Homebrew CitrAGB - 3DS GBA Emulator (WIP)

[Slushie3DS]

There is no enough Cubic Ninja for everyone, unfortunately

Yeah, the dev team was dismembered, and they stopped producing the game. I had to pay $20 for my copy, and it still has yet to come.

 

[davhuit]

Only 20? The price changed again? Because when smealum said the name of the exploit, the price was like 500 dollars in some places...

If yes, I suppose the new gateway exploit might have played a role in that change of price.

 

[Slushie3DS]

Only 20? The price changed again? Because when smealum said the name of the exploit, the price was like 500 dollars in some places

The price has went down a lot. You can get the cartridge used for about $18. I got lucky and caught a guy that was selling them new, never out of the plastic, for only $20. Considering that my local super market sells older 3DS games for around $5, it is kind of expensive.

 

[Kikirini]

I lucked out and got both of my copies for $10 all together.
Hopefully we can eventually get the homebrew launcher on gateway. It seems like a lot of devs like the 3dsx format.

 

[Slushie3DS]

I lucked out and got both of my copies for $10 all together.
Hopefully we can eventually get the homebrew launcher on gateway. It seems like a lot of devs like the 3dsx format.

I know I do. I've been kind of coding my Atari 2600 emulator in the dark while I wait for my game.

 

[davhuit]

The price has went down a lot. You can get the cartridge used for about $18. I got lucky and caught a guy that was selling them new, never out of the plastic, for only $20. Considering that my local super market sells older 3DS games for around $5, it is kind of expensive.

Nice to know. Most of the things I use are .cia files and don't require it but I'll probably buy one when the emulator will be in a more advanced stage then.

 

[hippy dave]

Yeah, the dev team was dismembered,

Wo, brutal.

 

[Slushie3DS]

Wo, brutal.

Kek, you know what I mean.

 

[FPSRussi4]

Not sure if it's been stated or if it's an error on my end, but I followed the installation instructions on newest build and it keeps kicking me to home menu, saying "There's been an an error and the system will now restart."

 

[Keylogger]

Any news/update about this?

 

[Elrinth]

it crashed for me when i navigated to the folder where I had the game roms.

 

[Chrisser_75]

it crashed for me when i navigated to the folder where I had the game roms.

try to put them into the root path

 

[ubergeek77]

Is this still being developed? When this got released, it all but killed the previous emulator, and that one would at least run at 50% speed. This one hardly runs at all...

 

[Kikirini]

I'd also love to know if this is still alive. I'm really excited for the future of this emulator.

 

[Cube]

Is there already out a .3DSX file for NinjHAX users?

Thanks!

 

[daxtsu]

It clearly says "get the .3dsx file here" under install instructions in the first post, so yes.

 

[Cube]

Thanks!

 

[shinyquagsire23]

Is this still being developed? When this got released, it all but killed the previous emulator, and that one would at least run at 50% speed. This one hardly runs at all...

It's still being worked on (somewhat). Still trying to find a better way to invalidate the instruction cache, because that is the only thing stopping this emulator from running at full speed. Like, seriously. If you look at the demo in the main post, it runs 100% full speed, with sound. The main issue is that while gpsp can cache frequently called routines, it cannot cache new routines before they happen, so games like Pokemon and the like will lag pretty hard in that area due to having a crapton of subroutines.

The problem with flushing the icache is that this can normally only be done in ARM11 kernel mode. Smea's method involves unmapping and mapping a section of memory to force the flush/invalidation, but it's horribly laggy. I was looking at yifanlu's work and I was maybe thinking it could be done via kernel hax or maybe a nop slide or something, idk.

 

[ubergeek77]

It's still being worked on (somewhat). Still trying to find a better way to invalidate the instruction cache, because that is the only thing stopping this emulator from running at full speed. Like, seriously. If you look at the demo in the main post, it runs 100% full speed, with sound. The main issue is that while gpsp can cache frequently called routines, it cannot cache new routines before they happen, so games like Pokemon and the like will lag pretty hard in that area due to having a crapton of subroutines.

The problem with flushing the icache is that this can normally only be done in ARM11 kernel mode. Smea's method involves unmapping and mapping a section of memory to force the flush/invalidation, but it's horribly laggy. I was looking at yifanlu's work and I was maybe thinking it could be done via kernel hax or maybe a nop slide or something, idk.

Thanks. I'm glad to hear it hasn't gone away. It's likely the game I was trying to run was simply poorly coded, too, as every game is different.

Is what you need for the backbone of this emulator solely contained in ninjhax? Do you think you'd get a speed boost if you moved this over to that ARM/Spider/browser hack we saw a few days ago?

 

[shinyquagsire23]

Thanks. I'm glad to hear it hasn't gone away. It's likely the game I was trying to run was simply poorly coded, too, as every game is different.

Is what you need for the backbone of this emulator solely contained in ninjhax? Do you think you'd get a speed boost if you moved this over to that ARM/Spider/browser hack we saw a few days ago?

Depends. With ARM11 kernel access I'd definitely be able to get some good speed boosts, because instruction cache invalidation is just 4 instructions (assuming this isn't what's killing the speed in the first place). After looking at what they have I believe I'll be able to change a few things to get it running as a .cia, although the same cache invalidation issue will still be there which is no bueno. I would prefer to stay away from the browser stuff (or anything which relies on exploitation to gain ARM11 kernel or ARM9 kernel), but it may be needed, idk. Fact of the matter is though, without the ability to actually invalidate the instruction cache, it will be difficult to properly implement dynrec emulators for anything really. It does run full speed though, provided every piece of the program is stowed away in the recompiling caches. This was the case with my demo.gba and as such it runs at full speed, no issues with video or sound at all (and it's beautiful). It *might* be possible to recursively scan the entire ROM and recompile it that way, avoiding invalidation altogether, but that also would be pretty tricky.

EDIT: OK, *used* to run at full speed. Apparently on 4.5 it runs a crapton slower (as in, 50%) for some reason, but I haven't done any code changes so no idea what could cause this regression other than whatever Nintendo was using for mapping memory being slower on 4.5 than on 9.2). Can anyone test the demo on 8.x or 9.x and set a feel for how fast it is on those versions?