05-17-2013, 04:13 PM -
Depends on what you call 'finished'. I hate to give those vague answers every time but that's all we got for now, so here it goes:
The core will never get really perfect, at least not in the near future, since we're HLE'ing the hardware, but we'll get to a point where most things that are implemented work accurately enough for now and what's not implemented we will probably only find out that it exists when not having it causes some problem.
We'll probably get to a stage where the devs will focus on get things running, sometimes focusing on improving accuracy and sometimes focusing on improving speed, which will involve development on every part of the emulator. That's how it *usually* works with this kind of emulator.
Our CPU is currently being interpreted. Writing an Interpreter is the most basic form of emulation, it is relatively easy to achieve accuracy on that part if you have all the info you need about the chip, which we don't, and we're not talking about any chip here, it's an extremely complex one.
The thing is, interpreters are really slow. So at some point in the future, when the interpreter is good enough, we'll probably want to have a dynamic recompiler wrote for the CPU, which is a much more efficient way of emulating it, but is harder to program and harder to get accurate emulation with. In other words, it will have even more bugs and glitches, and that's why it's important to code a interpreter first, so we can debug stuff later.
So considering all that, I don't think we're getting the core done anytime soon, but it'll be playable, eventually. If that's gonna happen this year is hard to say, but I'd doubt it.
The core will never get really perfect, at least not in the near future, since we're HLE'ing the hardware, but we'll get to a point where most things that are implemented work accurately enough for now and what's not implemented we will probably only find out that it exists when not having it causes some problem.
We'll probably get to a stage where the devs will focus on get things running, sometimes focusing on improving accuracy and sometimes focusing on improving speed, which will involve development on every part of the emulator. That's how it *usually* works with this kind of emulator.
Our CPU is currently being interpreted. Writing an Interpreter is the most basic form of emulation, it is relatively easy to achieve accuracy on that part if you have all the info you need about the chip, which we don't, and we're not talking about any chip here, it's an extremely complex one.
The thing is, interpreters are really slow. So at some point in the future, when the interpreter is good enough, we'll probably want to have a dynamic recompiler wrote for the CPU, which is a much more efficient way of emulating it, but is harder to program and harder to get accurate emulation with. In other words, it will have even more bugs and glitches, and that's why it's important to code a interpreter first, so we can debug stuff later.
So considering all that, I don't think we're getting the core done anytime soon, but it'll be playable, eventually. If that's gonna happen this year is hard to say, but I'd doubt it.