This greatly helps the person taking the > photos of the dies narrow in on where to take better pictures and fix the dump for > good.
Hope to see some examples of the difficult spots and the correct answers for them, these could be helpful for the future typings. Like, http://cs.sipr0n.org/static/sega_315-5678_xpol/sega_315-5678_xpol_23_27.png - row 8, column 3 - is it 0 or 1? Too bright to be 0, but no bright "circle" in the middle to be 1.
> Hope to see some examples of the difficult spots and the correct answers for them, > these could be helpful for the future typings. Like, > http://cs.sipr0n.org/static/sega_315-5678_xpol/sega_315-5678_xpol_23_27.png - row 8, > column 3 - is it 0 or 1? Too bright to be 0, but no bright "circle" in the middle to > be 1.
I'd like to know more about what causes these difficult spots so that they can be better mitigated, if at all possible.
If I recall correctly, the steps for decapping a chip are:
1. A small amount of the package is ground off in order to provide a divot for the fuming nitric acid. 2. A small amount (just a few drops, possibly even less) of fuming nitric acid is placed in the divot. 3. After a rather short amount of time, the fuming nitric acid is neutralized with acetone or some other alkaline material. 4. Return to step 2, until the die is sufficiently exposed for imaging.
I seem to recall reading somewhere that fuming nitric acid doesn't eat away silicon or precious metals like gold, but regular nitric acid can and does, and moreover, fuming nitric acid turns into regular nitric acid upon reacting with water in the air. Thus if the chip is allowed to process too long, it could damage the silicon die itself, and if it's not allowed to process long enough, there could still be flecks of the package left on the die.
Is that the source of these problem spots in this case, or is there some additional potential source of defects in the photos?
IDrinkHF will know much more about this, but it has been suggested that some of the "too hard to tell" bits are due to "damage on die." The bits seem to be recoverable with some additional work though, so having the monkeys point out the damage is as important as a perfect typing.
Hopefully a better explanation will come at some point, 'cuz that's all I know! /Andrew
> IDrinkHF will know much more about this, but it has been suggested that some of the > "too hard to tell" bits are due to "damage on die." The bits seem to be recoverable > with some additional work though, so having the monkeys point out the damage is as > important as a perfect typing. > > Hopefully a better explanation will come at some point, 'cuz that's all I know! > /Andrew
It's a bit better, but I have no idea what "damage on (sic) die" means. Off the top of my head I wonder if it could be physical degradation due to time, or if it's physical degradation due to the way it was decapped, or any number of things.