Arcade Machine Dev Kits

Makes sense, the Naomi and the Dreamcast are essentially the same system.

 

I could only wish, it would mean that all the testing and control dev hardware would be pci slotted or something
, and to tell the truth I wouldn't be surprised at this point if the dev unit was closer to something like a Dimm attachment, and firmware for a naomi that allows for 2way comm.

I suppose so, getting thing out of that place would be super tight in either case.

 

One could only wish for a library and some queues to code with, as easy as it would be to just make a mame box and program a game, I would much rather program a cf card for a Naomi (there is just something romantic about it) /drool. I think the release would point towards openGL programming. but who knows... just want... There is a point where mystery needs to take a hike

 

Not true. Several Naomi games were made by third-party developers, such as Capcom, Namco, and Tecmo to name a few. There were also smaller companies, like G.rev, for example.

 

x

 

CPS games were almost certainly NOT developed-developed on X68000, despite the hardware being admittedly INSPIRED by the X68000.

I'm tired of the persistent myth that developers do productivity work on similar architecture to the CPU they target, it's ignorant romanticism by non-developers. In just about every case there is ZERO-incentive to do this.

The most likely CPS1 game component to be developed on the X68K would be the music and sound effects, which actually runs on a completely different processor.

Basically every Japanese developer into the Windows-era used NEC workstations for their main development environment, that is, aside from NEC-competitor platforms (Fujitsu, Sharp). The reason being they are ubiquitous and the best tool sets were written for them. Game developers generally don't make their own tools unless they have to, and when they do (as Capcom would have) they go with what's practical, which would not be the X68K. Sure it was ahead of its time and powerful and might have had some decent 68K tools, especially for developing X68K software... That's irrelevant; the X68K was an entertainment/media computer, it didn't have nearly the industrial support so to cross-develop for a new platform you'd have to build tools to make more tools. To assume they did this would be extremely unreasonable.

If you want to know how arcade games are developed you can break it into parts:

-Hardware

Arcade hardware is typically custom. When a new game is developed the hardware engineers recycled circuits from older games then implemented new improvements. Basically all arcade hardware into the 90s were developed using discrete logic, you know, 74 series chips. This allowed TONS of clever circuits and ideas to be copied by competitors (and enabled bootlegs) so by the mid-80s companies started integrating a lot of the logic into gate-array or standard-cell ASIC (nothing like FPGA). ASIC were developed using CAD systems purchased/rented by the ASIC manufacturer. Basically you input your discrete logic into the system, verify operation, then order chips. Once you get the chips you can assemble finalized hardware.

Generally the launch games would start out running on discrete logic prototypes, while the finalized ASIC hardware is being cooked, then would be carried over to the final hardware.

Arcade games did not have/need custom development hardware until emulator (as in in-circuit-emulator, not software emulator) technology couldn't keep up in the 90s. At that point the options were to develop custom development hardware such as the Sega Katana kit when it was practical, or just use software debuggers which by the 90s weren't a big deal because CPUs had cycles to burn and games were built upon operating systems and libraries already.

-Main program

The main program for just about every 2D arcade game would be developed using a hardware emulator/debugger (again in-circuit-emulator). These systems typically had emulation RAM to simulate ROM and were highly integrated with an IDE so they were a one-stop solution. Basically all you have to do is socket the processor and insert the emulator's pod. If the developer couldn't afford an emulator they would have to develop their own tools, which generally means the software route: bootstrap a "ROM monitor" program with a communication channel to their workstation and physically hack in some RAM to the program ROM's address space. No hardware debugging.

-Graphics

Graphics are developed using custom drawing/animation/tile layout tools. This is for the most part a necessity because custom hardware = custom tile composition.

If the artists' workstation didn't have an adequate (color) display adapter to visualize their work and they couldn't afford a framebuffer card they made an "artist's tool", basically a CPU program in communication with the workstation to assist in positioning tiles and updating the color palettes, then a "ROM emulator" on the graphics ROM bus which allows tiles to be updated dynamically.

Once the art was complete they'd export the tiles and animation metadata to give to the programmer who would test it on their own hardware (suited to programming instead of art development) by burning an EPROM.

When the game is first started and there aren't any graphical resources, placeholders are used to assist in developing the game engine.

-Sound

Arcade sound is near-universally decoupled from the main program allowing it to be developed alongside the main program independently. Nobody wasted time developing custom sound engines for each game, or worse writing code/song, they incrementally developed low-level sound engines across many games and then composed music using music markup compilers or trackers which could be implemented on any platform. A CPS1 tracker was probably developed for X68K, CPS2 no because there wouldn't have been hardware or tool incentives to do so.

With custom sound hardware (as in most boards) sound would be tested via a "sound tool" (a monitor program running on the sound processor that relays data from the workstation over another communication channel to the sound generator).

--------

So what's the message? Arcade developers made extensive use of off-the-shelf components/existing commercial tools. Their "dev kits" were hand-cobbled hardware and software that interfaced with the very same PRODUCTION PCBs you can buy today. It's also possible for many games to have been developed without ANY custom hardware, again by using numerous hardware emulators. All encompassing console-style hardware kits and SDKs don't exist, and not just due to money and time constraints, but also due to the architecture of arcade games. Because there were only a handful of development environments assembled per game (<=1 for each developer typically), and the generic nature and high cost of the tools which could be reused or sold off there's little chance any environments whatsoever have been preserved intact. Maybe an ex-employee kept their toolchain... that's about the best you can hope for. Today if you want to do retro arcade development the only sensible solution is to use modern assemblers, write your own tools, and test/debug via software emulation.

 

Interesting stuff.

 

Extremely informative post. I was merely parroting what I've seen in various other places around the net. From reading on this board in other topics I had figured arcade game development was a hodge-podge process, but never figured it would be -that- tough.

 

Yes, Calpis is right. In the good old days, Atari designed the actual board themselves and developed on it - although if the hardware hadn't been finished, they might use a different game board. At first, they used their own ICE boxes, but later they just bought one designed for the CPU. They had a cluster of VAXen that they used for programming, and a lot of tools were written in-house. Here's what an Atari coin-op hardware guru told me:

I guess the main difference is that back then, they were really pushing their hardware to the limit - programmers didn't get in-game credit because it would take up too much ROM space!! The games evolved with current technology - they were always checking out new chips. Conversely, things have progressed a lot nowadays and, as he said, we have JTAG. I guess technology has evolved that much that it was feasible to have a universal platform (such as Naomi) and more cost effective, too. And, of course, we now have animators and musicians - jobs that would probably have been done by programmers when things were basic!

 

About the developments tool chains, I have finally found what i searched.

Many big Arcade Manufacturers used the X680X0 professional workstations to code and make the graphics of their coin-op on it.

Capcom for instance, replaced their SMC-70 computers on which their made all their 8 bits title graphics, this until 1990.

Strider for instance was drawn on a Sony SMC-70 computer, and programmed on a PC-8801.

From 1990, capcom shelved their old machines and replaced them all by Sharp X68000 ! So yes Most CPS1 and CPS2 games on Sharp X68030 until 1995 where produced with this computer ! Many capcom people confirmed this on the site Video Game Densetsu.

All this explain the 98% carbon copy matching of the CPS games. the First CPS games where made on X68000 with the 10mhz 68000, and the first CPS2 games made with the X68000 XVI with the 16mhz 68000.

Capcom, but Konami as well, and even SNK (some games only) made their games with the sharp X68000. Toaplan, made Outzone, Truxton and so on on X68000 with their Toaplan TED editor. TED for Tile Editor. They also coded the games on the X68000, under OS/9 from Microware for X68000.

They used a huge 20 button custom joystick to paint the graphics. Taito, capcom, SNK, and others used this joystick built by Taito. (compatible on X68000 and PCs)



The graphic toolchain running on X68000 was coming from Konami's own toolchain.

In fact only Sega for instance and IREM used full PC computers as toolchain for their coin-op games.



the picture above shows the X68000 XVI used by taito to make Kaiser Knuckle (1994) and Kirameki Star Road (1997).

During the development of Street Fighter II (1991), Capcom stored its SMC-70 custom and replaced them with computers from the X68000 series. Like the PC-98, the X68000 was a workstation originally developed for professional use (unlike the PC-88 which was marketed towards hobbyists).

Akira Yasuda talked several times about the graphic editor used by Capcom, saying that it was based on Konami’s own graphic tool (it’s unclear whether it has something to do with Yoshiki Okamoto who left Konami to join Capcom in 1984). TCE (for Tiny Character Editor) originally ran on an SMC-70 Custom with which was equipped Capcom when Yasuda joined the company in 1985.

He used a keyboard to interact with this tool -he has never been fond of drawing with a mouse- and it seems that this tool was later converted to X68000 when Capcom decided to replace its workstations in 1990.

Its longevity within the Japanese industry can be explained by many factors, including the fact that it took years for many companies to fully transition from 2D to 3D.

The X68000 was a powerful computer for everything related to 2D and the companies that still used it already had all the necessary development tools to work. The graphic data of games such as Galaxy Fight and Far East of Eden: Kabuki Klash / Tengai Makyō Shinden, two Neo-Geo games released in 1995, made on Sharp’s computer.

ZackmannX, a former Taito employee, sent a mail to one of Sharp’s higher ups in the late 90s, asking him to change his mind after the company announced that it would no longer produce its decade old workstation.

How funny isn't it ? The PC-88 was a leisure computer lol ! And the X68000 was the top ranking machine used for Coin-op games by 90% of the industry !

(source : Video Games Densetsu)

 

Here's something else:



A dev board for Gaiapolis' music for the x68000 made by Konami

 

Nice find, dude!

 

According to the SNK Hardware specification manual they used a PC-9801
For programming/sound and a art box also known as character development unit which had a RGB monitor,bus mouse,EP rom writer,Epson gt-6000 scanner and scsi external HDD

 

Sega used wide range of computers, afaik ~in mid 90x they was Sun Spark, HP-UX, PC with DOS or Mac.
in late 90x (i.e. Dreamcast/NAOMI times) proposed dev.setup was like: SGI Indy for gfx development, PC with WinNT for overall dev.tasks, Apple Macintosh (PowerPC-based) for sound and music production.