Dreamcast Emulator Redream 1.5.0 Progress Report

(To my and my brothers' initial disappointment, my parents didn't understand why we wanted a PS2 and got us a Dreamcast instead, then refused to get a second game system. One of my brothers still has and plays that Dreamcast.)

For sort of 80/20 compat the PS2 is easier. You can HLE a couple of the processors, the GPU side of things is from the fixed function days, and has a split VRAM like your PC. The PS3 also has a nasty habit of making the SPEs make up for the anemic GPU and what would traditionally be pixel shader post processing tasks.

For Dolphin level full compat, IMO the PS3 is easier. The complexity memory subsystem and just plain number of jobs in flight meant that synchronization was very explicit between the components, as you couldn't really get away with "I'm just going to assume some happens before relationship is valid because I counted the requisite number of cycles". So you can rely quite a bit more as an emulator author that the game authors will make their intentions clearer. There's also more legacy in the PS2; the IO processor is itself a full PS1, that's how backwards compat worked. So you're really looking at all the work to make two emulators in one.

The PS3 does not have self-modifying PPC code (SCEI forbade it), which means the PPE blob can be compiled ahead-of-time (RPCS3 converts it into LLVM). The SPE data can self-modify, however, but (to my knowledge) does not require extensive synchronisation, therefore each SPE core can be put on a thread.

The PS2 code has fairly extensive use of self-modifying code; Naughty Dog in particular will frequently load parts of the executable in and out of memory on both the PS2 main processor (the EE) and the PS1 processor (the IOP), and rely on the synchronisation between these two separate processors to be fairly tight. Trying to make the EE and IOP separate threads running simultaneously breaks this synchronisation, so the EE and IOP have to run on the same thread.

Additionally, the PS2 has two vector units; VU0 is associated with the EE (it can be used as a floating-point SIMD unit in the EE instruction stream) and VU1 is associated with the GS [the PS2's GPU, the Graphics Synthesizer] Interface (GIF) (it can directly output primitives to the GS). This means that VU0 needs to run on the same thread that the EE runs on (because there is instruction stream interlocking), and VU1 needs relatively tight synchronisation to the GS (it is feasible to put it on its own thread, but games can be quite picky with timings)

Timing remains my main problem with emulation. Input and output lag have gotten really bad and so hard to predict until you start actually messing with the equipment you'll be using, and even then little config quirks or the way things are hooked up (adding a receiver to the mix, for instance) can throw it all over the place. Lots of games become unplayable or unreasonably hard with just a little extra lag, so it's a real problem.

For most games, VU borrowing "cycles" from CPU can help a great deal, so can enabling the threading. The defaults are the most compatible, but also slowest.

Also, GSDX is the only gpu plugin that works semi-decently, and only with actual DirectX. Opengl is all over the place.

Most games work fine even on a ten year old computer, with these settings, as long as you're on Windows.

Linux support could be better, but there's no manpower for that. It is kept alive by one or two developers who use Linux, most of the team being on Windows.