Homebrew Official [5.5.0/1][WIP] LiteNESU - A terrible (userspace) NES emulator

[ShadowOne333]

Wow.
First the Gameboy emulator for Wii U and now this?

Wii U homebrew is sure getting really interesting.
Can't wait for a way to boot Homebrew without relying on the browser exploit. (And HDD support of course)

 

[Phantom64]

Wow.
First the Gameboy emulator for Wii U and now this?

Wii U homebrew is sure getting really interesting.
Can't wait for a way to boot Homebrew without relying on the browser exploit. (And HDD support of course)

Gameboy emulator? What?

 

[DeslotlCL]

Wow.
First the Gameboy emulator for Wii U and now this?

Wii U homebrew is sure getting really interesting.
Can't wait for a way to boot Homebrew without relying on the browser exploit. (And HDD support of course)

wasnt the gameboy emulatpr only for 5.3/5.4? This nes emulator is for 5.5!

Btw fantastic news! But how will this load the roms?

 

[ShadowOne333]

Gameboy emulator? What?

There was a post last week about a guy who made a Gameboy emulator (alpha version) for Wii U.
I will post a link once I find it, I hope it didn't get closed or something.

--------------------- MERGED ---------------------------

Here we go:
http://gbatemp.net/threads/game-boy-emu-for-5-3-2.417294/

That's the one.

 

[DeslotlCL]

There was a post last week about a guy who made a Gameboy emulator (alpha version) for Wii U.
I will post a link once I find it, I hope it didn't get closed or something.

--------------------- MERGED ---------------------------

Here we go:
http://gbatemp.net/threads/game-boy-emu-for-5-3-2.417294/

That's the one.

ya, but thats for 5.3, that isnt news for us in 5.5.x

 

[rw-r-r_0644]

Wow, nice work!
For Mario Bros corruption it may be caused by 2 things:
-ROM Corruption
-emulated RAM corrupted/wrong adress (most likely)

 

[daxtsu]

Right now, it renders each pixel one at a time. As far as I can tell, this is the main reason why everything is so slow (the OSScreen method for putting a pixel onto the screen is called 61k+ times each frame!) I don't really want to add the extra load of upscaling on top of all that until I can find a better rendering method (GX2 maybe? If anyone knows how that works please let me know)

Check out the description on @vgmoose 's game, by storing the coreinit pointers and simply passing those around to your various functions, you can get some drastic speedups.
https://github.com/vgmoose/space

In addition to Kakkoii's suggestion, you might want to look into editing the pixels in a RAM buffer and then sending it to the GPU when you're done, instead of poking thousands of pixels to the screen per frame. That should be significantly faster as well, if the OSScreen functions support it. If not, maybe look into how GX2 works and figure out how to draw a textured rectangle (or two triangles drawn to be like a rectangle), with your pixel data as the texture.

 

[TotalInsanity4]

What are the odds that an ELF version could be forked by yourself/someone else to take full advantage of hardware on 5.3.2 and 5.4?

 

[link270]

Awesome! The more of this type of thing that gets out there the better! Great job!

 

[ploggy]

Emulators..? YAY \O/ Will be trying this too.

Thanks

 

[shibi]

omg this is exciting! i hope more like this come out soon! Kudos to the developer!

 

[QuarkTheAwesome]

But how will this load the roms?

I've got some really overcomplicated instructions on GitHub. I admit it's not the best.

What are the odds that an ELF version could be forked by yourself/someone else to take full advantage of hardware on 5.3.2 and 5.4?

I've only got a 5.5 console so I can't, but if someone else wants to I have no objections. As long as they tell me if they fix one of my bugs ;3

Will definitely start working on those rendering tricks this arvo. Thanks!

--------------------- MERGED ---------------------------

Wow, nice work!
For Mario Bros corruption it may be caused by 2 things:
-ROM Corruption
-emulated RAM corrupted/wrong adress (most likely)

I'm probably gonna start work on a version with a simple debugger once I get rendering sorted out, hopefully that'll tell us which one ;3

 

[Deleted User]

Check out the description on @vgmoose 's game, by storing the coreinit pointers and simply passing those around to your various functions, you can get some drastic speedups.
https://github.com/vgmoose/space

Why did you pass them in a struct to every functions? It looks like it's just a bunch of dynamic pointer to the functions. It looks a bit like the OpenGL extention where you have to get the pointer of the functions before to use them (glBindBufferARB for exemple).

Why not making a kind of dynamic loader that make them available in the global scope of the application at startup? There something that make it impossible in the elf loader?

 

[vgmoose]

Thank you for making this homebrew! I'm surprised at the compactness of the NES emulator code. Cool stuff!

Why did you pass them in a struct to every functions? It looks like it's just a bunch of dynamic pointer to the functions. It looks a bit like the OpenGL extention where you have to get the pointer of the functions before to use them (glBindBufferARB for exemple).

Why not making a kind of dynamic loader that make them available in the global scope of the application at startup? There something that make it impossible in the elf loader?

My only reason is it would crash whenever I tried that! I figured that that may have been the reason they were looked up on each call, so the passing is just a workaround for the global definitions crashing.

 

[Deleted User]

My only reason is it would crash whenever I tried that! I figured that that may have been the reason they were looked up on each call, so the passing is just a workaround for the global definitions crashing.

Yes I guess your code run without the loader so global are unusable, but since there is an elf loader now it should be possible maybe

Edit: The example https://github.com/dimok789/hello_world from dimok789 use this

 

[vgmoose]

Yes I guess your code run without the loader so global are unusable, but since there is an elf loader now it should be possible maybe

Edit: The example https://github.com/dimok789/hello_world from dimok789 use this

So that example is for a HBL ELF, and I've got a branch here that does it that way. As seen here it no longer uses the struct-passing method. However! I haven't tested if this works on 5.5.x or not.

I have a suspicion that exporting the functions like that requires kernel access, but am happy to be proven wrong

 

[QuarkTheAwesome]

After a bit of work, I've made the OSScreen methods global (similar to what @romain337 said) and it definitely seems a bit quicker. I've also cleaned up the drawing helpers I wasn't using, so that all good.
I couldn't upload a binary to GitHub ("something went wrong." I think their error messages are a bit too specific!) so I've attached it below.

Also, if anyone knows their way around GX2 that'd be awesome since it seems to me like OSScreen can't provide me a better option than rendering the pixels one at a time (happy to be proven wrong!).

EDIT: @Kakkoii Done, thanks!

 

[Kakkoii]

After a bit of work, I've made the OSScreen methods global (similar to what @romain337 said) and it definitely seems a bit quicker. I've also cleaned up the drawing helpers I wasn't using, so that all good.
I couldn't upload a binary to GitHub ("something went wrong." I think their error messages are a bit too specific!) so you'll have to compile it yourself for now.

Also, if anyone knows their way around GX2 that'd be awesome since it seems to me like OSScreen can't provide me a better option than rendering the pixels one at a time (happy to be proven wrong!).

You can attach a zip containing the binary to this forum post instead of the github.

 

[QuarkTheAwesome]

Finished a basic debugger here. I guess I'm gonna be spending some time with SMB now ;3
It's probably RAM corruption - I've noticed a few weird things just while making sure the debugger works.

Also, has anyone else noticed that you can't declare a global variable with an initial value properly? It seems like you have to add 1 to the value you actually want.

 

[Relys]

Great work thus far.