Patch to Solve conflicts between Snk games and 4Mb

[ExCyber]

It just has 4MB in two 2Mx8 chips.

 

[keropi]

Originally posted by Shakey Jake33+Jun 30, 2003 @ 03:35 AM--><div class='quotetop'>QUOTE(Shakey Jake33 @ Jun 30, 2003 @ 03:35 AM)</div><div class='quotemain'> <!--QuoteBegin-mal@Jul 1, 2003 @ 10:32 AM

So how does an AR4M+ cope with both?

I seem to remember the cart itself auto-switches, as in cuts some circuits and opens some others.... the actual hardware itself changing. [/b][/quote]

ar+ worx perfect with both, tested my self

 

[Tagrineth]

Originally posted by ExCyber@Jul 2, 2003 @ 10:29 PM

The connections between the RAM chip, the circuit board, and the cartridge logic.

As far as I can remember, on my Action Replay 4M Plus, the address bus on the cart slot is just connected to the DRAM through a pair of muxes. This would not provide the same memory map as what would be on the 1MB cart specced by Sega - if a game just used the 1MB RAM where it was officially mapped it should work fine since they overlap, but any access outside the RAM area for 1MB can't be guaranteed to be compatible (it probably mirrors on a real cart, while it would just read/write normally on a 4MB one).

So how do you explain why SNK's 1MB-only games fuck up on native 4MB cartridges?

At least mine is plausible. It also makes sense because the earliest AR4M+'s with the 4MB capability had one erroneous connection, and incidentally, the faulty boards were made before the official 4MB cartridge was released! Sounds like Sega modified the spec to me. But the ones with the wrong connection for 4MB still work perfectly for 1MB. I can attest to this; my AR4M+ is one of those, and needed a trace cut and a new wire added to replace it.

 

[antime]

The "faulty" Action Replays use the device ID of 1MB cartridges, so the games that require the 4MB expansion refuse to boot. I doubt Sega changed the spec, more probably Datel didn't know that the ID was different for 4MB carts.

 

[Tagrineth]

Originally posted by antime@Jul 4, 2003 @ 03:13 PM

The "faulty" Action Replays use the device ID of 1MB cartridges, so the games that require the 4MB expansion refuse to boot. I doubt Sega changed the spec, more probably Datel didn't know that the ID was different for 4MB carts.

I like that reasoning.

 

[antime]

The games in question obviously check for the presence of both the 1MB and 4MB carts (or neither), since they work with all 4MB Action Replays.

 

[ExCyber]

So how do you explain why SNK's 1MB-only games fuck up on native 4MB cartridges?

I'm not saying there's any substantial difference between a "corrected" AR4M Plus and a Sega 4MB cart; my stuff about the ID and mirroring applies to 4MB vs. 1MB as specified by SOA Technical Bulletin 47. I suspect (from poking around in Samurai Spirits 4, not just idle speculation) that the "SNK" carts are actually Sega carts, or at least designed to the Sega specification.

more probably Datel didn't know that the ID was different for 4MB carts.

More probably EMS was modifying Datel's Action Replay Plus (1MB) design and didn't have any info on the ID byte at all.

 

[M3d10n]

Hmm... throwing another log in the fire:

In my KOF '97 box it says that it supports both 1MB and 4MB RAM carts. Maybe there are differences among the different 4MB carts avaliable...

 

[ExCyber]

In my KOF '97 box it says that it supports both 1MB and 4MB RAM carts. Maybe there are differences among the different 4MB carts avaliable...

Has anyone confirmed that KOF '97 in particular does have problems with the Action Replay 4M Plus? I can't do a meaningful test because mine appears to be defective (either bad memory or too leaky, per consistent SFZ3 graphics corruption).

 

[Tagrineth]

I don't know about KOF, but there is another game that supports both 1MB and 4MB, so they probably have different device ID's that the games request, allowing the AR4M+ to use alternate logic paths for 1MB and 4MB. It would make sense, considering that games that can support BOTH expansions don't make the poor AR4M+ explode. =)

 

[ExCyber]

I don't know about KOF, but there is another game that supports both 1MB and 4MB, so they probably have different device ID's that the games request, allowing the AR4M+ to use alternate logic paths for 1MB and 4MB.

The game can validate both IDs, but the game doesn't request a particular ID, it just reads whatever ID the cart supplies at 4FFFFFF and decides what to do based on that. That means that to be able to change the ID, the AR would have to know what ID the game wants before the game reads the ID. The only real way to do this would be to detect the game itself by other means (title/number check, header hashing, etc.) or provide a manual switch.

 

[ExCyber]

So far it looks like SS4 treats the 1MB and 4MB carts identically; the detection routine just returns zero if either is present or -1 if any other ID is read.

 

[shinhoshi]

The only SNK games I have tested are KOF 96 and SS 4 and they have those graphic glitches when used with the 4MB cart.

A curious thing is that in SS 4 if you take the same character (on VS) there are no glitches at all.

On KOf 96 the things go more randomly but it always happens the same with the characters. I mean it's the same glitch EVERYTIME.

On SFZ3, Pocket Fighters, FF Revenge the cart just plays fine.

I openened the IP bin with an hex editor and found a string I thought it was the cart type... I haven't investigated it more and doubt it changes anything if I change it.