It's Time


EvilDragon posted on Sep 23 2005 at 11:54 PM said:
Same goes for SNES: Already pretty good on GP32, so why shouldn't it run perfect on the gp2x?

A lot of the top-GP32 developers are interested in the system (some are alreadly fiddling around with it), so we can be sure there will be good stuff on it.

So, why do we need to see more? The display is the same, and I can imagine a fullspeed running SNES on the same display as on the GP32.



Well quite honestly ED I have doubts about SNES being "perfect" on GP2X. Better than GP32 ,certainly, but "perfect"? I consider "perfect" to be running at fs0 WITH good sound and WITH transparencies. I don't think that will happen. It was running slow on the GP32 with crappy sound and NO transparencies, big difference. Both of those things alone require alot of horsepower.

For one thing a dual 333 MHz PSP struggles with some SNES games and that is more powerful than the GP2X.

I would expect SNES to run at FS1 or 2 with sound set on low quality and no transparencies on the GP2X. Playable certainly, perfect no.

Then there is the question of who will make a SNES emu? The original OSNES9x GP32 team has gone to PSP. Ports from linux will run poorly as they aren't optimised. It will take DrMD level ASM core tweaking to get it to run well. This may take some time.
 
Last edited by a moderator:
Besides, Reesy said that if the OSNES9XGP coding was messed with, it could be A LOT better. All we ever got was version 0.3. That's hardly a finished project. The GP32 should definitely be capable of playing SNES about as well as it can play Genesis. DrMD is at about 90-95% perfection. I'd be willing to guess that SNES could hit at least 75% perfection on the GP32. Now why would a machine with 2 processors that are each about 40mhz faster than GP32's SINGLE processor be unable to play perfect SNES? Plus, I remember people talking about different ways of accomplishing transparency... They said something about having the image that's supposed to be transparent made out of straight parallel lines separated by 1 pixel. There was also talk about making the transparent image a checkerboard. They both looked good... isn't there a way to do that with the SNES too, or would that mean someone would have to go thru every image in every game and edit them all? If so, then nevermind, that's not an option.

It seems to me as though the GP32's emulation was limited by a lack of effort, not a lack of power from the machine.
 
Magus 86 posted on Sep 24 2005 at 01:22 AM said:
Now why would a machine with 2 processors that are each about 40mhz faster than GP32's SINGLE processor be unable to play perfect SNES? Plus, I remember people talking about different ways of accomplishing transparency... They said something about having the image that's supposed to be transparent made out of straight parallel lines separated by 1 pixel. There was also talk about making the transparent image a checkerboard. They both looked good... isn't there a way to do that with the SNES too, or would that mean someone would have to go thru every image in every game and edit them all? If so, then nevermind, that's not an option.

It seems to me as though the GP32's emulation was limited by a lack of effort, not a lack of power from the machine.

That depends on how well both of those cores will work to emulate SNES. I remember Squidge saying that there is bus contention with those cores. This will impact using both cores. With both cores it will be faster but how much? All of this is still pretty unknown right now. It is a wait and see kind of thing. You may have NEAR perfect SNES, you may not, it remains to be seen.

No one has touched SNES on the GP32 since .3 and who will port and optimise for the GP2X? No one knows who will do it and how far someone will get. Many times someone will do a quick port that runs ok but slow. But then to get it to run well it becomes alot of work and the coder gives up and abandons it in an unfinished state (like OSNES9XGP) Again too many unknowns, I wouldn't assume anything either way.
 
Last edited by a moderator:
DaveC posted on Sep 23 2005 at 08:03 PM said:
EvilDragon posted on Sep 23 2005 at 11:54 PM said:
Same goes for SNES: Already pretty good on GP32, so why shouldn't it run perfect on the gp2x?

A lot of the top-GP32 developers are interested in the system (some are alreadly fiddling around with it), so we can be sure there will be good stuff on it.

So, why do we need to see more? The display is the same, and I can imagine a fullspeed running SNES on the same display as on the GP32.



Well quite honestly ED I have doubts about SNES being "perfect" on GP2X. Better than GP32 ,certainly, but "perfect"? I consider "perfect" to be running at fs0 WITH good sound and WITH transparencies. I don't think that will happen. It was running slow on the GP32 with crappy sound and NO transparencies, big difference. Both of those things alone require alot of horsepower.

For one thing a dual 333 MHz PSP struggles with some SNES games and that is more powerful than the GP2X.

I would expect SNES to run at FS1 or 2 with sound set on low quality and no transparencies on the GP2X. Playable certainly, perfect no.

Then there is the question of who will make a SNES emu? The original OSNES9x GP32 team has gone to PSP. Ports from linux will run poorly as they aren't optimised. It will take DrMD level ASM core tweaking to get it to run well. This may take some time.


Ok. Now we're going to try to kill a sad rumor. The second cpu on the psp isn't candenced at 333mhz. Even worse, that second cpu is so restricted that homebrew developers are able to use less than 10 percent of it right now.
And the psp have a 3d accelerator, wich is pretty good for today's comercial games, but bad for 2d homebrew or emulation. On the other hand, the gp2x have a 2d accelerator.

Oh, and did we said gp2x have more memory?

gp2x seems to be an emulation paradise compared to the psp. So stop complaining. Then again, we'll have to wait and see. But if you want to compare PSP's specs to the GP2X specs, gp2x is the winner, for emulation and homebrew.
 
Last edited by a moderator:
DaveC posted on Sep 23 2005 at 09:36 PM said:
Magus 86 posted on Sep 24 2005 at 01:22 AM said:
Now why would a machine with 2 processors that are each about 40mhz faster than GP32's SINGLE processor be unable to play perfect SNES? Plus, I remember people talking about different ways of accomplishing transparency... They said something about having the image that's supposed to be transparent made out of straight parallel lines separated by 1 pixel. There was also talk about making the transparent image a checkerboard. They both looked good... isn't there a way to do that with the SNES too, or would that mean someone would have to go thru every image in every game and edit them all? If so, then nevermind, that's not an option.

It seems to me as though the GP32's emulation was limited by a lack of effort, not a lack of power from the machine.

That depends on how well both of those cores will work to emulate SNES. I remember Squidge saying that there is bus contention with those cores. This will impact using both cores. With both cores it will be faster but how much? All of this is still pretty unknown right now. It is a wait and see kind of thing. You may have NEAR perfect SNES, you may not, it remains to be seen.

No one has touched SNES on the GP32 since .3 and who will port and optimise for the GP2X? No one knows who will do it and how far someone will get. Many times someone will do a quick port that runs ok but slow. But then to get it to run well it becomes alot of work and the coder gives up and abandons it in an unfinished state (like OSNES9XGP) Again too many unknowns, I wouldn't assume anything either way.


Well, okay... let me put it this way... Perfect SNES emulation is about 95% possible... but whether or not we'll actually get it depends on whether or not someone actually codes it. I'd do it if I knew how to.
 
Last edited by a moderator:
Cahaz posted on Sep 24 2005 at 01:49 AM said:
Ok. Now we're going to try to kill a sad rumor. The second cpu on the psp isn't candenced at 333mhz. Even worse, that second cpu is so restricted that homebrew developers are able to use less than 10 percent of it right now.
And the psp have a 3d accelerator, wich is pretty good for today's comercial games, but bad for 2d homebrew or emulation. On the other hand, the gp2x have a 2d accelerator.

Oh, and did we said gp2x have more memory?

gp2x seems to be an emulation paradise compared to the psp. So stop complaining. Then again, we'll have to wait and see. But if you want to compare PSP's specs to the GP2X specs,  gp2x is the winner, for emulation and homebrew.

Who is complaining? I am just saying I have doubts. I am not saying it is impossible I am just saying that you can't assume that perfect SNES on GP2X is a sure thing. Who really knows how much of the second core will be useable on the GP2X, it could only give 10% increase as well. There may be bus contention and architecture issues keeping it from being very useful in an emu. It was designed for movie decoding not emus so it is not known yet. Basically what I am saying nothing is a given, SNES may be very doable at near perfect, it may not, it may be somewhere in between we just don't know either way yet.
 
Last edited by a moderator:
I gues the biggest Performance hit at snes and other 2D platform emus is scaleing and transperencys.
This could all be done in Hardware with next to no Speed loss on the original System. But it eats a big portion of speed when you have to do it in Software.

Now on the gp2x we have a 2d Accerlator wich should be able to do many CPU expensive functions in Hardware with little performance loss, this should give a bigger performance boost than a few more mhz would be capable off.
 
Exactly, scaling, transparencies, etc. can all be done through the HARDWARE, so this will NOT slow down the emulation.
And then we have a 200 MHz processor (which is the STANDARD speed, I'm sure it can overclock higher) and a second processor which can also do a bit more work than just scaling and transparencies.

The SNES on GP32 runs pretty fast, if you don't use sound. And sound is only that slow because the sound mixing is still done in C.

So, take for example the first processor for just the emulation and the second one for sound emulating and mixing and the bit hardware scaling and transparencies.

It is PRETTY hard to do, but it is possible. And of you look how well DrMD runs on the GP32, you know it's possible.

As the gp2x also has an ARM processor, you can use the Assembler core from OpenSNES9x, optimize it, and use it for the gp2x SNES emulator.

So we may have to wait a bit - but not for too long, in my opinion.
 
No ones wanting to optimise the SNES code for the GP32 then? I'd of thought it would be a popular one? or is it there's alot of code to work with???
 
Without getting too bogged down in tec specs, the GPX2 is being built with fast 2D emulation in mind

The PSP wasn't

I pre-orederd mine on the basis of being part of a fun community trying to keep the joy of gaming alive

It reminds of the Spectrum scene way back, with guys coding in their bedrooms, only with the [Bush]Internets[/Bush] communication is that much easier between us geeks

If you wanna play FIFA 2456 or constantly dance around $ony's firmaware upgrades then go grab yourslef a PSP fingerprint magnet

If you want to wait til the GPX2 is out and running Scummvm etc...then do just that

:)
 
EvilDragon posted on Sep 24 2005 at 08:48 AM said:
Exactly, scaling, transparencies, etc. can all be done through the HARDWARE, so this will NOT slow down the emulation.
And then we have a 200 MHz processor (which is the STANDARD speed, I'm sure it can overclock higher) and a second processor which can also do a bit more work than just scaling and transparencies.

...

But can it really be used for mode 7 and transparencies? Will both CPU cores be able to communicate fast enough for emulation at a decent framerate? Does this other chip even support transparencies and do they work the same as SNES?

I still say these questions are mostly unanswered until the hardware is out. I wouldn't jump to conclusions either way.
 
Last edited by a moderator:
From the sound of recent sound of the info dump Codesmith gave on the "supposedly leaked" documents (and assuming his margin for error isnt too high and that the PDFs he read from were accurate enough), it seems like the two CPUs will be able to talk to each other just fine, so they can pass instructions back and forth rather quickly and easily, the only question in my mind now is can they pass that information to the screen fast enough for the emulation to be fast. Theoretically if the pipline from the processors to the screen can handle the output I'm not sure it'd even be a noticable problem but if there was a bottle neck it seems like it'd be there.
 
Back
Top