Emulation requests


This might be a bad time to ask this, but is there any chance of a native pandora port of an nes emulator that runs fullspeed?
 
I think the situation will improve regarding optimizing existing emulators if and when a couple things happen:


1. People start making their source code easily accessible and make things as straight forward as possible for other people to build. I think its preferable to host a project on something like github rather than simply making tarballs available of source code because it allows others to see change history and makes it easier to submit patches.


2. More units being available will increase the chances that there are more developers with the skills to do optimizations.
 
The hardware getting more complex (like the GPU) is a problem, yes. There could be some effort spent to understand how to use it in low level, more like it has been on PSP, but that's difficult of course. Still the same methods used to optimize in GP32/GP2X could still be used though, but it's not happening nearly as much as it could be.

Well, the issue is: Most emulators run fine without the need to optimize them.


SNES9x runs perfect for most games, UAE4ALL as well, etc.


While it would be great having optimized stuff here that needs less CPU Power to achieve the same, I think a lot of devs don't see a challenge here.


On the GP2X, it was a challenge to get SNES properly working, on the GP32 it was MegaDrive emulation.


Devs took on that challenge, the community followed their efforts and was happy - but these days, it's not really appreciated anymore.


If a dev takes the challenge, the only thing he achieves is a few FPS more or needing less overclock.


On the GP2X, optimizing meant going from 5fps to fullspeed on SNES, sacrificing a bit of compatibility.


There are only a few systems on the Pandora where such optimization is really needed, like N64, NDS, PSP, DreamCast, PSX.


You and Exophase did PSX and it's almost perfect. N64 core was perfected by Ari, but I think the plugins could be optimized.


No idea if PSP or Dreamcast would even work... but that would be the only challenges I could think of (except for DosBOX / QEmu).


The next-gen consoles are far out of reach, the old ones work fullspeed without optimization... so it's sad, yes, but it seems the amount of work needed to gain only a little is not attractive for devs.
 
I don't think the P2 discussions will motivate the devs to push the P1 to the limit either.


At the end of the day we have a large library of fantastic software on Pandora, and with or without the best optimization I am extremly greatful to the developers who have provided the OS optimization and Emulators. The only thing the community can do to encourage more optmization of things like N64 is to thank the devs for their work and make donations in recognition.


To the original poster, many thanks for your video! I must admit to be being a bit sad that my 1Ghz won't be able to manage N64, I suspect I was being overly optimistic! Performance wise, 800Mhz P1 and 1000Mhz P2 look quite similar I guess, but still some nice improvements in certain games. :)
 
There are only a few systems on the Pandora where such optimization is really needed, like N64, NDS, PSP, DreamCast, PSX.


You and Exophase did PSX and it's almost perfect. N64 core was perfected by Ari, but I think the plugins could be optimized.


No idea if PSP or Dreamcast would even work... but that would be the only challenges I could think of (except for DosBOX / QEmu).

Sounds like a new competition for me ;-)
 
Let's make an emulator competition.


Or simply some bounty for the N64 games we want to play.
 
This has been raised before, but there are some PNDs which deserve a gold star and PCSX is one of those. I'm totally in favour of a gold star bounty system.


If a developer buys a full price Pandora, releases an awesome application which is considered the brass ring of it's type. Give them a refund on their pandora at a minimum.


I'll happily throw money into that pile if ED wishes to organise it.
 
A lot of devs that did tricks like that on the old systems probably have a family now and no time to code.

Well, Slaanesh seems back in action: http://www.gp32x.de...r-gp32-in-2012/


Maybe someone shoud contact him, since he owns a Pandora.

I guess the fact he wants to do a standalone 1943 emu on the GP32 really shows the passion to drive hardware to the max! Just imagine if someone like that would be able to spend time on an N64 emulator.


Anyway, if a bounty was created I pledge EUR50 for work on the N64 emulator.


/C_S
 
Last edited by a moderator:
What would be the expectations for emulation


http://en.wikipedia....o_game_consoles

  • First - Fourth gen should work out of the box on the original.
  • Fifth gen should work on the 1ghz/P2?
  • Sixth gen plus is almost out of the question.


http://en.wikipedia....ld_game_console


is even more fun. Most of these should be able to be emulated.

If we could get fifth gen working, that would be a dream!


Taking into account NGC super mario galaxy requires a dual core (Max supported threads) at 5.0Ghz to play 60fps at 1080p, I am quite confident that there isn't a dev on the planet that could make that game run on a Pandora at playable levels* without framskip. :)


*50fps PAL mode
 
Last edited by a moderator:
I actually only see backwards trend. uae4all got Cyclone pulled out instead of fixing it (the problem is actually in glue code there, not Cyclone itself). Snes9x got asm cores pulled out too instead of fixing them. I'm not aware of any new uses, except maybe FBA, but that came from GP2X too AFAIK.

The hardware getting more complex (like the GPU) is a problem, yes. There could be some effort spent to understand how to use it in low level, more like it has been on PSP, but that's difficult of course. Still the same methods used to optimize in GP32/GP2X could still be used though, but it's not happening nearly as much as it could be.
How many active devs were you to understand well the arm architecture (aka able to write optimized code for it) in these gp2x days ? I can easyly count 6 that are not around here. (oh and I wasnt part of the community at this time so I probably forgot a few in my count). Now we only have left :


- you (notaz) : kernel, sdl, your emulators, fixes in others emus too (enough for one man realy ;)


- MT-H : X-COM releases, theses optimizer filters


- Exophase : pscx gpu plugin


- Ari64 : mips->arm recompiler


And I'm not realy sure I can count the last too as active.


So as always the problem here is : work-force. Without new fresh blood growing your rank, there little chance this to happen. But the chance to get more are thin : working for android can make you some more bucks than here so capables dev go there.


I for one would love to join your rank, but I had barely coding knowledge when I received my pandora and I'm already proud of what I'm able to do. But the optimisations you're talking about are still far away from my league.
 
I actually only see backwards trend. uae4all got Cyclone pulled out instead of fixing it (the problem is actually in glue code there, not Cyclone itself). Snes9x got asm cores pulled out too instead of fixing them. I'm not aware of any new uses, except maybe FBA, but that came from GP2X too AFAIK.

The hardware getting more complex (like the GPU) is a problem, yes. There could be some effort spent to understand how to use it in low level, more like it has been on PSP, but that's difficult of course. Still the same methods used to optimize in GP32/GP2X could still be used though, but it's not happening nearly as much as it could be.
How many active devs were you to understand well the arm architecture (aka able to write optimized code for it) in these gp2x days ? I can easyly count 6 that are not around here. (oh and I wasnt part of the community at this time so I probably forgot a few in my count). Now we only have left :


- you (notaz) : kernel, sdl, your emulators, fixes in others emus too (enough for one man realy ;)


- MT-H : X-COM releases, theses optimizer filters


- Exophase : pscx gpu plugin


- Ari64 : mips->arm recompiler


And I'm not realy sure I can count the last too as active.


So as always the problem here is : work-force. Without new fresh blood growing your rank, there little chance this to happen. But the chance to get more are thin : working for android can make you some more bucks than here so capables dev go there.


I for one would love to join your rank, but I had barely coding knowledge when I received my pandora and I'm already proud of what I'm able to do. But the optimisations you're talking about are still far away from my league.

Which again shows that supporting Android is an important strategy for Pandora.


I don't really understand how Gingerbread runs on our Pandora. Is it running directly on the hardware, or is it going through a hypervisor?


If it is the first one, do we have a chance that 1Ghz units could run N64 via Android? Or is just too clunky?
 
I don't really understand how Gingerbread runs on our Pandora. Is it running directly on the hardware, or is it going through a hypervisor?


If it is the first one, do we have a chance that 1Ghz units could run N64 via Android? Or is just too clunky?

Directly on the hardware. No idea about N64 on Android ;)
 
N64 on android doesnt run that much better than mupen on the pandora, i get slightly better performance on goldeneye but thats clocking upto 1.6ghz so cant imagin running it on the pandora will be as good as it aint going to clock up that high.
 
There are only a few systems on the Pandora where such optimization is really needed, like N64, NDS, PSP, DreamCast, PSX.
You and Exophase did PSX and it's almost perfect. N64 core was perfected by Ari, but I think the plugins could be optimized.


No idea if PSP or Dreamcast would even work... but that would be the only challenges I could think of (except for DosBOX / QEmu).


The next-gen consoles are far out of reach, the old ones work fullspeed without optimization... so it's sad, yes, but it seems the amount of work needed to gain only a little is not attractive for devs.

IMO there are actually lots of consoles that are at least a "maybe", but only with a very efficient implementation and not naive ports (if an emulator even exists).. they're just less common and popular platforms.


For example:


Jaguar/Jaguar CD


3DO


PCFX


Apple Bandai Pippin


Various arcade


Later Macs


GP32


N-Gage


Gizmondo


And to some extent I at least think Saturn emulation could get a lot faster with the right VDP1/VDP2 implementations. And if you're going to even consider PSP or Dreamcast I'd consider NDS to be on a similra level.

Which again shows that supporting Android is an important strategy for Pandora.

I don't know about that. Most of the emulators on Android are just straight ports of emulators from other platforms without any real new optimization.
 
N64 on android doesnt run that much better than mupen on the pandora, i get slightly better performance on goldeneye but thats clocking upto 1.6ghz so cant imagin running it on the pandora will be as good as it aint going to clock up that high.

It was a shame to hear n64 on the 1ghz via Android was still not fast enough. Does this take advantage of the GPU? Is there anyone in the community that might work on this?
 
The people who originally worked on it, like Ari64 and Adventus were not heared of for some time now unfortunately.


The GPU is used for the rendering, but there might be some potential in optimizing the gpu plugin, as the emu seems to be gpulimited a bit.


Stradian wanted to do a new build on fresh codebase, but nothing materialized until now.
 
It was a shame to hear n64 on the 1ghz via Android was still not fast enough.
N64 on the 1Ghz looked quite brilliant natively, noticeably better than on the 600mhz machines which were already getting close.

Really? I thought the YouTube of Zelda MM was a bit choppy, perhaps it was a poor choice of game to test. Can you point me to some 1ghz N64 videos? I expected them to be overflowing, but I'm struggling to gauge what the performance is and judge the fps.


Thanks for taking the time to let me know this WizardStan, I was getting a bit sad about my forthcoming 1ghz, but I hope some of the lucky 20 will be kind enough to share more videos and experiences this weekend. I think they are all too busy having fun to post!


Going a bit OT my 1st batch can clock to 1ghz without issue and never locks. How much lifetime am I sacrificing by doing that do you think?
 
Back
Top