Haze says: No-Intro is wrong!

Post by **root** » 26 Aug 2006 11:51

Rif: 790-1 Haze says: No-Intro is wrong! \ alcoatjez on 26th August 2006, 12:51 wrote:

It's not that bad, but there is a very interesting discussion on the MAMEworld Forums in this thread: http://www.mameworld.info/ubbthreads/sh ... 1156513423
IMHO it does matter, but it will be almost impossible, since it means redumping ALL ROMs by screwing your carts open.

===============================================

Re: Haze says: No-Intro is wrong! \ Yakushi~Kabuto on 26th August 2006, 16:30 wrote:

Not only No-Intro but goodxxx as well. Your title is wrong too

MESSdevs reply:

Haze

I don't trust any of the databases.

To my knowledge none of them document REAL dumps (how many rom chips in the cart, the labels on the rom chips, all Sega In-house games should have official EPR labels afaik) Also as well as being merged I'm pretty sure every single Genesis rom has been byteswapped so that all the ASCII is readable compared to a normal dump. (Every *real* dump Guru has done with proper tools has required byteswapping on loading, and had non readable Ascii)

There is no such thing as a 'good' database of Megadrive of SMS dumps at this time. It would be nice if people worked towards creating one, but most console emu users don't care about doing things the right way which for me as a Mame developer is very, very annoying.

Bart Trzynadlowski

> IMO those who don't care aren't necessarily lamers/kiddies, it's just that 'under the
> hood' rigourous hardware documentation isn't the be-all-end-all for everyone who is
> involved with emulation. Actually I'd say those people are very much in the fringe
> group. Most really are in it just for fun.

It's not even really a matter of rigorous hardware documentation, it's a matter of where to draw the line on what needs and needs not be documented. Consoles see cartridges as a part of their address space -- it therefore makes perfect sense to dump them as images that way. All the game data is preserved.

A database of the physical construction of various games would be fine (although useless to just about everyone, including me), but it would have to take into account that games were produced differently throughout their lifespan. I believe Snorter pointed out that some games originally came on multiple ROM chips and were later moved over to a single ROM to lower costs. Either way, the Genesis didn't care.

As far as I'm concerned, a console emulator is supposed to emulate the console. It ought to run existing games and new homebrew software. In this context, a flat binary dump for the Genesis makes the most sense. If games are to be split into files corresponding to physical ROM chips, then there's going to have to be an additional control file specifying how to load the ROMs thereby creating more problems to deal with.

I think flat binary dumps for Genesis games are a technically superior solution: They take advantage of the fact that files can store an arbitrary amount of data (remember: computer file != ROM chip), they are easier to examine and modify because they are laid out as the developer and the actual hardware would see them, and, if necessary, they can be decomposed into files ready to be loaded onto ROMs under the control of external data not needed by emulators to perform their jobs.

An external database is arguably better for documentation purposes.

Neither Haze's approach nor the existing approach are more "correct." I just think it's completely absurd to care about something as arbitrary as how many ROM chips appear in a cartridge (which is itself the discrete medium of game data since the console "ends" at the cartridge connector) and what their labels are while the code that actually emulates the system is a complete misrepresentation of how the system actually works. Sure, emulators may produce pixel-perfect results, but they can rarely help you deduce how the actual hardware is structured or how it works because the emulator authors don't know or care about that.

R. Belmont

I'm with you Bart, and personally I think UNIF on the NES gets it exactly right (the data's in an easy-to-emulate format *and* you have all the data about the individual chips) but this is a religion for Haze, not a technical matter. I'd advise you to just stay out of the way.

===============================================

Re: Haze says: No-Intro is wrong! \ ElBarto on 26th August 2006, 18:15 wrote:

This mother f****r is right and this is the goal of my secret project, I can tell it now, need to be faster than him now

Well exactlly he is not fully right.
I explain :
I'm currently writing a multi emulator, based on module (dll) for each system and each CPU. I've started with Mega Drive (because I know wel this system). I've never see that before but on actual emu (even Kega, the best MD emu) you can't emulate properly a game genie or sonic and knuckles, on mine you could cause I emulate this PCB of each cart. I'm currently fill my database of cart with chip info and other things for each rom.
More new later.

===============================================

Re: Haze says: No-Intro is wrong! \ Yakushi~Kabuto on 26th August 2006, 18:21 wrote:

You are awesome Elbarto!!

On a not related note, please fix the Mega Drive dat

===============================================

Re: Haze says: No-Intro is wrong! \ geezup on 27th August 2006, 11:02 wrote:

that sounds great ElBarto !

===============================================

Re: Haze says: No-Intro is wrong! \ etabeta on 27th August 2006, 17:06 wrote:

@elbarto: it would be interesting a cooperation with haze, don't you agree? hazemd0.05 starts to add 2 'proper' dumps (nhl 94 & robocod) and it would be wonderful to increase this number

@alco: i always thought that no-intro goal was to identify best possible dump already existent (that usually meant the good & proper dump), not to introduce better way to dump & preserve images (even if a natural implication of nointro goal is that when a better format emerges it should be considered as primary against the older one)... so haze's point is more: nobody (mainly referred to dumpers) is trying to preserve correctly the old carts more than a direct attack to tools authors

my 2 cents

===============================================

Re: Haze says: No-Intro is wrong! \ NGEfreak on 27th August 2006, 19:26 wrote:

Sounds interesting, but how does he plan to address the following issue? Include the same ROM info twice?

I believe Snorter pointed out that some games originally came on multiple ROM chips and were later moved over to a single ROM to lower costs.

===============================================

Re: Haze says: No-Intro is wrong! \ Yakushi~Kabuto on 27th August 2006, 19:32 wrote:

Sounds interesting, but how does he plan to address the following issue? Include the same ROM info twice?

such things happened in arcade PCBs too
MAME tags them as (set1) , (set2) ,...

===============================================

Re: Haze says: No-Intro is wrong! \ KingHanco on 27th August 2006, 21:46 wrote:

HazeMD 0.05a is release.

So what you going to do? Try it or leave it alone?

KillerBee.

===============================================

Re: Haze says: No-Intro is wrong! \ alcoatjez on 28th August 2006, 09:39 wrote:

@etabeta: You are right. I just wanted to start the discussion

===============================================

Re: Haze says: No-Intro is wrong! \ Yakushi~Kabuto on 28th August 2006, 15:44 wrote:

My humble opinion.

Haze idea isn't new at all since we @no-intro have discussed the matter with emu authors for ages. And that's simply what MAME already does for ages.

But what exactly does he intend to do? Answer: documenting ROM chips.
This isn't related to emulation at all and that's why emu authors never bothered so far.

More importantly, redumping everything from scratch is considered utopian by MESSdevs. Hundreds/Thousands of dumped carts are most likely not available anymore for documentation purpose.

Conclusion: Haze idea is certainly not wrong, but it isn't new either, it isn't that important either and most of all is utopian.

===============================================

Re: Haze says: No-Intro is wrong! \ ElBarto on 30th August 2006, 07:56 wrote:

@elbarto: it would be interesting a cooperation with haze, don't you agree? hazemd0.05 starts to add 2 'proper' dumps (nhl 94 & robocod) and it would be wonderful to increase this number

I contact him 2 days ago, still no response.

===============================================

Re: Haze says: No-Intro is wrong! \ ssjkakaroto on 30th August 2006, 18:00 wrote:

I didn't understand why the proper dumps of those 2 games only had 1 file each, if Haze said that all chips of a game should be dumped individually, in the same fashion of arcade dumps.
Do the Genesis cartridges only have 1 chip?

Thanks

===============================================

Re: Haze says: No-Intro is wrong! \ hydr0x on 1st September 2006, 16:45 wrote:

I didn't understand why the proper dumps of those 2 games only had 1 file each, if Haze said that all chips of a game should be dumped individually, in the same fashion of arcade dumps.
Do the Genesis cartridges only have 1 chip?

Thanks

i don't know anything about the technical aspects of genesis carts but it's not uncommon that only 1 rom is used on a pcb/cart

Post by **root** » 13 Nov 2007 14:37

Rif: 790-2 Re: Haze says: No-Intro is wrong! \ gigadeath on 13th November 2007, 14:37 wrote:

I just read this thread, I was searching Google for "MD proper dumps", since I vaguely recalled Haze saying something about them.

Now I have a gamebit to open the carts (needed to read the serials), so in the last few days I tried collecting info about proper chip-by-chip dumping, using a real programmer.

Do you all wanna know why YK says a proper dumping project is pure utopia (and rightly so)? Because a programmer that supports MD maskroms would cost me 300€.

===============================================

Re: Haze says: No-Intro is wrong! \ denzilla on 13th November 2007, 22:06 wrote:

What a killjoy

I assume a tool could be coded to separate the data once the rom chips are cataloged?

===============================================

Re: Haze says: No-Intro is wrong! \ mikers on 14th November 2007, 05:45 wrote:

It could, but you'd have to know where the individual ROMs are (afaik).

===============================================

Re: Haze says: No-Intro is wrong! \ Yakushi~Kabuto on 14th November 2007, 11:17 wrote:

As our media database grows, we will eventually be able to achieve the goal of spliting ROMs as they are in the PCB. An eprom reader is not necessary, as we can read the ROM size on the chips and split accordingly. We also know in most cases, the byte arrangement, (MD ROMs are byte swapped for instance).

Retrospectively I must say:
Haze says: No-Intro is wrong!
No-Intro says: Who is this guy?

Does a guy stealing code from his fellow devs and crediting himself for it, is in position to distribute good and bad points to the communities, hardly. Instead of showing off, he should start making some real contributions to the community. Fact, Haze is not able to code from scratch unlike the devs he stole code from to make him a reput. I couldn't care less what Haze has to say about No-Intro.

===============================================

Re: Haze says: No-Intro is wrong! \ tetsuo55 on 14th November 2007, 11:44 wrote:

thw worst part is:

Only a few carts would have to be dumped at first, using this data we could probably fix every other dump.

However if this was not possible, everything would indeed need to be redumped using this system

Can the dumping system be used for other systems than just MD?

===============================================

Re: Haze says: No-Intro is wrong! \ denzilla on 14th November 2007, 11:48 wrote:

My personal feelings are that data is data and as long as its backed up/verfied, who cares how many rom chips were required to store it originally. Its not like someone taking a rom 20 years from now is going use 4 eprom chips to construct one cart if he can fit it all on one chip. Would this also mean that all emus would have to be recoded to load the new rom storage type?

===============================================

Re: Haze says: No-Intro is wrong! \ FitzRoy on 18th November 2007, 02:39 wrote:

It doesn't matter if data which could be stored on a single rom chip gets split into multiple chips on some carts. It is totally superfluous and redundant to try and store roms in a way that mimics the physical separation of data which could exist or function in one chip. NES data, on the other hand, has separate chips with different types of data that should not be combined in its rom format. The only reason they function in emulators is because a subjective header format is attached to them which tells the emulator where the relevant data for each chip begins and ends in the file. Most arcade games are the same way and have different chips for different types of data, but are correctly stored.

I would be somewhat concerned about the byteswapping thing if he is correct about that. If original data is being converted to something else in the dumping process, then that is bad.

===============================================

Re: Haze says: No-Intro is wrong! \ ElBarto on 18th November 2007, 14:27 wrote:

In fact it matter if the data was store originaly on diferent chip. Take the NES for exemple.

- Say we have a game release on Europe, in Germany and France
- Both of the release is exclusive to the country (all in French and German respectively)
- It could be possible that the PRG rom is the same on both but the CHR one is different right ? Why would the have redone some code of the game if just the text differs ?

So in that exemple we have a duplicate PRG rom.
I've not seen that already but it could be possible.
Hope you understand what I mean.

===============================================

Re: Haze says: No-Intro is wrong! \ Haze on 19th November 2007, 13:45 wrote:

As our media database grows, we will eventually be able to achieve the goal of spliting ROMs as they are in the PCB. An eprom reader is not necessary, as we can read the ROM size on the chips and split accordingly. We also know in most cases, the byte arrangement, (MD ROMs are byte swapped for instance).

Retrospectively I must say:
Haze says: No-Intro is wrong!
No-Intro says: Who is this guy?

Does a guy stealing code from his fellow devs and crediting himself for it, is in position to distribute good and bad points to the communities, hardly. Instead of showing off, he should start making some real contributions to the community. Fact, Haze is not able to code from scratch unlike the devs he stole code from to make him a reput. I couldn't care less what Haze has to say about No-Intro.

stop talking out of your ass.

If you dump a MD rom with the proper tools the roms are byteswapped compared to the roms that current Megadrive emulators expect. This is common for pretty much all 16-bit systems. The text in the roms should not be human readable without futher byteswapping. I'm not saying proper tools are cheap, but for MAME we expect dumps done with proper tools.

I've not stolen any code and credited myself with it, so stop talking complete shit about me behind my back. HazeMD is based on MAME because it's a side-project of me improving the Megadrive emulation in MAME, writing a standalone emulator would be completely pointless for that. Nowhere do I say that it isn't based on MAME. I also developed it out of interest in emulating a large number of the pirate carts that nothing else emulates correctly (because they don't handle the protection devices and require hacked roms.) Compatibility is high, it correctly runs a good number of games that nothing else does, and I'm happy with it. It's also open source, and everything I've discovered is free for anybody else to use, the same can't be said for several of your other beloved console emulators, whereby they think that keeping the information to themselves somehow makes them special.

I've been working on MAME for a long time, I've emulated a large number of games, I'm in close contact with a lot of people with professional tools, I know exactly what I'm doing and exactly what I'm saying, and I'll stick by my statement that the console scene and their methodologies suck.

===============================================

Re: Haze says: No-Intro is wrong! \ Yakushi~Kabuto on 19th November 2007, 14:55 wrote:

In fact it matter if the data was store originaly on diferent chip. Take the NES for exemple.

- Say we have a game release on Europe, in Germany and France
- Both of the release is exclusive to the country (all in French and German respectively)
- It could be possible that the PRG rom is the same on both but the CHR one is different right ? Why would the have redone some code of the game if just the text differs ?

So in that exemple we have a duplicate PRG rom.
I've not seen that already but it could be possible.
Hope you understand what I mean.

Unlike you I'm not an expert in MD hardware, since as you should know, I'm only into SNES and FDS. Do MD ROM chips have their size printed on them, much like Nintendo ROM chips. If that's the case, do we really need an eprom reader? Would the result be different if we byteswap and split the current roms?

I think all we need are the scans of the PCB.

===============================================

Re: Haze says: No-Intro is wrong! \ gigadeath on 19th November 2007, 16:27 wrote:

OMG Haze here!

As I stated, the problem is not the will, it's the money. I'd redump all my carts in a flash with proper equipment, no matter if I've already dumped them other ways, but I don't want to spend an extra 300€ for a piece of equipment which would be useless to me for everything else, while I can play my MegaCD AND dump with it.

It's only an economic issue, sorry.

===============================================

Re: Haze says: No-Intro is wrong! \ Yakushi~Kabuto on 19th November 2007, 20:01 wrote:

After reading Bart Trzynadlowski post again, the question of whether dumps should be considered as individual chip data or the data that ends in the pin connectors is still relevant.

Take the example of Final Fantasy VI (Japan)
It can be 3x8 Mib chips or 16+8 Mib chips.
But either way, the data ending in the cartridge pin connectors is still the same.
The console hardware is reading the cartridge pin connectors, so the console hardware doesn't make a difference whether there are 2 or 3 ROM chips. Therefore having a single dump zipped (as opposed to multiple dumps in a zip) makes perfect sense for accurate emulation.

Having a driver to map the data chip by chip is totally unaccurate.

Sorry but after much thoughts, Haze is wrong. Splitting the dumps isn't going to help proper emulation.

===============================================

Re: Haze says: No-Intro is wrong! \ FitzRoy on 20th November 2007, 01:35 wrote:

In fact it matter if the data was store originaly on diferent chip. Take the NES for exemple.

- Say we have a game release on Europe, in Germany and France
- Both of the release is exclusive to the country (all in French and German respectively)
- It could be possible that the PRG rom is the same on both but the CHR one is different right ? Why would the have redone some code of the game if just the text differs ?

So in that exemple we have a duplicate PRG rom.
I've not seen that already but it could be possible.
Hope you understand what I mean.

Yeah, that's a good point and in that scenario I would have to go with the simplest method of simply adding it to the name. My conventions are different than yours, so don't mind the serials. Fictional example:

Monkey Boy (NES-MB-EEC, NES-MB-NOE).chr
Monkey Boy (NES-MB-EEC).prg
Monkey Boy (NES-MB-NOE).prg

===============================================

Re: Haze says: No-Intro is wrong! \ sibio on 20th November 2007, 04:49 wrote:

In my opinion it would be useful to documentation purpose, to the final emulation result it doesn't matter too much...but having the split roms gives the idea of how the PCB is built and the rom chips labeled.
Certainly it would be a very long, time consuming and costly work to document everything to the smallest detail but for purists it would be the better idea of emulation community ever.
Haze idea IS NOT wrong, it's only a bit extreme but understandable and fascinating somehow (the more I think about it, the more I like it

)

No-intro could open a paypal donation account to help the best dumpers to acquire rare carts and dumping equipment. I would donate a few bucks!

===============================================

Re: Haze says: No-Intro is wrong! \ ElBarto on 20th November 2007, 10:12 wrote:

Unlike you I'm not an expert in MD hardware, since as you should know, I'm only into SNES and FDS. Do MD ROM chips have their size printed on them, much like Nintendo ROM chips. If that's the case, do we really need an eprom reader? Would the result be different if we byteswap and split the current roms?

I think all we need are the scans of the PCB.

No they don't have size printed on.
We do not need PCB scan, this is useless (I mean we can trust each other in what a dumper see on a cartridge).
For MD we need two thing :
- The ROM(s) Chip MPR code (for naming, revision etc ...)
- The cart PCB revision (171-XXXX) because it's with this that you can know how the multi rom on the PCB are "connected"to each others. This is very usefull for games with saves, lock on cartridge etc ...

Post by **root** » 20 Nov 2007 10:26

Rif: 790-3 Re: Haze says: No-Intro is wrong! \ ElBarto on 20th November 2007, 10:26 wrote:

stop talking out of your ass.

If you dump a MD rom with the proper tools the roms are byteswapped compared to the roms that current Megadrive emulators expect. This is common for pretty much all 16-bit systems. The text in the roms should not be human readable without futher byteswapping. I'm not saying proper tools are cheap, but for MAME we expect dumps done with proper tools.

I think we all totally agree with you about this point.
The fact is the only reason that we've not switched to byteswapped rom is because that your emulator is the only one who support that and only for the roms that you note on the source code (which for me is the only default of MAME, not having an external game list but that's not the question).

I've not stolen any code and credited myself with it, so stop talking complete shit about me behind my back. HazeMD is based on MAME because it's a side-project of me improving the Megadrive emulation in MAME, writing a standalone emulator would be completely pointless for that. Nowhere do I say that it isn't based on MAME. I also developed it out of interest in emulating a large number of the pirate carts that nothing else emulates correctly (because they don't handle the protection devices and require hacked roms.) Compatibility is high, it correctly runs a good number of games that nothing else does, and I'm happy with it. It's also open source, and everything I've discovered is free for anybody else to use, the same can't be said for several of your other beloved console emulators, whereby they think that keeping the information to themselves somehow makes them special.

I don't know what Y~K is talking about.

I've been working on MAME for a long time, I've emulated a large number of games, I'm in close contact with a lot of people with professional tools, I know exactly what I'm doing and exactly what I'm saying, and I'll stick by my statement that the console scene and their methodologies suck.

And you know the reason : Cowering.
If this guys have not starting doing mess with console game roms, emulator author haven't been started to support incorrect rom format.
No-Intro really want to change that, have proper dump in proper format but we are not as known as you, emulator authors won't change their way to do a thing because of us.
Hope we can find a way to work together for proper emulation.

===============================================

Re: Haze says: No-Intro is wrong! \ alcoatjez on 20th November 2007, 17:44 wrote:

@Haze: I have already contacted the Guru about dumping systems like Vic-20, VideoPac and MSX. I'm waiting for his reply.
Do you know how much a device costs that can dump standard EPROMs or MASKROMs?

===============================================

Re: Haze says: No-Intro is wrong! \ tetsuo55 on 21st November 2007, 15:43 wrote:

About splitting the roms into seperate parts:

This is valid, even in systems like snes, just look at how superfx and other special carts work, these carts could be much more accurately emulated if all the data was split.
However for the games without a special chip its not quite as important, as long as we can verify that the eventual data is 100% correct (which we currently cannot do without testdumping a few maskroms)