Capcom CPS3 arcade core reaches MiSTer FPGA in new beta release

The MiSTer FPGA scene has reached another important milestone with the first public-facing beta release of a Capcom Play System III core, bringing one of the most demanding and historically awkward 2D arcade platforms closer to accurate hardware-level preservation. For retro gaming fans, arcade collectors and FPGA enthusiasts, this is more than another compatibility update. It is a sign that MiSTer’s open hardware ecosystem is still pushing into territory once considered too complex, too memory-hungry or too fragile to reproduce convincingly outside original arcade boards. The first supported titles are Street Fighter III: New Generation and Red Earth, two games that immediately show why CPS3 has such a strong reputation among 2D arcade fans. Both are visually rich, animation-heavy games from Capcom’s late-1990s fighting game era, and both rely on hardware that was very different from the cartridge-based arcade boards many players associate with classic Capcom. The beta is already playable, although it remains a work in progress, with visual issues and timing details still being investigated.

A difficult arcade system comes to FPGA

Capcom launched the CP System III in 1996 as the successor to CPS2, the arcade platform behind many of the company’s best-known 1990s fighting games. CPS3 was powerful, specialised and focused heavily on 2D performance at a time when much of the arcade industry was moving towards 3D graphics. In visual terms, it was a late-generation 2D showcase, capable of huge sprites, fluid animation and rich backgrounds that still look technically impressive today.

Its commercial life was short. Only six games were released for the platform, but the list includes some of Capcom’s most respected fighting titles: Red Earth, Street Fighter III: New Generation, Street Fighter III 2nd Impact, JoJo’s Venture, Street Fighter III 3rd Strike and JoJo’s Bizarre Adventure: Heritage for the Future. That small library has helped give CPS3 a slightly mythical status. It is not broad, but it is highly concentrated.

The hardware was also unusual. CPS3 games were distributed using CD-ROMs and security cartridges, with data loaded into memory on the arcade board. This made the system more flexible than earlier fixed-ROM board designs, but it also introduced a major weakness. The security cartridges relied on battery-backed data. When batteries failed, games could become unusable. For arcade operators, collectors and preservationists, that fragility has long been part of the CPS3 story.

That context makes an FPGA core especially meaningful. Emulation can already run CPS3 games well on modern computers, but FPGA development aims for something different: recreating the behaviour of the original hardware through logic rather than simply running software that imitates it. In practice, that can mean very low latency, accurate video output and a setup that feels closer to original arcade timing when connected to modern screens or CRT displays.

Why MiSTer users care

MiSTer FPGA has become one of the most important platforms in retro hardware preservation because it sits between original hardware and software emulation. Built around an FPGA board, it can be configured to behave like different consoles, computers and arcade systems. Each core describes the logic of a particular machine, allowing the hardware to be reconfigured for that system.

That approach has made MiSTer popular with players who care about input latency, video accuracy and faithful reproduction of classic machines. It has also become a serious preservation platform, especially for arcade boards that are ageing, expensive or difficult to maintain. Every new high-profile arcade core expands what the platform can offer without relying on original, failing hardware.

CPS3 is a particularly demanding target because of its memory requirements, graphics pipeline and CD-based design. The first MiSTer beta focuses on No CD arcade ROM sets and games that fit within a 64 MB SDRAM footprint. That technical limitation shapes the initial release, but it also shows careful engineering. Getting a complex system like CPS3 running within standard MiSTer memory constraints is a major part of the achievement.

For users, the immediate benefit is access to two landmark Capcom titles through MiSTer. For developers and preservation-minded fans, the more important point is that the architecture is now alive on FPGA and can be improved over time.

Street Fighter III and Red Earth lead the first wave

The first two games are a sensible starting point. Street Fighter III: New Generation was Capcom’s long-awaited continuation of its flagship fighting series, and although it was controversial at launch for replacing much of the familiar Street Fighter II cast, it became an important foundation for the later Street Fighter III entries. It also introduced the animation style and technical personality that would later make 3rd Strike one of the most admired fighting games of all time.

Red Earth, known in Japan as Warzard, is a different kind of showcase. It mixes one-on-one fighting with fantasy role-playing elements, large boss characters and a more experimental structure than Capcom’s tournament-focused fighters. It was also the first CPS3 game, making it a natural fit for the core’s debut. Its big sprites and distinctive art direction are useful tests for any attempt to reproduce the hardware accurately.

Starting with these games allows the beta to demonstrate both technical range and visual character. Street Fighter III puts focus on fluid character animation and responsive fighting mechanics. Red Earth stresses scale, effects and unusual presentation. Together, they make a strong opening statement for the core.

The rest of the CPS3 library is expected to follow in stages. That staged rollout is important because each game can expose different hardware behaviours, timing issues and edge cases. A core can appear stable in one title and still require significant work for another, especially on an arcade platform with complex graphics and memory handling.

Still a beta, not a finished product

Although the arrival of CPS3 on MiSTer is big news, this release should be understood as a beta. The current build is playable, but it is not final. Known issues include visual glitches, text artefacts, occasional solid colour squares, horizontal positioning problems on certain screen elements and missing full-screen zoom effects. Audio and pause behaviour also still need refinement, while DMA timing remains an area for verification against real hardware.

That may sound like a long list, but it is normal for a complex FPGA arcade core at this stage. Early betas often expose the hardest problems only after more users begin testing different games, displays, settings and memory configurations. The point of a beta is not perfection; it is to move from internal development to wider testing while the major pieces are already functional.

For players who simply want a flawless plug-and-play experience, waiting for later public releases may be the better option. For enthusiasts who enjoy following core development closely, the current beta is exactly the kind of moment that makes the MiSTer scene interesting. It shows the system working, but also shows the technical work still happening underneath.

No CD support is a practical choice

One detail likely to stand out is the use of No CD ROM sets rather than original CD images. For purists, that might sound like a compromise. In practice, it is a pragmatic decision. CPS3 originally used CD-ROM media to load game data into memory, but once that data is available in a suitable form, recreating the physical CD-loading process is not necessarily useful for gameplay or preservation on MiSTer.

The important question is whether the resulting behaviour matches the arcade system once the game is running. For an FPGA implementation, avoiding CD image handling can reduce complexity and keep development focused on the actual hardware logic that affects play, graphics, timing and audio.

It also fits the way most MiSTer users interact with arcade cores. The platform is valued for quick loading, clean configuration and accurate output, not for recreating every inconvenience of a 1990s arcade installation process. Preserving the game experience does not always mean preserving the slowest or most failure-prone part of the original setup.

Why CPS3 still matters in 2026

CPS3 matters because it represents the final peak of Capcom’s proprietary 2D arcade hardware. By the late 1990s, arcade technology was changing rapidly. Sega’s Naomi platform, 3D accelerators and home console convergence were reshaping the industry. Capcom’s CPS3 stood slightly apart from that trend, delivering some of the most elaborate 2D fighting game visuals ever produced.

That makes it historically important, even with only six games. Street Fighter III 3rd Strike remains a competitive and cultural touchstone. Red Earth has gained more attention over time as players revisit Capcom’s experimental catalogue. The JoJo titles have their own dedicated audience, helped by the broader popularity of the manga and anime franchise. The platform’s small library has aged unusually well because it was built around strong art direction rather than short-lived 3D spectacle.

CPS3 is also important because original hardware is not getting younger. Security cartridge issues, optical drives, flash memory and ageing boards all create long-term preservation problems. FPGA cores cannot replace the historical value of original cabinets and boards, but they can reduce dependence on fragile hardware for everyday play, study and community access.

What this means for arcade preservation

The CPS3 beta reinforces a broader trend: FPGA preservation is no longer limited to early consoles and simple arcade boards. Developers are now tackling increasingly advanced systems with complex memory maps, custom chips and demanding video behaviour. Each successful project expands expectations for what can be preserved at a hardware-description level.

This does not make software emulation obsolete. Mature emulators remain essential, flexible and often more accessible. They support save states, debugging, netplay, scaling options and a huge range of devices. FPGA and software emulation serve overlapping but distinct roles. The value of MiSTer is that it offers another path, one focused on hardware timing, low latency and authentic display options.

For arcade preservation, having multiple approaches is healthy. Original boards, software emulation, FPGA cores, documentation projects and community testing all contribute to the same goal: keeping games playable and understandable beyond the lifespan of their original parts.

Not for every FPGA device

The CPS3 core also highlights an important reality in FPGA gaming: not every FPGA platform can run every core. Some users may hope for support on smaller handheld FPGA devices, but CPS3 is simply too large and demanding for certain hardware. Its memory and logic requirements put it outside the practical reach of platforms with tighter resource limits.

That is not a failure of optimisation. It is a hardware ceiling. FPGA devices vary widely in capacity, memory bandwidth and available logic elements. A core that is possible on MiSTer may be unrealistic elsewhere, especially when the goal is accuracy rather than a heavily simplified approximation.

For the MiSTer ecosystem, this strengthens the platform’s role as a serious arcade preservation device. It also reminds users that FPGA development is not magic. These cores are the result of careful trade-offs, deep hardware research and many hours of debugging.

A strong signal for MiSTer’s future

The arrival of CPS3 gives MiSTer one of its most exciting arcade updates in recent memory. It brings two important Capcom titles to the platform now, with the promise of more to come as development continues. It also demonstrates that the community can still push into difficult late-1990s arcade hardware, not just refine older systems that are already well understood.

There is work ahead. The beta needs fixes, testing and broader game support before it can be considered mature. But the hardest psychological barrier has already been crossed: CPS3 is running on MiSTer, and it is playable.

For retro gaming fans, that is enough to make this release worth watching closely. For preservationists, it is another step towards keeping fragile arcade history alive in a form that people can actually use. For MiSTer owners, it is a reminder of why the platform remains one of the most compelling projects in retro computing: it keeps turning once-impossible hardware targets into living, playable cores.