The 6502 CPU has been decapped and
simulated at a low level, so we see exactly why
the unofficial instructions behave how they do through the compressed microcode and the post-processing. Likewise, the CIC microcontroller and its clone Rabbit have been decapped in the process of making
CIClone and
Super CIClone. Would it be worth it to do the same for the PPU, to help figure out the reason for other quirks such as power-up behavior or OAM refresh behavior? Or to do the same for the NES CPU, to see the details of the clock divider, DMA unit, sound unit, and the exact way that decimal mode was disabled?
Sure, go for it.
I mean, if we were able to find the sprite bug (disabling sprites early) recently-ish, then there could be other treasures waiting.
Worth it? I think so. If it were, it would be possible to remove the PPU from a console and drop in perhaps an FPGA board that is an exact copy of the PPU except for being able to output a higher quality signal like S-Video or RGB. And just in general it would be great if the exact NES circuitry were preserved forever.
If someone is able to understand the blueprints (I'm definitely not !) then of course it'd be worth it.
That way those who say "emulation may never be perfect" will be proofed wrong
(from a NES ROM demo that ironically works as it was supposed only under some emulators).
Also keep in mind that there might be slight difference between revisions, such as those weird OAM behavior.
I would love to know what actually happens inside the PPU! The APU should be interesting too, but I'd be more excited about the PPU.
I also think that this would be an awesome project. Unfortunately, I don't have the equipment nor the skills to assist.
Would it be possible to just ask Ricoh if they would publish what they have on these chips?
Do these chip designed contain any current market value or any undiscovered secrets that they would want to keep hidden?
I can donate PPUs, I don't have any extra CPUs though. Would be great to see it emulated, sounds like that's what the guys who did the 2600 chips wanted to do.
Would be interesting to look at mappers too, more out of curiosity though. It would be interesting to compare the PPU to the TI chip in the Coleco also.
I fully agree that this whole community would benefit from a complete disassembly and microanalysis of the NES's CPU and PPU. Somebody on here's gotta have access to this process.
It'd also be useful in the development of more interesting mappers that could exploit any uncovered quirks in addition to dicovering some of its flaws.
If anybody is qualified to do this and will dedicate the time, I can send samples of multiple PPU & CPU types and revisions (ditto on the other components that'd be useful to R-E).
I can try to extract stuff out of Ricoh and Sharp, but getting docs on 30-year old HW is tough, even when they are willing to release them.
I already need to wiggle some sttuff from Sharp and Mitsumi, so I'll see what I might achieve. I have some other video chip schematics from other consoles that people may want or might find interesting for private trade or gift.
I won't talk about what they are in detail in public, to save me from both harassment and possible legal issues from exposure. If you know this stuff inside-out and want anything, send me a PM.
Likewise if you want to undertake the PPU decap, PM me and we'll talk about it in detail.
I am the guy who did the die shot of the 2A03 seen in the visual6502 talk mentioned in the other thread, you can see all the material here:
http://uxul.org/~noname/chips/ (here is a 300MB image:
http://uxul.org/~noname/chips/stitch-devig-color/1.png). The CPUs destroyed in the process came from Memblers and hyarion. I lack the resources to do the polygon drawing by myself, though.
If anyone wants to send me PPUs to make die shots of, that's fine (PM me for my current address), but I cannot promise that I will find the time to do the decapping and shooting in the next few months. In any case, if I don't find the time personally, I will forward the PPUs to other people from the visual6502 project who may, but we already got a lot of other chips, so don't expect immediate results.
The 2A03 pictures look really great.
I didn't know this was on here too:
http://uxul.org/~noname/chips/rp2c02-0-9g5-14/stitched/
Text on it is:
G2C02
(C) RICOH 1986
So, is this really the PPU? I'm really curious to see what the PPU does internally. There are so many details that could be cleared up by observing these chips like this...
Yes it is it... Look at the markings at the top
Anyone want to donate one of each MMC to the decap effort so that we can read off the real netlists? Imagine 100% accurate MMC5.
tepples wrote:
Anyone want to donate one of each MMC to the decap effort so that we can read off the real netlists? Imagine 100% accurate MMC5.
What is a "netlist" in the context of a mapper?
A
netlist is a low-level description of all the gates in a circuit and how they are connected, written in a simple HDL that resembles Boolean algebra. One might construct this by decapping a mapper, taking a photomicrograph, and identifying gates. (Lately, Quietust has been
reading the PPU and APU photos to discover not-yet-documented behaviors, but at a higher level than netlists.) The resulting netlist can be synthesized into a CPLD, FPGA, or ASIC, allowing accurate clones of more complex mappers like MMC3, RAMBO-1, FME-7, later VRCs, and MMC5.
tepples wrote:
Anyone want to donate one of each MMC to the decap effort so that we can read off the real netlists? Imagine 100% accurate MMC5.
Sure i can donate some ICs, that is if you find anyone willing to decap them and share the result
If this is possible, then would it also be possible to decap the PPU from a Vs. machine and recreate it to make RGB NES mods easier and cheaper.
marvelus10 wrote:
If this is possible, then would it also be possible to decap the PPU from a Vs. machine and recreate it to make RGB NES mods easier and cheaper.
The normal PPU is just as effective for this if not moreso.
[/quote]The normal PPU is just as effective for this if not moreso.[/quote]
Maybe I got this wrong, but I was under the impression the NES PPU could not output RGB, therefore replacing it with one from the Vs. machine allows for that.
Or is it a matter of the NES PPU doesnt have the output externally but is quite capable internally if it was reproduced with the right outputs.
marvelus10 wrote:
If this is possible, then would it also be possible to decap the PPU from a Vs. machine and recreate it to make RGB NES mods easier and cheaper.
Decapping a Vs PPU will not let us make new RGB PPUs. It will only let us verify that our external measurements as to how its output differs from a composite-only PPU are correct.
We also already have a decapped composite-only PPU and know how to (with expensive hardware) convert its output to RGB.
HardWareMan can donate UMC's
UA6527P/6538 chips for decapsulation: Worth it?