Cycle Emulation

It would run slower than molasses.

>>4095364
Because actual real programmers aren't full-on autists like Byuu. Anyone else could've fixed that missing chopper shadow on one nobody-cares game without that, but no, he had to go full retard.

>>4095364
At that point the developers mostly weren't thinking in terms of individual cycles. Programming N64 or PS1 was a more "modern" experience than NES or SNES.

>>4095401
This, and also, they used things like SDRAM or RDRAM where you can't get perfect cycle accuracy due to random cycle misses.

>>4095364
Inject N64 on your Wii U.

Hack your Vita for PSX.

>>4095364
Because it takes an i7 just for accurate snes emulation.

Simple answer.

SNES = 3.58MHz 65C816, 1.5 MIPS + PPU
Cycle Accurate SNES emulator = Needs i5 to run flawlessly

PSX = 33.8688MHz R3000 30 MIPS + 66MIPS Geometry co-processor + 32-bit GPU
Cycle Accurate PSX emulator = ?????

N64 = 93.75MHz R4300i 125MIPS + 62.5MHz R4000 micro-code re-programmable RCP
Cycle Accurate N64 emulator = ?????

Even if they could make a cycle accurate N64 or PSX emulator, and I'm not saying that they could, it would require more hardware than what's currently on the market. Just getting the SNES cycle accurate requires a halfway decent modern machine to run smoothly. Keep in mind it's only recently that higan/bsnes was considered runable on your average machine, it used to be strictly for those with high end systems. And that's just a 4MHz processor.

Post-SNES system architectures were powerful enough to

1) use higher level languages to code in (cpu was fast enough that programmers didn't have to worry about clocks, nor could they easily)

and 2) utilize pixel frame buffers rather than tile-based graphics, obviating any need for tightly controlled timing used for special effects (primarily interrupts and HDMA).

Why not just get the real consoles instead of frying your brains with trying to get emulation to work correctly?
It's not like the systems are even expensive, and for expensive games, there's everdrives.

Emulation is a godsend to try games for free, but if you really want to enjoy a game, there's nothing like the real deal.

>>4095506
I'm an original hardwarefag like you but some people prefer to use their PC or something else as an all-in system with a few more options. Besides, even if most of those consoles and games will probably work for a while they might become rare in our lifetime. Having a digital equivalent that *just werks* would be great by then.

https://github.com/tj90241/cen64

>>4095484
SNES gets a bit complicated with multiple CPUs running at different clocks. The base system has the main CPU based off the master clock and the sound CPU running off the sound clock, then there are a bunch of other components that run at different rates derived from those clocks.

When you add things like cartridge coprocessors with their own CPUs shit gets bogged down even more. The big hitters in that department are CX4, SuperFX and that shogi game with the ARM CPU. IIRC first person to run the Yoshi's Island title screen at full speed on bsnes accuracy did so with a heavily overclocked i7 2600k. SA-1 is much easier because it runs off the same clock as the main CPU.

So systems with a single CPU and not much else going on tend to perform much better. Probably the reason we had decent GBA emulation circa 2001, despite it being clocked much higher than SNES+friends.

>>4095504
Good answer.

>>4095506
It's important for preservation purposes.

Friendly reminder that NO emulation will EVER be accurate.

>>4095632
Higan. You could argue that hardware (like your monitor) makes the emulation innacurate, but tgat's on the hardware side of things, not a problem with the emulation itself.

>>4095678
Literally no emulator is accurate, nor will it be. Not even higan.

>>4095632
>>4095680
>no reasons listed

>>4095680
>Literally no emulator is accurate
Higan

>>4095680
Email this to byuu and see how triggered he gets.

>>4095678
>>4095687
Higan is perfect until a bug is found, patched, then it's perfect again. This has happened hundreds of times over the years.

There is a thread on nesdev at this very moment where certain people are discovering new things about the SMP and are insisting that byuu fix his APU cycle emulation to match yet he's losing his shit about it and fighting it non-stop.

The SNES part of Higan is loads better than the competition of which there is practically none, but it is by no means perfect. The rest of the cores are nothing special as far as accuracy goes.

>>4095703
>There is a thread on nesdev at this very moment
source?

>>4095706
https://forums.nesdev.com/viewtopic.php?f=12&t=16140

>>4095697
But not even byuu think Higan is completely perfect. He actually think today's CPU's are still not fast enough for completely accurate emulation

>At the moment, I can't even conceive of a time when I might be able to properly emulate the SA-1's bus conflict manager that inserts stalls into the slave CPU when the master CPU is accessing the same bus. We simply don't have fast enough CPUs for this. And it's not just me saying this, even _Demo_ from ZSNES once mentioned how it would require a ~10GHz CPU to emulate the SA-1 perfectly.

>>4095687
>>4095703
Higan is imperfect. Emulation my definition is imperfect and inaccurate. Higan will never be perfect.

>>4095731
>my definition
lol.

you don't have to keep shitposting. Someone already posted as to why that is, and I agree.

>>4095632
consolers aren't accurate either
a snes jr will behave differently than a first revision snes

>>4095632
It doesn't need to be. I would be happy if N64 emulation was even 1/4 as accurate as the current SNES/NES emulation.

>>4095730
Isn't that bullshit though? The SA-1 only stalls the main CPU whenever they access the same address. You'd just need to check that and idle() the main CPU for a while.

>>4095737
And can't disc drive consoles vary in loading times too? I can see that being a nightmare for speedrunning.

>>4095364
>cycle-accurate emulation post-SNES is a mess all the time

>What are MAME/MESS, Dolphin Gamecube, Virtual Jaguar, Retroarch, mGBA, Visual Boy Advance, and Riivolution 24

>>4095730
>He actually think today's CPU's are still not fast enough for completely accurate emulation
Then he should've worked on a NES cycle accurate emulator, desu.

>>4095747
I remember something about FF9 speedrunners using a PS2 in fast mode because of the much faster loading times. Most speedrunners I've heard of prefer physical hardware anyway, so non-perfect emulation is a non-issue. The only big exception I know of is the TAS crowd and they're autistic about accuracy to the point of being dangerous.

>emu1 is more accurate is more accurate than emu2
because it passes ten more exotic controller tests despite failing nine serious hardware tests

>>4095747
Especially for the Sega CD and Many early PS1 games.

Is Mesen currently the most accurate NES emulator?

>>4095730
>>4095740
Byuu gets autistic about his coding.

I'm pretty sure anyone else could make an emulator with the same level of accuracy while only needing 1/3 the processing power.

>>4095752
Dolphin is not cycle accurate, nor does it claim to be so.

That said, it's accurate enough that Nintendo's servers can't tell the difference, so you can actually buy things off the eShop with Dolphin.

>>4095792
there always hope for the future.

>>4095516
>they might become rare in our lifetime
I wonder when hardware is going to become rare.
Some people are chasing "rare" games now, but maybe it's time to get a spare console or two just in case.

>>4095801
That depends.

CD/DVD based consoles will go first, as replacement disk drives get harder to find, although I'm sure SD card to disk drive adapters will continue to be a thing.

Cart based consoles will be around for a long time though.

>>4095798
No there isn't. Emulation accuracy is decreasing at a rapid rate with every new generation. There's no such thing as accurate emulation as it is, anything past 5th gen is a lost cause.

>>4095904
>no such thing as accurate emulation, anything past 5th gen is a lost cause.
Bullshit and garbage lies. What makes you so sure that anything past 5th gen is a lost cause, huh? Just because its decreasing now doesn't mean its going to be like that forever. Where there's a will there's a way. Ask the people who were able to successfully tackle Sega Saturn emulation and those that cracked the copy protection for games on it. If that can happen for a usually hard pain to program for as the Saturn, why can't the same be said for the PS Vita or Wii U?

>>4095904
>as time goes on hardware becomes more and more standardized, even across vendors
Today at the abi level it's like all x86 and at the api level it's mostly BSD syscalls + calls that are proprietary but often semi-publicly (easily attainable to whoever cares) documented.

Every generation we get closer and closer to requiring only a compatability layer, not full on emulation. It's not like it used to be where everyone made their own special hardware that had 0 public documentation, and anyone could make their own hardware addons as well, not to mention weird hacks/quirks.

Gen 5+ was a lot of POWER variants, while more exotic to most people, and much less standard than today, even that was less crazy than it used to be gen 5-, I say that even with crazy shit like the xenon and cell in mind because while they're special variants they're still based on something people have used and not 100% unique.

>>4095982
>Every generation we get closer and closer to requiring only a compatability layer, not full on emulation.
I should add to this, the biggest issues today seems to be artificial security, not platform specifications. The creators go out of their way to keep secrets, obfuscate as much as possible, and make plenty of systems specifically to prevent reverse engineering.

The biggest challenges today are getting information and circumventing security, not implementing the system itself.

>>4095364
>How come no one has been targeting cycle-accurate emulation for anything post-SNES?

You need over 3ghz on a single core for that accuracy for the SNES

I haven't been keeping up with it but consumer processors are plateauing around 4ghz. I have no idea how much speed the PS or N64 would need in theory. Probably not something we'll see for a long time.

>>4095703
Link it I want to read

Also consider than byuu is the world's foremost super nintendo expert and autist. Even if something worked perfectly he would throw it out to satisfy his autism. I'm totally open to the possibility that those people could just be wrong.

>>4095392
stop posting him, I spent six hours yesterday reading about him

>>4095771
Mesen, PuNES, and one other who's name escapes me right now.

>>4096076
Look below that post genius.

>>4095364
Because all the computers powerful enough to do that are tied up faking climate data. kek

>>4096084
I'm obsessed with him while Chris Chan hasn't been doing anything interesting.

>>4095475

Crap. A cortex a53 can do it.

>>4096104
I was saving my attention span for this 5 pages of autism

Thanks

>>4095703
>There is a thread on nesdev at this very moment where certain people are discovering new things about the SMP and are insisting that byuu fix his APU cycle emulation to match yet he's losing his shit about it and fighting it non-stop.

God fuck you. I just read 5 pages to find out this was a lie.

He didn't blindly accept information he didn't know the quality of and suspicions of a fellow aspie because surprise, empiricism is fucking useful and everything else is dogshit when developing emulators. They sorted that out, he wrote the changes, and they did tests but hit a wall. It's still ongoing. But it looks like the changes will happen.

Also this inaccuracy only affected testing, it doesn't affect a single game.

>>4096343
looks like Terry updated his website today

>>4095561

Dead-end project.

>>4095506

>Why not just get the real consoles instead of frying your brains with trying to get emulation to work correctly?

Spoken like someone who tried emulators once in the early Aughts and then never touched them again. Get with the times, dad; front-ends make LLE emulators dead fucking simple to run and play (game compatibility is a non-issue because they are LLEs).

>>4095789

This. It's been pointed out time and time again he could optimize his fucking code and not lose accuracy with a single game. He refuses to, because his code is supposed to be a self-documenting software reference for the SNES hardware too.

How well it functions as that I could not say. Not my forte.

>>4095816

What about the electrolytics? Aren't they already having trouble with some computers from the late 70's, despite museum-quality storage and exhibition?

>>4095738

Not gonna happen, dude. System architecture is even more complex (and powerful) than the Saturn, and that took Saint Ryphecha to bring about accuracy-focused emulation.

>>4095904
>>4095940
>>4095982
>>4095998

It'll happen. What the emulator runs on may not even by recognizable to us as a "computer" , but it *will* happen.

Just don't count on being alive for some of 'em, ya know?

>>4095451
>Hack your Vita for PSX.
Psp right? Vita didnt have good ps1 support last i checked.

>>4096641
N64 emulation just need a new team to write a new, from scratch emulator for it.

The people behind Project 64 seems to be more interested these days in finding new ways to include shady adwares in their emulator than making actual progress.

The fact that N64 games run better on Dolphin (see Ogre Battle) than a fucking dedicated N64 emulator say it all.

>>4096641
>System architecture is even more complex (and powerful) than the Saturn
People need to stop repeating this. The N64 is not more complicated than the Saturn. It's less complicated than the PS2 as well.

>>4096454
Not really but okay.

>>4095561
Cen is too busy being an alcoholic manlet to ever finish a project.

>>4095364
MESS N64 core is so slow it can't be debugged properly.

>>4095757
>The only big exception I know of is the TAS crowd and they're autistic about accuracy to the point of being dangerous.
This. If console-accuracy is your highest priority, BizHawk is the best solution for a catch-all emulator. Otherwise, this page shows the most accurate emulators for a given platform:
http://tasvideos.org/EmulatorResources.html
Accepted but not preferred tab if you want to see alternative to BizHawk

>>4096638
Can't you replace capacitors?

>>4096504
>Also this inaccuracy only affected testing, it doesn't affect a single game.

Modern higan SNES development in a nutshell. There's no real reason to go above v98 or so where he removed the balanced profile that used a per-scanline renderer that was ~2x faster with only a missing shadow in Air Strike Patrol.