State of the N64 emulation scene

I.S.T. · Post by **I.S.T.** » Thu May 01, 2008 5:12 pm

Nach wrote:
I.S.T. wrote:I see you've come out of your hibernation, Nach.
And I come with much new software.

Excellent to get away from everything for a while, and just code till you drop.

And now I am curious. What'd ya code? Stuff for personal use or something?

And LOL to Deathlike2.

Post by **Nach** » Thu May 01, 2008 10:11 pm

I.S.T. wrote:
Nach wrote:
I.S.T. wrote:I see you've come out of your hibernation, Nach.
And I come with much new software.

Excellent to get away from everything for a while, and just code till you drop.
And now I am curious. What'd ya code? Stuff for personal use or something?

You'll find out when it's ready. Might need another month of work though. Lets hope I can finish it up soon.

Nightcrawler · Post by **Nightcrawler** » Fri May 02, 2008 1:27 pm

Hopefully these new developments will be announced on a working webpage somewhere.

I'm going to start linking people to RHDN for NSRT. Double

Post by **grinvader** » Sat May 03, 2008 10:45 am

Mooglyguy's advancing WAY faster than I expected. This is good news.

mudlord · Post by **mudlord** » Sun May 04, 2008 1:25 pm

Yeah, its really cool.

Though, recently he hit a major speed snag: coverage emulation.

Though that certainly hasn't been researched much, according to MooglyGuy's blog, with it, brings around a 16-32x speed hit. And in his test viewer, rendering one scene in SM64 with low level coverage emulation takes 3 seconds.. Ouch....

declan · Post by **declan** » Mon May 05, 2008 3:37 am

Does anyone know if Pj64 is still being devloped?

Another thing we should incorporate into this thread is the state of the DS emulation scene..... it leaves a lot to be desired!

mudlord · Post by **mudlord** » Mon May 05, 2008 4:07 am

Does anyone know if Pj64 is still being devloped?

It is on haitus at the moment.

No ETA.

3 seconds...... wow.... thats fast.

Yes, though accuracy focused individuals won't care, as long as its accurate emulation, some people don't even care if it runs < 1 FPS, which is what MESS does currently with coverage emulation on.....

Another thing we should incorporate into this thread is the state of the DS emulation scene..... it leaves a lot to be desired!

Martin made some major progress with No$GBA, still, its closed source, so there is no knowing how accurate it really is.

franpa · Post by **franpa** » Mon May 05, 2008 4:30 am

what is coverage emulation?

byuu · Post by **byuu** » Mon May 05, 2008 4:53 am

Yes, though accuracy focused individuals won't care, as long as its accurate emulation, some people don't even care if it runs < 1 FPS, which is what MESS does currently with coverage emulation on.....

It's cool that he made it optional. I do wonder how other emulators get around the problems when coverage emulation is turned off, though. Perhaps that approximation should be the fallback in MESS.

Martin made some major progress with No$GBA, still, its closed source, so there is no knowing how accurate it really is.

I didn't mean to imply it was impossible to gauge that sort of thing, just that it's a lot harder. It becomes a matter of trust and personal tests rather than simply code observation. Not that I really think anyone's going to lie about such things.

neo_bahamut1985 · Post by **neo_bahamut1985** » Mon May 05, 2008 5:23 am

What do you mean by "coverage emulation"??

Gil_Hamilton · Post by **Gil_Hamilton** » Mon May 05, 2008 6:53 am

Seconds per frame? Ouch.

mudlord · Post by **mudlord** » Mon May 05, 2008 9:18 am

what is coverage emulation?

Coverage is used in one of the render modes of the Nintendo 64:
"Use coverage as pixel alpha".
And "coverage as pixel alpha" is used in games like Star Wars Rogue Squadron, so its needed.

For a basic idea of what coverage can entail, read here:
http://www.freepatentsonline.com/7280120.html
http://forum.beyond3d.com/showthread.php?t=32553

It's cool that he made it optional. I do wonder how other emulators get around the problems when coverage emulation is turned off, though. Perhaps that approximation should be the fallback in MESS.

Yeah, I applaud him for doing so. I also congratulate him for attempting coverage as pixel alpha emulation as well. With almost all HLE plugins, shaders are used for colour combining, since they are dedicated for those operations. I heard D3D has a decent fixed function pipeline, and I know OGL's sucks.

Seconds per frame?

Yep, the price you pay for pure LLE

.

Gil_Hamilton · Post by **Gil_Hamilton** » Mon May 05, 2008 9:53 am

mudlord wrote:
Seconds per frame?
Yep, the price you pay for pure LLE .

For now. I assume there's some optimization that can be done.

And, of course, future hardware will run a lot better.

The modern ZSNES would probably run at seconds per frame on a system that ran the oldschool ZSNES builds reasonably fast.
...
Not that I'm bringing my old K6-2 up just to test the theory.

I know moving from a socket 754 Sempron to a Core2 moved me from almost being able to run BSNES at full speed to being able to run it significantly faster than full speed in most circumstances.

Certainly, having it as an option now is a good idea. To further the "when I was an emu kid" analogy...
It's a lot like how the old SNES emulators(that supported transparencies) didn't FORCE you to use a 16-bit graphics mode instead of an 8-bit one. Or how KGEN98 had a "raster effects" toggle.

etabeta · Post by **etabeta** » Mon May 05, 2008 2:46 pm

mudlord wrote:Yes, though accuracy focused individuals won't care, as long as its accurate emulation, some people don't even care if it runs < 1 FPS, which is what MESS does currently with coverage emulation on.....

actually, none of those wip are currently in MESS (but they'll be eventually)

and coverage is not implemented yet even in MooglyGuy's local code (to my knowledge), so those are only educated guesses and there could well be space for optimizations (some are already planned [1]). not that I expect any miracle or full speed, but yet 3 SPF might not be the final result

[1] recently, someone may have noted that MAME has started to exploit multicore and 64bit CPU to improve speed (e.g Midway Seattle and Zeus hardware went from slideshow speed to > 25/30 FPS)... next step seems to be an abstract DRC framework by Aaron Giles which could allow a DRC for RSP... and MESS would inherit any MAME improvement

ZH/Franky · Post by **ZH/Franky** » Mon May 05, 2008 4:56 pm

Hey guys, in how many years do you think we'll be able to get full speed, perfect accuracy ps3 emulation?

Post by **Nach** » Mon May 05, 2008 5:41 pm

Franky wrote:Hey guys, in how many years do you think we'll be able to get full speed, perfect accuracy ps3 emulation?

Probably by 2067, if we make it past 2038. Although Intel's 80 core CPU each running at 20 GHz might make it happen sooner.

I.S.T. · Post by **I.S.T.** » Mon May 05, 2008 8:05 pm

Franky wrote:Hey guys, in how many years do you think we'll be able to get full speed, perfect accuracy ps3 emulation?

Given we don't have perfect accuracy SNES emulators yet...

at least 25 years from now, if not many, many more.

DancemasterGlenn · Post by **DancemasterGlenn** » Tue May 06, 2008 8:25 pm

Nach wrote:
Franky wrote:Hey guys, in how many years do you think we'll be able to get full speed, perfect accuracy ps3 emulation?
Probably by 2067, if we make it past 2038. Although Intel's 80 core CPU each running at 20 GHz might make it happen sooner.

Haha. I'm so glad you're back.

Gil_Hamilton · Post by **Gil_Hamilton** » Wed May 07, 2008 12:17 am

Nach wrote:
Franky wrote:Hey guys, in how many years do you think we'll be able to get full speed, perfect accuracy ps3 emulation?
Probably by 2067, if we make it past 2038. Although Intel's 80 core CPU each running at 20 GHz might make it happen sooner.

Well, we survived 1999. And if we can whip 2.3 dozen nuclear wars, 72.9 alien invasions, AND Lavos all at once, I think we can handle anything.

Post by **odditude** » Wed May 07, 2008 3:38 am

Gil_Hamilton wrote:
Nach wrote:
Franky wrote:Hey guys, in how many years do you think we'll be able to get full speed, perfect accuracy ps3 emulation?
Probably by 2067, if we make it past 2038. Although Intel's 80 core CPU each running at 20 GHz might make it happen sooner.
Well, we survived 1999. And if we can whip 2.3 dozen nuclear wars, 72.9 alien invasions, AND Lavos all at once, I think we can handle anything.

But what happens when all the machines that rendered all those nuclear wars and alien invasions set themselves up the bomb?

creaothceann · Post by **creaothceann** » Wed May 07, 2008 3:16 pm

odditude wrote:
Gil_Hamilton wrote:
Nach wrote:
Franky wrote:Hey guys, in how many years do you think we'll be able to get full speed, perfect accuracy ps3 emulation?
Probably by 2067, if we make it past 2038. Although Intel's 80 core CPU each running at 20 GHz might make it happen sooner.
Well, we survived 1999. And if we can whip 2.3 dozen nuclear wars, 72.9 alien invasions, AND Lavos all at once, I think we can handle anything.
But what happens when all the machines that rendered all those nuclear wars and alien invasions set themselves up the bomb?

We should let them play e.g. tic-tac-toe against themselves, until they're so fed up with it that they want to play a nice game of chess instead.
It's the only way to be sure.

declan · Post by **declan** » Thu May 08, 2008 5:16 am

Forget about ps3 emulation for now.

Do you recon it would be possible to emulate the GameCube with curent chip sets (Anthlon 64 and Core2 Duo)???

EDIT: If we can't even tackle n64, then thered no hope of gamecube.
The Gekko is i think about 0.5 GHz and the Flipper about 160 MHz or something like that.

Deathlike2 · Post by **Deathlike2** » Thu May 08, 2008 5:25 am

declan wrote:Forget about ps3 emulation for now.

Do you recon it would be possible to emulate the GameCube with curent chip sets (Anthlon 64 and Core2 Duo)???

If we're having trouble with N64, what makes you think the GC is viable? You're talking about significantly increased complexity, by leaps and bounds.

Use the brain please.

I.S.T. · Post by **I.S.T.** » Thu May 08, 2008 6:05 am

declan wrote:Forget about ps3 emulation for now.

Do you recon it would be possible to emulate the GameCube with curent chip sets (Anthlon 64 and Core2 Duo)???

EDIT: If we can't even tackle n64, then thered no hope of gamecube.
The Gekko is i think about 0.5 GHz and the Flipper about 160 MHz or something like that.

If Dolphin is to be believed, a higher speed Athlon 64 X2 and Core 2 duos can run GC emulators just fine.

Also, Gekko is 485 MHZ and Flipper is 162, to be exact.

Post by **Nach** » Thu May 08, 2008 12:07 pm

Deathlike2 wrote:
declan wrote:Forget about ps3 emulation for now.

Do you recon it would be possible to emulate the GameCube with curent chip sets (Anthlon 64 and Core2 Duo)???
If we're having trouble with N64, what makes you think the GC is viable? You're talking about significantly increased complexity, by leaps and bounds.

Interesting thing about the GameCube is that it uses a PowerPC processor and an ATI video card.

On a multicore PPC and on a machine with an ATI video card, it may just be possible to actually run the two main processing units directly, and use the other cores for working out the extra sync math and quirks.

Hence why I think emulating the XBox shouldn't require as much horsepower as say the PS2. Since when parts aren't actually emulated, but are able to be ran just like the original (if possible), extra overhead for that component isn't needed.