Will this hopefully support cd games or cartridge only? Also are we going to have a way to use the VLM even if we have to use wavs again? OK make it three ?'s, does cannon fodder run?
- CD has never been a priority - with there only being a handful of CD games there (and me not having the JagCD or any titles for it) it's a long way back on the list.
- The sound capture support has been rewritten to conform to Emu's ioservice specs but the hardware support hasn't been tested. It may be a little trickier to get working than before, because it needs to be more accurate (to support loading Spectrum games from tape - it can't afford to drop any samples) and because the timing master system has changed a lot. However, the ioservice approach means that substituting a different backend (such as the wav file method) wouldn't be at all difficult.
- Cannon Fodder worked fine last time I had a look at it.
paddy said:
With this being a port of T2K,does that mean it will have the same graphical bugs as well as the limited compatability and sound sync issues ? ,is the project open source meaning you can fully optimise it
as well as develop it further.
Hmm. Well...
- the GPU and DSP core are broadly the same as T2K's although with some small bugfixes, cycle timing rather than instruction counting and accuracy improvements
- the core of the blitter hasn't really changed at all - the code's completely unmaintainable, making any change to it usually breaks it. I've started on a rewrite but it's slow going.
- the 68K core is now mine instead of Starscream, which is somewhat more accurate and notably emulates prefetch. While that's absolutely essential for ST emulation, it should be a wash on the Jag.
- multi-CPU (target) timing actually has some accuracy involved. The naive model that T2K uses was only really equipped to play T2K - anything that actually made parallel use of the 3 CPUs would almost inevitably fail. Emu's got a fair scheduler that divvies up cycles accurately between the chips
- the new timing makes it quite a bit slower than it was but that's probably necessary for accuracy.
- The hugely enhanced debugger has let me track down quite a few small issues - some interrupts were being handled wrong and that was breaking D2K's classic and plus modes; I'd put a dummy value return from some register that was breaking AvP's enemy AI - things like that.
The goal is the same as for all Emu's targets: 100% compatibility. The complexity of the Jag means there's still a fair way to go.
The project is not open source, although I'm quite happy to share the source with someone who can convince me they have something to contribute and are prepared to agree that the core and emulators are mine
. Emu's primary purpose is still for me to play with all the emulation I want to do over the rest of my lifetime, and in the very odd view in my head I'm wary of letting other people in because they might do things I would find fun to do myself
. I view emulation as more like sudoku than anything else.
Which isn't to say that that view is set in stone, but that's the way it is for now.
I haven't tried Chequered Flag. If it uses Z then it definitely won't work until the new blitter.