Vote: What style M92 Multi would you buy?

[AlxUnderBase]

Simple one for me too… if can use old style eeproms much better

 

[AlxUnderBase]

Vote skip, and port them all to M107 and mess everyone up

How many collectors can put hand or have m107 hardware ?

 

[kuze]

Vote skip, and port them all to M107 and mess everyone up

Really excited for your M84 multi board by the way Irem rocks!!

 

[Brentradio]

I also would like to add another option:

c) I would buy the first M92 multi that is available.

 

[jassin000]

port them all to M107

I'm 99.9% sure this has already happened in private.
Problem is M107 boards while cheap are pretty rare, too rare to craft a multi around IMO.

 

[ShootTheCore]

I also would like to add another option:

c) I would buy the first M92 multi that is available.

THIS!

 

[dewmansnk]

Would an M92 that loaded ROMS via an attached Raspberry Pi (similar to Naomi netbooting, but way less flaky) be something that folks would want? Since we are in a component shortage you got to get creative! The Linsen RV901T is a Spartan 6 FPGA platform, with SRAM and Ethernet Phy. There is a fantastic base design, written in Litex python, that has an Etherbone (Wishbone + Ethernet) stack implemented. There is some companion python code and an example design that allows you to write data to wishbone addresses over ethernet. This allows you to write to addresses inside the FPGA fabric from a python script running on a PC/Pi attached via ethernet. All on a $35 board that is 'currently' still easy to find on Amazon/AliExpress

I have been thinking that using the IO pins on the RV901T (its about 64 IO pins, 32 on each connector) you could add SRAM to the wishbone bus and then be able to write to those SRAM addresses via the python script on the attached Pi/PC. If the SRAM was configured with the right data, and attached to the M92 at the right spots, that should be most of the battle. You would be using darn near all the IO pins to do it (there are some more on another connector, but it starts getting kludgy) and some gaps in my understanding of how the PALs come into play, jumper settings, other nuances I don't even know about, etc. I know I could look through all the MAME source, but maybe this multi should just be open sourced and we get to the finish line that way.

For a start, I can pry whip up something in KiCad that would be 2 PCBs, with SRAMs and some interface logic and get it attached to the Linsen board. This would be targeting MT2 board first, but once the thing actually work, spin-off layouts are not all the hard. Its moving pin headers around and leaving space for the Linsen board to be attached - not all that much layout work, and layout is my fun to me, hence the last design (M72 ROM board) was all layout and no design work. So for the areas I would need help in:

-The PALs - would be nice to make a universal PAL and not have the FPGA have to do the PAL address mapping as well. The Linsen board has buffers that are either configured as output or input - if the Linsen only needs to bootstrap the SRAM it can all be configured as outputs - if it needs to start implementing the PAL addresses mapping, some of those Linsen outputs will need to be switched to inputs, and that means cuts/jumpers on the Linsen card.
-I suck at Python, and so while the base design is fantastic, its written in Litex (A fraction HDL language based in Python) and I am still struggling through how to make changes. If someone has worked with Migen/LiteX and could help move the ball forward in that area.
-The PC/Pi side python loader (see above point) so while the example python code is available, It would be helpful if someone knew how to take a ROM set and write the files to the appropriate addresses on the wishbone bus (which would have SRAM attached to receive the code). How to potentially split up the ROM files so that they could all be written with 8-bit data width to the wishbone bus at arbitrary addresses.
-A ROM set with decrypted audio?

 

[jassin000]

Would an M92 that loaded ROMS via an attached Raspberry Pi (similar to Naomi netbooting, but way less flaky) be something that folks would want?

In this case I would say no... People have been waiting for years!
The ROMs/conversions have been patched/created long ago, so all we need at this point is a stupid/dumb bank switcher with DIPs on it.
No FPGA/CPU required PERIOD!

Get this device out the door first, then later go back for the "Champion Edition" or "Ultra" revision...
Supporting all the PCBs, with LCD screen selector and microSD loading (launching in 2067 at this fucking rate *snicker*).

 

[dewmansnk]

In this case I would say no... People have been waiting for years!
The ROMs/conversions have been patched/created long ago, so all we need at this point is a stupid/dumb bank switcher with DIPs on it.
No FPGA/CPU required PERIOD!

Get this device out the door first, then later go back for the "Champion Edition" or "Ultra" revision...
Supporting all the PCBs, with LCD screen selector and microSD loading (launching in 2067 at this fucking rate *snicker*).

I am suggesting the quickest path to a physical multi. Having the ~14 game M92 ROM set on EEPROMs is not viable. The PCBs I have suggested are pretty simple, and could be cobbled together in a few weeks. The rest is essentially a software issue that I have suggested we open up to the community. You mentioned not wanting to deal with burning EEPORMs previously, yet suggest a multi format that could have 20+ EEPROMs as @rtw roughed out.

 

[jassin000]

Having the ~14 game M92 ROM set on EEPROMs is not viable.

Its not, this would be preformed with large capacity TSOPs or other modern storage medium.
Take a look at the Namco System 1 multi (not the Sega System 16 multi) to see what I mean.

See also @twistedsymphony 's post #13.

So NO, NOT the fastest path... Sorry (not sorry/all this arguing about what would be better is what delayed this project for years in the first place).

 

[twistedsymphony]

I actually built a proof of concept multi based around both arduino and Pi, the biggest issue is they're slow as balls to write to the ROMs because they don't have have enough GPIO

 

[dewmansnk]

I actually built a proof of concept multi based around both arduino and Pi, the biggest issue is they're slow as balls to write to the ROMs because they don't have have enough GPIO

I really like the Pi for cheap bulk storage and with Etherbone being fully implemented already on the Rv901T, it seemed like a pretty good pair for solving the issue of having a large library of games loadable into some memory format that works well with arcade PCBs. I didn't even mention the 64Mb of 200Mhz SDRAM that is part of the reference design on the RV901T, so you could almost get away not even need any external SRAM and just use the SDRAM already there, but we are going for the simple approach, so just using the RV901T to bootstrap onboard SRAM was the quickest approach.

 

[twistedsymphony]

some memory format that works well with arcade PCBs

you're literally hand-waving the only piece of this that actually matters.

 

[jassin000]

I personally HATE Pi's!
They are slow, prone to microSD corruption (I reimage my SMB-OPL Pi at least once a year due to corruption) and at least with the 3/4 too damn power hungry for what they are (aka I should be able to run the GD thing with a 5v 1amp PSU just like the Pi 1/2's).

Pi 4 has two microHDMI ports for Christ Sake, WHY?
Is someone actually using it as a dual monitor desktop PC? My god, the horror!

 

[dewmansnk]

Well 5V SRAM to be specific, in 8-bit width, stuff that Digikey has in stock.
https://www.digikey.com/en/products/detail/alliance-memory-inc/AS6C8008-55ZIN/4234598

 

[ShootTheCore]

I don't mind a multiple EPROM bank-switching setup. I've programmed EPROMs for dozens of Sega System 16 and Irem M72 kits free of charge for AP folks - having to program 15 EPROMs for a kit takes some time, but it's not that big of a deal. The upcoming Jaleco System 1 Multi is going to need around 30 EPROMs programmed per kit and I'm ready for it.

The flash chips Darksoft uses for the Namco System 1 Multi work fine too - I'm happy to program those for folks at no charge as well.

I'm not crazy about the Raspberry Pi approach unless it can be mounted & integrated in such a way that you don't have wires and power adapters running everywhere. Ask anyone with a Naomi NetBoot setup if they'd rather have that or a Multi cart that accomplished the same thing and I think you'd have unanimous consensus on the cart.

 

[dexster]

a MT2 multi using eeproms for m92 would be fantastic

Vote skip, and port them all to M107 and mess everyone up

it is all good Hammy - air assault and dream soccer are games in my regular rotation

truth is i would buy what becomes available first
(but jassin000 mentioned it previously build the easiest multi first then adapt/improve it over time)

 

[jassin000]

Ask anyone with a Naomi NetBoot setup if they'd rather have that or a Multi cart that accomplished the same thing and I think you'd have unanimous consensus on the cart

Someone gets it!

 

[Hammy]

Really excited for your M84 multi board by the way Irem rocks!!

M84 is real, got the schematic as proof

I'm 99.9% sure this has already happened in private.
Problem is M107 boards while cheap are pretty rare, too rare to craft a multi around IMO.

Yeah it's possible so not surprised. I hear M97 can be ported to it also.

 

[rtw]

I actually built a proof of concept multi based around both arduino and Pi, the biggest issue is they're slow as balls to write to the ROMs because they don't have have enough GPIO

I have moved from the arduino to the Raspberry Pi Pico, 4 USD