I think the approach to making a performant PS3 emulator, if one is actually desired (patent will be up in 10 years, about the same time frame for getting reasonable speed, and most systems are of course very functional still yet), is necessarily going to . It will have to be a ZSNES-style approach of game specific, patchwork hacks and little real accuracy; e.g., the code of the games should inform the design of the emulator, as opposed to the hardware.
Accuracy is a practical impossibility. To understand why, it's important to put the limitations and issues of the age in context. People are expecting 16/24/32 core processors on desktops in the next 10-15 years. While it would probably be feasible to emulate at high accuracy with this much power, these developments are themselves very unlikely to materialize because home wiring can't keep up with the juice required, and neither can residential power grids. We're likely to max out at 8 cores with hyperthreading (or 12 without). Which won't be enough to keep up with a PS3 in an accuracy dependent context. To go higher than this... is politically and logistically impractical. Few will be able to afford these high-juice systems (1%-ers?), and so the demand to produce software for them, let alone emulation, is likely to be non-existent.
Patchwork emulation is disliked by emulator authors for a number of reasons. For one, it puts them at odds with purists who demand "full system simulation", which is what "emulators" like Higan aim to achieve. Actually, there is no prerequisite that an emulator perfectly simulate its target system -- it need only reproduce the operation of the software, which merely requires getting accurate reproduction of the graphics, sound, and peripheral management if available. Anything beyond this is simulation, which is best left to computer scientists operating under reasonable financial direction (I can't think of a single good reason to make a perfectly accurate PS3 emulator, or even a PS2 emulator, frankly). Amongst the people who've been working on the more complex emulation projects over the past decade, a trend is discernable: they are 20-somethings who eventually migrate to jobs which require full system simulation. The experience of writing the emulator is their "proof in the pudding" of this highly specialized expertise, for which the eventual payoff is quite handsome. Simulation then is something which is professionally desirable, where hackish emulation doesn't offer the same prestige (again, the favor aspect of the historicists/purists, most of whom are institutionally entrenched (or at least connected) and powerful).
You could probably get decent performance with an analyzing translator/compiler, possibly a neural net. But who would write this? You'd need a mind capable of writing a PhD thesis in dynamic recompilation to write something that is likely to piss off people in high places. Not likely to happen.
So as has been said before, might as well just pick up one of those PS3s off Ebay or craigslist. If buying something used scares you, hire someone you trust to buy it for you.
Accuracy is a practical impossibility. To understand why, it's important to put the limitations and issues of the age in context. People are expecting 16/24/32 core processors on desktops in the next 10-15 years. While it would probably be feasible to emulate at high accuracy with this much power, these developments are themselves very unlikely to materialize because home wiring can't keep up with the juice required, and neither can residential power grids. We're likely to max out at 8 cores with hyperthreading (or 12 without). Which won't be enough to keep up with a PS3 in an accuracy dependent context. To go higher than this... is politically and logistically impractical. Few will be able to afford these high-juice systems (1%-ers?), and so the demand to produce software for them, let alone emulation, is likely to be non-existent.
Patchwork emulation is disliked by emulator authors for a number of reasons. For one, it puts them at odds with purists who demand "full system simulation", which is what "emulators" like Higan aim to achieve. Actually, there is no prerequisite that an emulator perfectly simulate its target system -- it need only reproduce the operation of the software, which merely requires getting accurate reproduction of the graphics, sound, and peripheral management if available. Anything beyond this is simulation, which is best left to computer scientists operating under reasonable financial direction (I can't think of a single good reason to make a perfectly accurate PS3 emulator, or even a PS2 emulator, frankly). Amongst the people who've been working on the more complex emulation projects over the past decade, a trend is discernable: they are 20-somethings who eventually migrate to jobs which require full system simulation. The experience of writing the emulator is their "proof in the pudding" of this highly specialized expertise, for which the eventual payoff is quite handsome. Simulation then is something which is professionally desirable, where hackish emulation doesn't offer the same prestige (again, the favor aspect of the historicists/purists, most of whom are institutionally entrenched (or at least connected) and powerful).
You could probably get decent performance with an analyzing translator/compiler, possibly a neural net. But who would write this? You'd need a mind capable of writing a PhD thesis in dynamic recompilation to write something that is likely to piss off people in high places. Not likely to happen.
So as has been said before, might as well just pick up one of those PS3s off Ebay or craigslist. If buying something used scares you, hire someone you trust to buy it for you.
Facebook
Twitter
Instagram