I don't quite agree: an emulator (including sources!) that can do 5 fps would be a major thing. You never know if some very clever hacker can squeeze enough extra performance out of it to make things playable, but this is much less likely to happen if he has to start a new project from scratch. Also, when (in the far future) there will be a Pandora 2, it is likely to have better hardware which is mostly downward compatible, so it would be easy to get the "slow" DC emulator for the Pandora1 to run at full speed on the Pandora 2.
In my opinion, someone should try to get the sources of nulldce from the original devs who for some reason had no time to finish the project. Then some of the super coders in this community can try to optimize that emulator. A fully accurate full speed emulator might not be feasible on the current Pandora hardware, but it could very well be possible to sacrifice some accuracy and get full speed.
Never underestimate how much room for improvement there is to optimize stuff if you know the hardware well enough. Just compare current Commodore 64 scene demos to the games that were released for it back in the days. The hardware is the same, but what people can do with it tends to improve over time. The Pandora hardware still has unused features (e.g. the DSP), if those can be effectively used by the emulator, much is possible. Starting with an emulator that uses the CPU to emulate everything, big speedups can be possible if those other features can somehow be used (e.g. NEON, GLES, DSP, etc. but also taking into account caching and pipelining and all those lowlevel things that can give speedups). You can also get a small speedup by eliminating the multitasking default linux kernel and doing things in single-task mode so no cycles are wasted by process scheduling etc. Don't assume that when somebody shows you a slow emulator, this means fast emulators are impossible. A slow emulator is just an important first step.
In my opinion, someone should try to get the sources of nulldce from the original devs who for some reason had no time to finish the project. Then some of the super coders in this community can try to optimize that emulator. A fully accurate full speed emulator might not be feasible on the current Pandora hardware, but it could very well be possible to sacrifice some accuracy and get full speed.
Never underestimate how much room for improvement there is to optimize stuff if you know the hardware well enough. Just compare current Commodore 64 scene demos to the games that were released for it back in the days. The hardware is the same, but what people can do with it tends to improve over time. The Pandora hardware still has unused features (e.g. the DSP), if those can be effectively used by the emulator, much is possible. Starting with an emulator that uses the CPU to emulate everything, big speedups can be possible if those other features can somehow be used (e.g. NEON, GLES, DSP, etc. but also taking into account caching and pipelining and all those lowlevel things that can give speedups). You can also get a small speedup by eliminating the multitasking default linux kernel and doing things in single-task mode so no cycles are wasted by process scheduling etc. Don't assume that when somebody shows you a slow emulator, this means fast emulators are impossible. A slow emulator is just an important first step.
