If there was an Emu project you could fundraise to complete, what would it be.

Ha - easier said than done! Back to the times of UltraHLE the power of PCs wasnt that much to correctly emulate the N64. And on that, there was almost no knowledge on the RCP (situation wich is almost the same today).

Emulate the N64 would mean :
a) emulate a R4300i (documented)
b) emulate a undocumented RCP

Lots of work...but still intresting

 

NullDC and PCSX2. I've retired my DC at this point, but still feel the urge to play the games from time to time, and playing them in high rez would be neat. I havn't retired my PS2 yet, but some games could really use a boost in resolution (wish Sony had mandated a min resolution like Sega did :-/)

edit: Also, a new thread in the 0th bit!? was starting the think it'd never happen

 

Actually the MAME n64 code is not bad... It emulates the RCP so LLE emulation is possible. I believe Jabo is writing a LLE graphics plugin for Project64 as well.

But it will be some time before computers are able to accurately emulate all the n64 hardware at such a low level.

 

Indeed. The RCP is totally powerfull.

 

I would want an Naomi emulator most of all right now but since things are moving very very slowly with that I would have to say 64DD emulator since the hardware has always interested me a lot and it would be great to finally have a go on the games

 

Yabuse DC, any Cd-i emu, a Super A'can emu, and a 64DD emu.

 

It's been proven over the last 10 years that the best emulators will always be the freest emulators.

It's also worth noting that emulator authors are often not able to perform tests on original hardware, either because of lack of equipment or knowhow, so they usually rely on existing reverse engineered docs. If the documentation isn't there though, no amount of donations will progress the project further.

Consider sponsoring documentation efforts instead because if the documentation is out there, people will certainly implement it for free.

 

Well for instance what documentation would we need to sponser for something like Yabuase

 

For a Saturn emulator it may be helpful to use money to purchase any unleaked Sega documents and reward reverse engineered breakthroughs.
These efforts may be actually made by the emulator authors, but I wanted to make the point that this is often not the case. Unless you are in direct contact with the authors, you won't know what they need to implement your function whether it's time, knowledge, money, motivation or all of the above. It's unlikely that there are many people actively reverse engineering the Saturn, if you could create interest (which costs nothing) that would go a long way. Donations are a great morale booster for authors, especially those of free work, but probably still won't get somebody to do your bidding.

 

speaking of Saturn emulators, SSR 0.8 R11 has been released August 25th. I tried it and it's a dream, give it a shot - Filtering and 32bit color seem to work fine too.

 

Speaking of dreamcast this talk reminds me of Bleemcast and Pscx dc.
Yet more examples of DC projects gone down the toliet.
Add these two to my list.

 

AFAIK, Yabause DC isn't even really optimized, it's just a simple port. If a devleoper(s) went through the code, focused on speed and optimized it for the SH4, the framerate could increase massively. I'm not much of a coder myself (VB, and that's about it), but maybe the approach, instead of emulation, is some kind of advanced virtual machine? The SH4 should be backwards compatible with the SH2 used in the Saturn, so instead of emulating itself, make a VM that has 2 virtual processors and use the rest of the computing power on emulating the 68k and VDPs.

Well that's my theory anyway.

Edit: Re-reading, I kinda sound like a jerk saying Yabause DC is 'just a simple port'. It takes a superhuman effort to port a huge emulation project to an unsupported platform without payment or support, and BlueCrab deserves an enormous amount of Kudos for his efforts, and I hope he continues the porting and development of Yabause DC until it is in a usable state, and beyond.

 

So why not mention that to Blue Crab?

Just make a note of it in the thread they got going at DCEmu

I'm sure imput would be appreciated.

http://www.dcemu.co.uk/vbulletin/showthread.php?goto=newpost&t=71498

 

Translating SH2 to SH4 would be something like dynamic recompilation, pretty tricky stuff and not very necessary. A regular old interpreted core is perfectly good in this case since the SH2 instruction set should map perfectly into SH4. CPUs are never the bottleneck anyway.

 

Surely translating = emulation? I was thinking of like Virtual PC or VMware: it uses your CPU to make another PC inside yours.
Like this: http://img.osnews.com/img/3863/vmware1.jpg
It's not emulation, since it uses the processor without converting any commands. The virtual XP sends a command, it travels through VMware and it's processed by the CPU natively.

Of course, a Saturn one would need to virtulize 2 CPUs, but the Dreamcast CPU is about 5-10x faster than the Saturn's, so it shouldn't be an issue.

This is all theoretical, by the way.

 

yes, but dont forget how the SH4 is a single processor, whereas it would have to emulate the interchanging nature of the dual SH2 and the shared bus etc. Throw the Sh1 in there as well, along with the other two chips and u ve got yourself quite alot of stuff to orchestrate.

 

Once the people working on the MAME/MESS Nintendo 64 core can get it functioning better with proper emulation of the RSP and RDP (the RSP is basically a MIPS chip with a bunch of wierd stuff) then it should be possible to use the existing MIPS dynamic recompiler to obtain the speed necessary to emulate the 2 MIPS CPUs, the RDP 3D display processor, the video interface plus the extra stuff (I/O, sound etc) on a modern PC.

 

While very true, that also has to be done with emulation. With my idea, there should be much less overhead for the processes themselves, and the CPU can focus on emulating the 68k and VDPs more.

 

hmmm maybe Model 3, and Naomi/2/GDrom would be nice

 

The thing is, SH4 isn't necessary exactly SH2 compatible. You always have to wrap instructions with some code to contain them in a fake memory space and make flags and operations binary compatible with the target system. If you're shooting for very accurate emulation, you need to emulate time as well. There's always that overhead. For dynamic recompilation, foreign instructions and memory addresses are translated on the fly or preemptively instead of having a static function to take care of the instruction; managing that is really complicated and best suited for either running a virtual CPU on the same type of CPU or running code on a CPU that normally is insufficient to emulate it in real time.

Running multiple CPU in parallel is easy to emulate, it's just very slow when you need to sync them. To speed it up, you can run one CPU ahead of the other(s) + VDP and catch them up at I/O. When you have 5+ things running in parallel though, it's going to be slow no matter what if you need any sort of accuracy. Dreamcast is just too underpowered to emulate Saturn very well. The best/most accurate emulators, specifically NES, (since few other platforms have remotely cycle/pixel accurate emulators) right now require gigahertz PCs to emulate a <2 Mhz CPU/PSG, <6 Mhz VDP accurately. Generally performance comes at the loss of precision.