FPGA SNES Complete

>Well, it's official, I just committed and documented the last core system feature for my FPGA SNES (direct color mode).

>I guess now it's time to start working on FPGA PSX. :)

>Just kidding, now it's time to start squashing some bugs and implementing other cool features that I've been wanting to add for the longest time!

>In other news, I've been giving this a lot of thought and I'm considering starting a Patreon for this project to see if I can make this thing into a marketable product. There is still a TON of work to do, but I think doing something like Patreon would at a minimum achieve to primary things:

>1) See if there _really_ is as much interest in a product like this as people claim
>2) Keep me motivated to continue working on this project

>But before I go down that route I'd like to hear from the nesdev community what it thinks about doing something like this.

https://m.youtube.com/watch?v=jllSzM24zxc

>>4171859
Great news! I still haven't even saved up enough for an Analog NT Mini.

Just adding to the list of all the cool new stuff I got to get like an Everdrive 64, Wonderswan flash cart, UltraHDMI mod, Gamecube HDMI, Phoebe, that Playstation ODE, Everdrive GB X, that Saturn MPEG card slot drive override, etc. All these wonderful fan made stuff. Yet, I still want a Mini SNES classic for Starfox 2 anyway.

>>4171859
wtf is so exciting about FPGA
i already have a real nintendos

The guy has also expressed interest in working on SuperFX and SA1 for SD2SNES

>>4171930
2Chips are starting to go, 1Chips and Minis will follow.

>>4171930
>>4171937
Additionally if one has a fully working FPGA implementation of the real thing, adding native support for higher resolutions + HDMI without negative effects like lag is possible.

>>4171859
>Hopefully coming to market soon.
Don't hold your breath OP. Once he remembers that Chinese hackers can buy one of these FPGA SNES's and dump the verilog code and then produce clones, he'll probably go back on the idea.

>I definitely have no plans on selling any products. All of my FPGA projects I do purely for three reasons: 1) It's something I enjoy. 2) It improves my FPGA design skills and keeps existing skills sharp. 3) I get to learn new things that I wouldn't have learned otherwise. As far as making it public I probably wouldn't have any issue with that if I could be sure that some company/person wasn't going to steal the code and use it to make money - but let's face it, that's probably exactly what would happen.
>Most importantly, I have no understanding of the legal issues surrounding the release of free cloned Nintendo hardware, cloned WDC processors (816 core), or cloned Sony processors (SPC700). That's a headache that I wouldn't want to deal with.

This is one of those "look what I can do" type situations. At BEST, he might do the SA-1 and SuperFX verilog code, but no promises.

Can you run it like an emulator?

Can someone explain?
What is FPGA? Are these emulator clone consoles?

>>4172150
Feild Programmable Gate Array

It is a chip that can be programed to mimic other chips.

>>4171920
>>4171931
>>4172127
I want to believe. Even if it's irrational, I just want to believe that SNES games like Starfox and Yoshi's Island will be playable on something that's not an emulator or the original console/cartridge and I want to believe that FPGA SNES consoles will be a thing.

I also want to believe that the same would happen with the N64 but something tells me that's a little too optimistic.

>>4172150
It's hardware simulating hardware as opposed to software emulating hardware. In theory it should allow for a much more accurate reproduction of how the real thing works and how the internal components react whilst operational.

Plus you get other advantages like no frame buffer which is completely unavoidable in software emulation.

"more accurate" is a scam marketing term by the people who sell these things and the retards who buy them.

This isn't all that different from bsnes which started claiming to be hardware perfect in 2008, yet a million bug fixes since then still doesn't handle certain obscure behaviours properly in 2017.

Any FPGA is still programmed just like a software emulator is programmed. There will be bugs and they can be patched, but be under no illusion that any hardware emulator is automatically better than a software emulator. Being able to run real carts and accessories natively with no extra effort is a massive plus, but that's one of the few clear ones.

As for this project I've been following it as much as anybody else, but a million red flags went off the moment he mentioned patreon. That platform is designed for content creators, but this guy is aiming at selling a product and there's a big difference. Given all the negativity associated between patreon and emulation communities, it gives the impression he's just out for money.

Take ikari_01 and the SD2SNES. When he finalized his design, he made and sold 100 units out of his own pocket and released the whole thing as open source, rather than begging for crowdfunding.

Time will tell I suppose. I don't like having more clueless idiots buying into the hype and parroting myths about FPGAs being more accurate. It's still an emulator after all.

>>4172446
Of course the FPGA is at the mercy of whoever is programming it but the truth is that an FPGA at least has the POTENTIAL to be more accurate than software emulation every could be.

Whilst it's cool that this guy has supposedly perfected his FPGA SNES, I'd rather just wait to see what kevtris comes up with because I know he won't settle for anything less than perfection and if any bugs do crop up he will fix them immediately.

>>4172448
>POTENTIAL

Bullshit.

The only clear advantage is being able to interact in real-time with peripherals, but a specially designed software emulator could do that if clocked high enough. In all other areas it's dependent on what the implementation supports.

>supposedly perfected his FPGA SNES

He only said he'd implemented the last major component. To put this in perspective, ZSNES has had every major component implemented for close to two decades.

The truth is a good FPGA based emulator is better than a shit software emulator, and a good software emulator is better than a shit FPGA based emulator.

It's very reckless to call something perfect when no implementation even comes close.

>Kevtris

Kevtris is awesome and I'll probably be first in line when the Zimba3000 comes out, but I've always gotten a "good enough" vibe from him when it comes to emulation instead of the mentally-impaired-ultimate-accuracy-at-all-costs-autist vibe that some other devs have.

The only thing I can think of right at this minute is him underclocking his NES core slightly to work with HDMI properly. I can live with something like that, but perfect is perfect and that's not perfect.

>>4172474
>The only thing I can think of right at this minute is him underclocking his NES core slightly to work with HDMI properly

It's worth noting that the Analog NT mini still outputs at the NES authentic 60.098fps over its analog outputs but needs the framerate to be bang on 60fps for most TVs to detect the HDMI signal. I won't work it out because my maths is shocking but I read that this equates roughly to the NT mini running 1 second slower over every ten minutes.

>>4172482
That explanation and the numbers sound spot on from what I remember. IIRC the SNES has a similar non-standard framerate. Despite what some people would claim, one frame every ten seconds isn't something perceivable, so it's a very minor thing in practice.

In theory you could run a genuine SNES and a HDMI clone side by side and the gap would eventually become noticeable. This is autism though, and the ageing oscillators on the real SNES probably have some impact on this.

I think a lot of emulators end up dropping the extra frame every ten seconds to maintain a relatively even 60Hz, whereas others also underclock.

File: 2c02.png (44KB, 1103x958px) Image search: [Google]

44KB, 1103x958px

>>4172446
>>4172474
Fuck you. Learn what an FPGA is before spouting this garbage. You can exactly synthesize any digital circuit specification following proper static and dynamic disciplines. The only thing you can't reproduce are hardware flaws like noisy buses, which are not part of any specification since they violate dynamic discipline, so they are erroneous to emulate anyway.

In other words, you can (and people do) decap and differentially etch an IC to be reverse-engineering, photograph it, and deduce the schematic (for instance http://www.visual6502.org/). Then it is a matter of taking what you find and putting it in terms of a Hardware Description Language. Done correctly, there is no ambiguity. You can EXACTLY reproduce the hardware specification.

I'm going through the motions right now to learn about semiconductor technologies, gain skill in reading NMOS, PMOS, and CMOS circuits straight from the silicon, and reverse-engineering the NES 2c02 PPU, so I can map it to VHDL. Out of the entire NES, the only things that would be impossible to emulate with just an FPGA are analog circuits and DACs, but these are only used in the output stages of the system, and nothing prevents you from incorporating these into your FPGA circuit.

If you disagree with this, that's like saying "it's impossible to look at a physical circuit and deduce it's structure and function".

>>4172474
>mentally-impaired-ultimate-accuracy-at-all-costs-autist
I never understood why people would whine about better accuracy. "good enough" is the attitude that gave us the ZSNES and the endless list of SMW romhacks that only work with that hacky shit. It's not like a more accurate emulator is going to cause anyone to lose anything they already had, so where's the complaint? There's a better option available to those who want it, and it's the worst thing ever? What gives?

>>4172832
>you can decap, etch

You CAN do that, but that's NOT what's being done here. The guy behind this claimed to have made a number of observations that aren't present in conventional emulators, but he hasn't exactly been forthcoming with evidence.

One of the big points I'm pushing is that perfect is the absolute highest you can go. That means a 1:1 implementation with absolutely no room for error. By his own admission, he's implementing each part to fit the behaviour as he imagines it. Nothing terribly wrong with this as it's what almost every emulator author does, but that's not necessarily the perfect reproduction you're claiming. It's definitely possible to achieve near-perfect accuracy without being a precise 1:1 implementation, but it shouldn't be marketed as such.

Even the visual6502 and visual2C02 implementations have had a shitload of systematic bugs due to the resolution of those scans and programmer error.

>I'm reverse-engineering the NES 2c02

It's already been photographed and simulated in visual2c02 by Quietust. If you're using this as a baseline to implement it in VHDL, you're not actually doing any reverse engineering.

I know I'm coming across as some piece of shit, but I'm definitely wrong in saying that a lot of people are overestimating and overselling FPGA emulators. There's also no reason why a software emulator couldn't achieve the same results. Hell, visual6502 and visual2c02 are software emulators.

>>4173454
Who's whining? Those are the best people but they're in very short supply. Even byuu has backed off from that, refusing to emulate SA-1 memory conflicts properly because it would butcher performance. It just so happens there have been a lot of SMW hacks that utilize SA-1 in recent years, so don't be surprised if a decade from now they're discovered not to work on real hardware.

Higan, literally the ZSNES of our generation.

>>4173585
>You CAN do that
I know.

>but that's NOT what's being done here
irrelevant to what FPGAs can do.

>Hell, visual6502 and visual2c02 are software emulators.
that run at 20Hz

>I'm definitely wrong in saying that a lot of people are overestimating and overselling FPGA emulators
I totally agree.

I didn't do the etching, but I am reading the photographs. Just because other people have done it doesn't mean it's not an interesting exercise.

>>4173585
>evidence

His FPGA passes the nintendo hardware test, which othe emulators cannot.

>>4173847
>irrelevant to what FPGAs can do.
A FPGA can simulate whatever you program it to simulate. The guy is programming his to simulate his own interpretation of SNES logic.

>that run at 20Hz
irrelevant to what software emulators can do

IIRC, didn't the Mesen guy port one of them to C# and it runs a lot faster?

>I'm definitely wrong
"not" wrong, fuck me.

>doesn't mean it's not an interesting exercise
Learning all this stuff is a very worthwhile endeavour, but ideally we need more people to figure out the parts we're still missing. It's discouraging for our scene to have so many people at an entry level who move away without making lasting contributions.

File: SNES Test Program.png (1KB, 256x224px) Image search: [Google]

1KB, 256x224px

>>4173870
Here is said test running on bsnes-classic.

Don't spread misinformation.

>>4172127
>Once he remembers that Chinese hackers can buy one of these FPGA SNES's and dump the verilog code and then produce clones, he'll probably go back on the idea
Where is my cheap MIST/MINIMIG then?

>>4174610
There needs to be high enough demand for the Chinese to bother making clones. That's why they make ED clones, since those are high demand. How many people are looking to get a replacement Amiga 500? It has it's fans, but no where near the level of the SNES.

>>4173905
>but ideally we need more people to figure out the parts we're still missing.
If you trace a motherboard with a multimeter, identify the off the shelf parts used, and decap and read the silicon off the custom ICs, you can draw up a full schematic of a system. There is nothing left to uncover.

Granted, with regard to the SNES, I don't know of any images of the decapped chips yet, but it's only a matter of time. I'd do it myself, but I don't know the chemistry well enough to do the differential etching, nor would I really be excited about having to deal with HF acid. On the other hand, a fully speced exact hardware clone of the NES can easliy be made these days.

>>4171930
New HD ready systems.

The AVS has been pretty great to me. I would love to have a SNES FPGA next so I don't have to wait for a HDMI mod for the same price.

From the looks of it he plans to sell it to the community as open source for 400k via crowdfunding.

Think we'll ever see this shit released?

File: 1492847125604.jpg (42KB, 572x303px) Image search: [Google]

42KB, 572x303px

>>4175563
>Open source
>Selling

Yeah, about that.

File: 1502077471321.gif (985KB, 500x211px) Image search: [Google]

985KB, 500x211px

Do people actually believe FPGA based systems are not real hardware?

Nothing is stopping you from taking that FPGA core and fabricating a real chip out of it.

>>4175898
Aside from the monstrous setup costs...

>>4175898
It is still at the mercy of whoever programmed it.

>>4175920
>>4176016
This is relevant to the post, how? He said "real hardware", that it is.

>>4176016
non-argument

>>4174610
>Where is my cheap MIST/MINIMIG then?
MIST is really cheap already, what the fuck are you talking about? The cheapest full system FPGA board there is actually. There are FPGA expansions only that cost even more.

>>4175920
That's it, it's not relevant, both a real chip and the FPGA work exactly the same.
FPGAs are a cheap way to make chips that have low demand.

>>4175920
It's no different than a C64 on a chip or NES on a chip, except you can actually update it over software instead of buying a new one.

>>4176016
I think those things have teams of people behind them replicating the real hardware on a transistor level, X-on-a-chip systems never get this bad rep, why do FPGA systems?

>>4176049
>>4176053
>>4176067
I'm not bashing it, I'm just saying that 100% compatibility and accuracy is still reliant one whoever programmed it.

>>4176067
>why do FPGA systems?

Because fags have no argument whatsoever and want so badly to equate it with software emulation.

>>4176079
I'm surprised this thread hasn't mentioned that FPGAs actually run ARM based emulators for the system. Kek.
(because some FPGAs include an ARM core for general purpose features)

I've heard this shit so much, I think the people who say that actually don't even bait, they just are this dumb.

Why do people want to equate it with software emulation anyways? I mean, I don't really see so much hate against X-on-a-chip devices, who don't they hate those too then?

>FPGAs are capable of being a perfect transistor level copy of the original and therefore more accurate than software emulators

>b..but what about Verilog/VHDL programming errors
>and the fact no-one has ever seen the transistor layout in the original
>and that software emulators are pretty close to perfect already
>and that even today many of the most obscure behaviors still aren't well understood

>NOT AN ARGUMENT
>ALL FPGAs ARE PERFECT
>SOFTWARE EMULATORS AREN'T
>KYS

this thread

>>4176156
>non-argument
Only proving my point.

>>4176156
>>b..but what about Verilog/VHDL programming errors
>>and the fact no-one has ever seen the transistor layout in the original
>>and that software emulators are pretty close to perfect already
>>and that even today many of the most obscure behaviors still aren't well understood
What do you think this is, 2003? I agree slightly with the last one, everything else is bullshit.

>>4176156
>and the fact no-one has ever seen the transistor layout in the original
Source? Even I have several naked chips from the original SNES.

>>4176181
>I agree slightly with the last one, everything else is bullshit.
We know what cause those behaviours, we don't know what's the 100% best way to replicate them.

>>4176186
>Source

I don't recall having seen any high resolution imagery of the internals of any major SNES component and definitely not anything close to suitable for this purpose.

Although I'd be very happy to be proven wrong. If there is something interesting out there, pls show.

I think there were hi-res photos of the DSP-1 when it was decapped, but Google isn't turning up anything helpful.

>>4176210
>I don't recall having seen any high resolution imagery of the internals of any major SNES component and definitely not anything close to suitable for this purpose.
>imagery
Yeah...
Those kind of people have their own hardware for reverse engineering. Like microscopes.

>>4176210
>I don't recall having seen any high resolution imagery of the internals of any major SNES component and definitely not anything close to suitable for this purpose.
maybe because it's illegal to share? get your own

>>4176215
>Those kind of people have their own hardware for reverse engineering. Like microscopes.
this, do people actually believe it's some neets free project he does on his spare time googling pictures?
no, most of the people who develop FPGA cores are actual current (or ex) IC engineers that do it in a team

File: 6502_top_op10x_BF_4677.jpg (3MB, 4677x5097px) Image search: [Google]

3MB, 4677x5097px

>>4176219
>illegal to share

But it's not? Any patents expired years ago and the photo rights belong to whoever took it.

>>4176215
>microscope

Doing it manually is awkward as hell and would take years. Do it that way if there is no other choice. Digital imaging is definitely the way to go.

visual6502 is the perfect example of what I'm getting at.

>>4176224
This one is some guy doing it out of his room in his spare time though. In general you are completely right about the level of professionalism involved and required in projects like this, but our scene probably falls out of that scope because commercial viability of doing so is negligible at best.

>>4176246
>Doing it manually is awkward as hell and would take years. Do it that way if there is no other choice. Digital imaging is definitely the way to go.
Who was talking about manually? I just said they source their own material, nobody said they don't use pictures....

>>4176246
>but our scene probably falls out of that scope because commercial viability of doing so is negligible at best
Vampire/68080 has several high ranking engineers and developers working for it.

>>4176246
>But it's not? Any patents expired years ago and the photo rights belong to whoever took it.
first, you would still be seeing Nintendo in court no matter what, big companies just fuck you even when you're right
second, nintendo's key chips are all custom made inside Japan

>>4176260
>high ranking engineer
like a general?

>>4176273
Yes....

Obviously high ranking as in respected, well known, longtime, important.

>>4176224
>most of the people who develop FPGA cores are actual current (or ex) IC engineers that do it in a team
source: anon's prolapsed asshole

the nice thing about FPGAs is it largely decouples logic design from implementation. it's a lot easier to design a logic schematic than the metal. this is made even easier using hardware description languages (HDLs), because instead of designing components like full adders on the transistor/gate level, you can use syntax like c <= a + b and the circuit is synthesized for you. it dramatically lowers the level of entry into IC development.

>>4176352
>source: anon's prolapsed asshole
see >>4176260

>>4176352
Most good cores for consoles and microcomputers where developed by teams, usually containing actual IC engineers.
It's fairly profitable to work for someone if they are going to commercialise it.

>>4176359
Add to that, obviously we aren't talking about some lesser known cores with shit compatibility but the most popular ones that also found commercial success.