After an entire weekend spent with headaches I've come to the conclusion that the PS Vita version of Minecraft can be ported to the 3DS, and it is an effort I... Am kind of unwilling to undertake. Not myself, no, but I'd gladly pay someone and recruit people. This is a massive undertaking but do hear me out this. In this here post I'll write out a bit of a detailed breakdown of what I have learned from a surface glance or two and what to do to get this thing done.
But first, why? We already have Minecraft on the 3DS! It's a complicated matter, the existing N3DS port is far from ideal, it uses pocket edition as a base before it became bedrock. That means that technically it's inferior, laggier, less optimized and simply feels worse. The problem is exacerbated by who made it. See, the port was a bit of a funky rushjob, originally most likely intended for 4J that plan fell through as 4J denied ARM development, so Mojang had to search until they settled on a shovelware company going by Other Ocean Interactive, based out of Canada. The port is sloppy, and inaccessible to modders. It's optimized poorly, and the source code is unlikely to ever leak. An actually good moddable Minecraft on the 3DS is but a mere dream, in any official manner that is.
Until now. The PS Vita source code for its Minecraft edition is the key to getting a better Minecraft version on the 3DS. First of all. It is well optimized as all derivates or LCE engine are. This naturally gets us closer to the goal of a community made MC port made with an official engine, RAM usage on legacy edition builds is comically low. However the key is the architecture of the Vita version specifically.
The PS Vita runs ARMv7, the 3DS runs ARMv6K, critically the K part. The 3DS features an FPU, which makes Minecraft barely possible. Of course it's far less capable than a vita but, the first order of business I see here is the architectural hurdle. We'd need someone who can strip out all ARMv7 related functions and replace them with more universal ones. In order to run well 4J did a lot of hackjob fixes, hardware specific lines of code that communicate with the CPU directly and critically, NEON systems. NEON is a way to batch compute stuff on ARMv7 as I understand it, it's one of the big painpoints here, much faster, helps with a lot but doesn't exist on the 3DS. Along with that there's also memory alignment, ARMv7 is much looser with it than ARMv6 is which means a lot of memory related instructions are also going to have to get refactored. Yuppie!
Then there's the big boy challenge, the PICA200's renderer. A derivative of OpenGL I believe, probably incorrect but varying sources tell me varying things. It lacks proper shader functionality and is overall bizzare. Of course in this theoretical we are gonna have to redo the renderer anyways, nothing in the legacy source code we have maps cleanly to the 3DS, But nothing is even remotely similar, it's either DirectX or Sony's mess. As I see it there are two routes to take. There's the wrapper, we keep the rendering engine's backend in-tact, and write a translator that reroutes the functions to the PICA200, basically just WINE but specifically for Minecraft for specifically the 3DS. While easier (I think) this comes at the cost of performance since we are adding more computing to the mix for the same result. Then there's the ballsy (and probably necessary) approach, the renderer replacement. We use citro3d to rewrite the rendering backend for the PSV edition entirely and cut out the middleman. This is ideal if we want performance which considering the hardware we definitely need.
In total the plan is:
1. Cut out and dummy out DRM and RAD dependencies, this is gonna render the UI and audio dead but we need the memory this saves, that and the libraries are hard to one by. That and network checks, PSN naturally doesn't exist on the 3DS.
2. Strip out ARMv7 specific code and replace it with standard C++ math, 3DS' FPU or otherwise other hacks. This is tideous, 4J most likely placed hundreds of these around for frame times since it was meant to all run on the same hardware.
3. Redo the renderer, either wrapper it or rewrite it. Both approaches are time consuming.
How long would this take? Hell if I know, weeks to even get a black screen if I had to guess with a few professionals, that which I am not. The project have to keep it under wraps, exactly the opposite of this thread. By that I mean keep the repo private till it's done cooking. Big Microsoft isn't exactly happy to see this whole thing go down.
If this thing we're to happen what'd be the current goal? Right now? Get anything to run, a black screen would be nice. I know for a fact it can be done so what we would need right now is for enough of a dent to be made for the game to compile to .CIA and run without crashing. Then print something, probably the ticks, see if the core engine has a heartbeat so to speak.
Now, if I were to be ambitious I'd say "DM me!" And other stuff of that variety, the truth is, this is probably not very feasible. And not with me as lead, I know jack about C++, and especially the 3DS. However, if you're willing to take a crack at this batshit insane idea I'd gladly try and help you get into contact with someone who would be more interested and technically qualified. I HAVE spent the last weekend scowering the ends of the earth for anyone with the qualifications and I have a few fat "maybe"s waiting on someone to lead the charge. Thank you for reading
But first, why? We already have Minecraft on the 3DS! It's a complicated matter, the existing N3DS port is far from ideal, it uses pocket edition as a base before it became bedrock. That means that technically it's inferior, laggier, less optimized and simply feels worse. The problem is exacerbated by who made it. See, the port was a bit of a funky rushjob, originally most likely intended for 4J that plan fell through as 4J denied ARM development, so Mojang had to search until they settled on a shovelware company going by Other Ocean Interactive, based out of Canada. The port is sloppy, and inaccessible to modders. It's optimized poorly, and the source code is unlikely to ever leak. An actually good moddable Minecraft on the 3DS is but a mere dream, in any official manner that is.
Until now. The PS Vita source code for its Minecraft edition is the key to getting a better Minecraft version on the 3DS. First of all. It is well optimized as all derivates or LCE engine are. This naturally gets us closer to the goal of a community made MC port made with an official engine, RAM usage on legacy edition builds is comically low. However the key is the architecture of the Vita version specifically.
The PS Vita runs ARMv7, the 3DS runs ARMv6K, critically the K part. The 3DS features an FPU, which makes Minecraft barely possible. Of course it's far less capable than a vita but, the first order of business I see here is the architectural hurdle. We'd need someone who can strip out all ARMv7 related functions and replace them with more universal ones. In order to run well 4J did a lot of hackjob fixes, hardware specific lines of code that communicate with the CPU directly and critically, NEON systems. NEON is a way to batch compute stuff on ARMv7 as I understand it, it's one of the big painpoints here, much faster, helps with a lot but doesn't exist on the 3DS. Along with that there's also memory alignment, ARMv7 is much looser with it than ARMv6 is which means a lot of memory related instructions are also going to have to get refactored. Yuppie!
Then there's the big boy challenge, the PICA200's renderer. A derivative of OpenGL I believe, probably incorrect but varying sources tell me varying things. It lacks proper shader functionality and is overall bizzare. Of course in this theoretical we are gonna have to redo the renderer anyways, nothing in the legacy source code we have maps cleanly to the 3DS, But nothing is even remotely similar, it's either DirectX or Sony's mess. As I see it there are two routes to take. There's the wrapper, we keep the rendering engine's backend in-tact, and write a translator that reroutes the functions to the PICA200, basically just WINE but specifically for Minecraft for specifically the 3DS. While easier (I think) this comes at the cost of performance since we are adding more computing to the mix for the same result. Then there's the ballsy (and probably necessary) approach, the renderer replacement. We use citro3d to rewrite the rendering backend for the PSV edition entirely and cut out the middleman. This is ideal if we want performance which considering the hardware we definitely need.
In total the plan is:
1. Cut out and dummy out DRM and RAD dependencies, this is gonna render the UI and audio dead but we need the memory this saves, that and the libraries are hard to one by. That and network checks, PSN naturally doesn't exist on the 3DS.
2. Strip out ARMv7 specific code and replace it with standard C++ math, 3DS' FPU or otherwise other hacks. This is tideous, 4J most likely placed hundreds of these around for frame times since it was meant to all run on the same hardware.
3. Redo the renderer, either wrapper it or rewrite it. Both approaches are time consuming.
How long would this take? Hell if I know, weeks to even get a black screen if I had to guess with a few professionals, that which I am not. The project have to keep it under wraps, exactly the opposite of this thread. By that I mean keep the repo private till it's done cooking. Big Microsoft isn't exactly happy to see this whole thing go down.
If this thing we're to happen what'd be the current goal? Right now? Get anything to run, a black screen would be nice. I know for a fact it can be done so what we would need right now is for enough of a dent to be made for the game to compile to .CIA and run without crashing. Then print something, probably the ticks, see if the core engine has a heartbeat so to speak.
Now, if I were to be ambitious I'd say "DM me!" And other stuff of that variety, the truth is, this is probably not very feasible. And not with me as lead, I know jack about C++, and especially the 3DS. However, if you're willing to take a crack at this batshit insane idea I'd gladly try and help you get into contact with someone who would be more interested and technically qualified. I HAVE spent the last weekend scowering the ends of the earth for anyone with the qualifications and I have a few fat "maybe"s waiting on someone to lead the charge. Thank you for reading
