Kismet

Did you know that USPS changed their shipping costs drastically a few days ago? Everyone that ships small packets for crowdfunding preorders is crying bloody murder. Personally, I don't care about the shipping cost in this one very specific case: Subtotal: $189.99 USD Shipping: $32.35 USD Total: $222.34 USD That is still cheaper than trying to acquire a working SNES.

Because these are relevant to reasons why people would buy a FPGA console, and not simply download illegal copies off the internet and play it on software emulators using they keyboard or xbox360 controller. Again, because elephant in the room is that the Mini NT, Super NT, and essentially every software emulator out there's primary purpose is to play games that can only be obtained from pirates. As has been repeated every few pages, if you can not use the original software (eg the physical cartridge) then it's purpose is piracy, not preservation. There are no physical MSU-1 cartridges, as it only exists as a FPGA, and is only part of the SD2SNES firmware. It's the pirates who generally do not care about what kind of damage they do that often results in things like FPGA consoles and FPGA flash drives being labeled as piracy tools when blobs of pirated material appear designed exclusively for the device. If you want to play with a MSU-1 romhack, you need to obtain the ROM, audio files, and the patch, of which acquiring all three are nearly always copyright infringement and not setup ready to be played on the SD2SNES, let alone emulators. I raise this concern once in a while because people get full of themselves about what kind of things kevtris is going to make available in a JB firmware when he hasn't even announced such a thing. If you could just buy the Mini NT or Super NT jailbroken, it would probably sell easier, but it would get smacked down by Nintendo pretty darn quick, because it would be promoting piracy. If it doesn't pirated games alone, then it's easy to go "well it only plays previously released cartridges", even if it can play everdrive or sd2snes carts, since a real snes can play those too. If at some point it's jailbroken and the pirates can't keep from ruining things, then you will end up seeing JB'd Super NT's on eBay, and then the awareness goes up, and good luck ever buying one at $190 ever again.

Did you read past the first 10 words? a) All MSU-1 content is copyright infringement, period. Good luck finding all the pieces needed to make them work. That is the practical problem with all game mods, that they only work with one specific version of a game that nobody physically has, and can not get by any legal means. Good luck people with PAL systems or non-English NTSC carts, or people who have other romhacks they would want to also use. b) The people doing the mods know this very damn well. So do the pirates, so the pirates just mod the game, release it in that form, and then people will incessently whine about how broken the game is when MSU-1 modded, when perhaps the bugs were fixed if you just patched a virgin rom correctly. Or maybe you really do have a rom dumper, and it turns out you don't have the right version of the game, and patching it instead produces garbage. If you can not see my point, don't bother replying, that goes for you too Wolf_

-EVERYTHING- produced for MSU-1 thus far has been copyright infringement, you're missing the point as always. People are not going to patch things themselves, they just go to the first guy on the internet who pre-patches everything and dumps it in a rompack. Then whatever broken state that patch is in, becomes the one people use and whine about. That is what idiot pirates do, they don't care if a patch was a work in progress or full of bugs. That's why so many broken dumps for all consoles exist in the first place. If you're not willing to dump the games yourself, then anything you acquire off the internet is at your own risk. Streaming/lp'ing from a SNES with the MSU-1 is also probably a reasonable, albeit unweildy way to work around Nintendo flagging videos if you don't use any music from the games. Given, Nintedo just hands out copyright strikes people who play Nintendo-owned games on emulators.

Amazing how many people decided this comment was worth jumping down your throat for. The MSU-1 is basically something that started out as an emulation-only thing, that was later made a standard feature of the SD2SNES, which is what enables all the MSU-1 romhacks to work on a real SNES. If gives a capability that would have been equal to having the BS-X/SuperDisc, and is primarily used to re-implement BS-X features. So yes it's a FPGA-only chip. That said, until the SuperDisc was discovered, it was basically a way to go "look what the SNES could have been capable with a CD-ROM", and then of course it turns out the MSU-1 exceeds what the SuperDisc was capable of. So making the suggestion of it's value is useless is missing the point, as the SuperDisc only has one known working unit out there, and it is impossible for anyone to own a legit SuperDisc either. For all intents kevtris could put the SuperDisc features into Super NT for a proof-of-concept, but anyone working on romhacks would rather target the more capable MSU-1. From a proof-of-concept point of view, the audio part is essentially unlimited because the cartridge expansion directly connects to the analog audio in the SNES, so games that utilize the MSU-1 typically have to hack the game to trigger the MSU-1 audio playback and mute the music that would normally play. This allows some games to actually have their CD OST's be utilized, or other fan-produced cover music to replace the original. That is an interesting use of the MSU-1. But, I digress, yes all the existing MSU-1 stuff is copyright infringement, and the amount of effort required to do the patching yourself is borderline impossible as there is no legal way to obtain all the prerequisite ROM's or audio data. So playing with the MSU-1 is basically going "Yeah this pressing so hard on the fair use and DMCA exception ( http://www.wired.co.uk/article/dmca-game-preservation-exemptions-abandoned-games) but let's not get too pissed off about it." I generally frown on blatent copyright infringement, but this probably one exception where I'm more interested in what people do with the MSU-1 more than I care about the developer's unwillingness to repackage/re-released games. There are games that I'd like to actually re-released with cd-audio, not because I think the original audio was poor, but because it had interesting soundtracks, and it's rather interesting to see how the OST audio fits.

Assuming it's some kind of proprietary thing, they could make it work on all future Retro NT devices. The pedantic thing that keeps happening in this thread is that "nothing" happens, and someone asks a question that kevtris hasn't answered and it goes off into cloud cuckoo land speculation. People so desperately want to believe the Super NT will be everything and the kitchen sink, and if it's not that Analogue will just produce another console next year that will. People have to be reasonable and understand that we're a very tiny market, and there are plenty of competing cheaper retro non-FPGA systems out there that do more, but have high latency and low accuracy as the tradeoff. If the Super NT is not what you want, don't buy it and then sit around waiting for a MD core.

They're reported all over the web, the fact that I have 4 dead systems, of which one is a SFC from Japan, one from the local game shop, one my sister bought years ago, and one from when it was brand new and stored in it's original packaging all have the same "light goes on, black screen" is 80% failure rate from trying to buy one without any information about it's internals. The working one I got of eBay. The SFC I got off eBay died the first day it was powered up. I would not be willing to buy another 5 to get another working one when they go for 75$.

The entire point is that people are going "oh the SNES is reliable" when all signs point to "nope" There are other things that can die, that are independent of the "System powers on, no picture/sound" problem, but most of those are "cold solder joint" or "bad cap" issues, which are aggravated by bad storage conditions, and can be fixed or parts can be salvaged from other dead systems. However as it currently stands, there is no solution or way of troubleshooting a SNES unless you personally acquire it, and it's entirely possible for them to die in transit or die in storage. So the problems do in fact point to an internal problem in the chip package of S-CPU/S-CPU A. The version 1 cpu core has a DMA bug that can crash the system. GPM is the minimum version you'd want to find, from previous work done on this topic https://forums.nesdev.com/viewtopic.php?f=5&t=16031 We figure serial numbers beginning with UN21 to UN26 are the GPM models (mine has a flush eject button, not raised, same as the SHVC-CPU-01's.) My GPM-02 begins with UN213 on a paper label. The two SHVC units that have intact shells have UN117 and UN127 at the beginning. The 1/1/1 SFC has a serial beginning with S103. So the eject button is only a clue for later models I believe, as none of the above have molded eject. The FCC text being molded is however a clue for GPM, as the SHVC models I have have the metalic label where as the GPM has the molded text.

There is no letter for 1's S-CPU goes: S-CPU (1), S-CPU A (2), S-CPU B (2) S-PPU1 goes: S-PPU1 (1) S-PPU2 goes: S-PPU2 (1), S-PPU2 A (2), S-PPU2 B (3), S-PPU2 C (3) If you've opened the system, the number is actually on the chip. CPU: S-CPU A, 5A11-01, 2JF 8F PPU1: S-PPU1, 5C77-01, 2HF 4Y PPU2 B, 5C78-03, 2HF 8Z On the 1/1/1's they all have -01 Here's an example of 1/1/1: http://s18.postimg.org/kkt2ioznt/20150914_165851.jpg No letters on CPU, PPU1 or PPU2 Here's an example of a 2/1/1 in a SHVC-CPU-01: https://i.ytimg.com/vi/IZ-3j4moFQI/maxresdefault.jpg Note S-CPU A but S-PPU2 without letters Here's a 2/1/3 in a SNSP-CPU-01 (PAL): https://i.ytimg.com/vi/mU4xWoD505E/maxresdefault.jpg Note S-CPU B and PPU2 C Here's a 2/1/2 in a SHVC-CPU-01: https://upload.wikimedia.org/wikipedia/commons/a/af/SHVC-CPU-01_F_01.jpg What would be interesting is to figure out which chip productions are the dead ones, but that will be unhelpful for people looking to buy ones they can't open. Just looking at this image, you can see two patterns of chips that are dead http://projectvb.com/nss/cpu.jpg

The GPM-02's are the only models that people seem to have that work of the multi-chip versions. Hence they're the most valuable for testing compatibility of everything. The APU version combines 4 chips into one (the SPC, DSP and ram chips for the APU to form S-APU) The "1chip" models combine the CPU, PPU1, PPU2 into a single chip, but still have S-APU/S-MIX. Apparently an issue on models with S-MIX is that they sometimes explode, which results in them not having sound. So there is also 1-chip models with S-MIX A. Chip arrangements are listed at https://console5.com/wiki/SNES When you look at the revisions of the SNES, a kind of history becomes obvious: SHVC-CPU-01, 1991 release model. Has either 1/1/1 or 2/1/3 versions of the CPU/PPU1/PPU2, issues with the CPU and PPU's became apparent mid-production and thus were revised, these are actually documented. Encoder chip can be BA6592F or S-ENC/S-ENC B (BA6594F) SNS-GPM-01, 1992, S-CPU A and S-PPU2 B, also S-DSP A, LM324 replaces two op-amps SNS-GPM-02, 1993, S-CPU B, S-PPU2 C, S-WRAM A, S-DSP A, LM324, SNS-CPU-RGB-01, 1994, Same as GPM-02 except S-RGB and S-MIX A replace earlier S-ENC video encoder and LM324 audio quad-amp. SNS-RGB-02, 1995, Same as RGB-01 except S-WRAM B, DAC changed to NEC 6379A SNS-APU-01, 1995, S-DSP A, S-SMP and the ram chips combined into a single S-APU chip SNS-CPU-1CHIP-01, have S-CPUN A, S-APU SNS-CPU-1CHIP-02, have S-CPUN A, S-APU SNS-CPU-1CHIP-03, have S-CPUN A, S-APU, S-RGB A (C-SYNC not connected) SNN-CPU-01 (Mini), 1997, S-CPUN A, S-APU, S-WRAM B, S-MIX, NEC 6379A S-RGB units may have "jailbars", and don't amplify the output for YPbPr output, S-ENC models amplify the output so component video mods are different for the two encoders. I don't want to be talking out of my butt for modding the consoles, but the "correct" output only comes out of SHVC-CPU-01/GPM-01/GPM-02 models, where as the RGB/APU/1chip models certainly treat the video signal differently, thus mods would be different. As SHVC-CPU-01 has both 1/1/1 and 2/1/3 models you have to consider that the 1991 models have at least two major versions (I've seen at least one 2/1/1 model in the serial number library, but it may have been repaired) Take a look at http://projectvb.com/nss/logs.htmand see how many times "replaced CPU" comes up. The vast majority are S-CPU A. S-CPU A are all the "2" versions. All models after CPU A are also "2"'s which tells me that B versions are a different production run. Likewise with the PPU2 C. PPU2 comes primarily in 1 and 3's which correspond to PPU2 and PPU2 B. I have not seen a PPU2 A in any photos anywhere on boards, only one photo where it has been removed from the board, so it's very likely that they're rare, or maybe mostly were in units not used in NTSC areas. So from a compatibility/durability point of view it's likely that only CPU B and PPU2 C are "good" versions, but you never know. Warmer climates may make chips deteriorate faster. At any rate. If you have a working SHVC-CPU-01, you have to take care of it. Even a NOS (New Old Stock) model that was kept sealed in it's original box will die the second you turn it on because the problem appears to be age/environment related. (As an additional piece of research, the EURO model, SNSP-CPU-02 is nearly identical to GPM-02 (CPU B, PPU2 C, no APU shield))

SHVC-CPU-01 has BA6592F SNS-GPM-02 has S-ENC or S-ENC B APU-01 (1 chip for the entire APU) appears to have S-RGB All 1-chip models have S-RGB A that I've seen online. I'll also note, that every SHVC-CPU-01 I have has different versions of the CPU/PPU and present/absent C67 cap. So it's not a 100% gaurantee, but research online suggests that BA6592F/S-ENC/S-ENC B are all pin compatible, and S-RGB/S-RGB are pin compatible. In theory you could salvage and swap the encoder chip but the blur/fuzz is actually part of PPU2 so you can't actually swap PPU2 on a 1-chip, but you could on a SHVC-CPU/GPM/APU, not that you'd want to. http://www.chrismcovell.com/gotRGB/snesblur.html

You can't dismiss the fact that I have 3 dead SNES's with that board, plus posts elsewhere on this forum and the broader internet have all pointed to the CPU's on these boards as being the culprit. More to the point the SHVC-CPU-01 SFC that died was a 1/1/1, obviously from the first japanese production run , so that's a problem with at least TWO different CPU revisions. There is no guarantee that you will ever find a working SHVC-CPU-01 in the wild, and I would never recommend buying that model for this point alone, and that's the vast majority of them.

You'll never find a working one however unless you can power it on in person, or have the person you buy it from show a recent video of it working.

That won't save it. A brand new in box SNES SHVC-CPU-01 that has been safely stored in an attic/garage/basement will die without ever being opened. There is something specific to these models where the CPU will just die with age, regardless of it ever being powered on. There's no warning sign. As I mentioned in another thread, I have three dead units and all of them are SHVC-CPU-01, and they all did the "power on LED, no picture" bit. The one that was kept in it's original box and packaging, did not power up after being shipped 500 miles and left in storage for a few years. The SFC one, initially did work, but then died the second time I tried to power it on. And yes, I used the correct power supply for that one. If I hazard to guess, all SHVC-CPU-01 models, regardless of being 1/1/1 or 2/1/3 models have this problem, and you will only know if you have this model by taking the cover off, as it's the only model with the "tin can" APU box. What seems to trigger the death of the CPU is powering it on after being shipped/stored, and it wouldn't surprise me if this was physical defect in the CPU package. GPM-02 however works fine. Which has all the same chips separate, but no RF shield around the box. Basically these models contain all the "best" versions of the CPU/PPU/APU chips before being combined in later models, and finally the 1-chip model. So from a reliability point of view, the 1-chip model (typically the "redesign" or "mini" model) would be the best one. That said, that model has lower compatibility, and lacks RGB capability, so it's only the best if you don't care about either of those, though it does have the best RGB output when modded for it. I can hardly wait for the Super NT to be released.

Back around 1996 or so, I used to play the SNES with a 5.1 system that had hand-built speakers for the front stereo pair. To this day, when I hear sound come out of "built in speakers" and soundbars, I'm like "yuck". That's the problem also currently presented with the FPGA snes projects as well. Because when you have a HDMI output the TV sends the audio to either it's internal audio DSP/DAC's or it may be configured in software to output through HDMI or S/PDIF. Since the SNES is 32Khz, that gives us a similar problem with the 4.5 scale at 1080p, where the target audio device might support 32khz, or might only accept 44.1khz, 48khz or 96khz. I would love to see the option of changing the output rate just like resolution, but that might screw up things that assume 32khz like white noise generation, and the few games that actually have matrix Dolby Prologic surround getting thrown out of phase.

We're really in a transition phase before 8K stuff arrives. Like if a 24" 4K feels kinda silly, imagine an 8K one. Where 4K and 8K are obviously better are home-theater setups which are more common in the United States single family homes. Apartments just do not have the real estate for large panels, and newer condos/townhouses often have wall mounts for 40" panels at most. However audio still plays second to picture resolution, and many people joke about how stereo 44.1khz is enough for everyone, and we've even started moving backwards to soundbars which throw 8 channels into a space that doesn't cast surround sound, and wireless speakers that are beyond terrible.

To date, I have only heard of 1-chip's having lower compatibility, not failure rates. But they're also 4 years newer and produced closer to the end of the production life of the SNES, so it's possible that the 1-chip might last longer in a situation where you can just replace the caps and voltage regulator. I have no content for my 4K IPS monitor, and the local telco which offers 4K TV only offers a few sports channels as 4K, and then uses Netflix app as a 4K upsell point. In all seriousness 4K is better, marginally. In games you see particle effects and higher detailed backgrounds, but most games, the foreground looks no different, and in the case of MMORPG games, typically the UI widgets are designed only for 720p play, and without an option to scale them, they become illegible. Most single player games are designed for the PS4 (non-pro) and to compromise on the Xbox One. So even if you have the 4K models of these consoles, you are mostly getting a "upscaled" experience, not an native one on games that aren't designed to use the 4K real estate. Like as an "early adopter" on 4K tech, I'd rather have waited a little longer, but it was difficult enough just to find a 24" 4K monitor to fit the space of the previous 24" 1920x1200 monitor, so I bought one that met the minimum requirements I needed for work, and decided that if a better 4K comes out, I'd just move this one over to the second monitor at that time. I would probably not recommend getting 4K in anything less than a 32" monitor/tv and to skip it entirely in a laptop. Although one of the nice draws of having it is having the desktops scale nicely (so much so that when you switch back to a native 1080p screen it feels eerie) but if you actually try to use a word processor or web browser without the scaling turned on, it's so tiny that often menu widgets become messed up. Adobe's CS4 software which I still use, is basically unusable at 4K, because the mouse cursors are invisible/wrap strangely, making it impossible to click on things. The thing I find annoying in some software (which includes some games) that use native and non-native (eg QT) widgets is that they are often cropped. So you'll see a right-click menu, where the text is made larger, but the width is fixed at the original size, so you only see half of the menu item's width, and then it wraps over top the next menu item rendering it impossible to read. A lot of problems with 4K are the same problems 1080p has back in Windows XP, when people were still using 640x480 as game native resolutions.

My parents have gone through 4 TV's in the time I've gone through one. They had a CRT, then they had a Philips HDTV, then they had a LG smartTV, now they have a Samsung 55" 4K. This was over about 18 years. The 4K is new this year. The only one that "burned" in anything was the LG, with a spot in the middle that appeared to be discoloring. However their PC screen is horribly full of burning blobs. Likewise my Samsung 1920x1200 screen has burn-in blobs around the top edge and bottom edge, where the bright parts are in the background wallpaper and windows taskbar respectively. So my 4K computer screen replaced that one. I've also replaced a Proview monitor in the same time frame that I used as a second monitor, but it was originally my primary monitor when I initially moved here. So I'd say in most cases, the average life of a screen not replaced for tech reasons is at least 5 years, with them lasting at most 10 before they absolutely will have enough burn in or backlight loss that will make them annoying to use if anything. This is by comparison to a CRT which dims over time with use, even if powered off due to ambient light keeping the phosphors lit. CRT's were far easier to burn in, often due to static text (Eg DOS software, UI widgets in arcade games) being lit up for long periods of time. The problem with Plasma screens has more to do with the underlying technology. Only Plasma screens made between 2006 and 2014 are really any good, with older models having burn in and phosphor loss like CRT's, and production has ceased on all plasma screens for the US market as of 2014.

Maybe there is a .. in the filename? Which if that's the case, that's actually a security/bug issue that should be fixed

This happens with every hot item. However one quick story... People will do this with damaged goods. If you have an item bought outside the return date, you buy a new item, swap the serial numbers, and then return the one that wasn't accepted. Store eats the "defective item" in their shrinkage, and returning person gets a new item. That is why VRA exists. Because most of these items go back to the manufacturer, and the store will not get it's money back on items that aren't "in new condition", this is why Best Buy and such tend to have open-box sales where they just dump all the stuff people returned into a bin and mark it down 40%. Sometimes you find things from 5 years ago in there. I once found a PC game from 1988 in a discount bin in 1998.

There needs to be a demand before there is investment. In the case of the FPGA consoles, a lot of the demand is driven by the 8-bit Nintendo, 16-bit Super Nintendo, 16-bit Sega Genesis (Mega Drive) in the US because these are the only consoles that were popular here, and nostalgia will color that demand. As a side-effect of implementing those consoles, anything else with the same parts (eg 6502 = Apple II, C64, Lynx, PCE. 68K = Amiga, Mac Classic, Z80 = Sega Master System, Game Gear, SG1000, Coleco, Gameboy) becomes much easier to implement. Portable units add additional complexity since you can't just slap a 3.6V lithium ion battery pack and a LCD panel and call it done. In theory, at least, you can reduce the complexity of the video driver if you make it use a COTS LCD panel (eg 320x240 panel would only require line doubling, and window-box the rest of the screen)

That mode select button doesn't do anything except switch it between 6 button and 3 button mode except when a 32X is used. Also only about 19 games supported 6-button mode. Hence my point still stands about trying to run SNES games with a MD controller. Assuming a SNES mode using MD controllers, still requires assuming that the 6-button controller is used, as SNES games typically use all the buttons. If people only have the three button controllers, then they can't play those games. Presumably you could make the FPGA use that select button as the select button for the SNES, but I think it would still be freaking awkward without being able to define the controller mapping at a per-game level. I almost think there is room for 8bitdo to actually produce master-system/mega-drive/atari compatible 8/18 button controllers if that ever gets developed. For the SNES, the only purpose of an 18 button controller would be to play those other cores, not that I think it wouldn't be useful in other contexts, just those number pads are ridiculously bad design (IIRC the coleco actually had a plastic insert overlay for the buttons.) It's a bit ironic that the touch-screen on the WiiU, DS, and 3DS basically are an evolution of that use.

I have a feeling this is not quite what the poster asked for. "Saving stats" (eg high scores) vs, "Saving states" which require saving the ram, registers, etc of everything. It might be a trivial thing to preserve the state of the RAM for games that don't use SRAM when the core is shut down, thus preserving the present state of the high score table, as if it were an arcade machine always powered on. However you'd have to know when to overwrite the RAM when it boots up again, otherwise you don't have the other chips state to synchronize. The SNES would be much harder since the entire sound hardware and external chips may have their own states to preserve. But for older games (eg Atari, Coleco, NES, etc) there's no SRAM in most carts, so you can pretty much snapshot the system ram and then assuming the game doesn't initialize the memory, overwrite it once it gets to a "booted" state. This would of course not work for anything you want to actually preserve the game state (eg a paused screenshot.) Like it would make more sense to save the state of the battery backed RAM in those games, and then be able to bring up a menu that lets you preserve the state of the SRAM so you have unlimited save slots (eg most RPG games) instead of two or three. It could even be done blindly. eg the FPGA just journals every write to the SRAM, so every time you save the game, it saves the SRAM as a new file, and you could bring up a menu and delete the nth oldest ones, or just go back to the previous save by date-time. But saving "state" of all chips is generally not doable, certainly not like a software emulator. I don't see why you'd need two copies of the FPGA to do it, you just need one copy, and a way to log changes to ram/registers, which is slow and bandwidth intensive. Two copies also doesn't permit RNG to work as intended, which is why I don't think that would be the right tree to bark up. I'll give you one example of save state-scumming that fails. When you record a game by logging the input and state of the ram only, if something is generated by RNG, the "state" stack gets smashed, and thus all input past that point is useless. Hence, saving the state in a hardware device gives you the option of freezing the state, and logging the registers, ram, and so forth, but it would still need a companion watchdog chip to do this (eg what runs the GUI), the FPGA couldn't do it itself, because it's in a halted state.

That's the thing. You can make passive cartridge adapters the way to "unlock" that core in the official firmware. Also if people wanted to make their own pin adapters, it would also unlock it the same way. It's not like you'd need a second firmware blob on the pin adapter, that would just make it a pain to fix bugs. Or people could quite literately make a 'everdrive' without the pin adapter, and just wire to the pins that the pin adapter would use, and n/c the rest. It's not like the pinout for the consoles are a big secret. There's also the problem of the SNES and MD do not have the same number of buttons. the SNES has 8, the MD has 4 on their release controllers (7 on Mark II genesis units, which were not compatible with all 4-button games.) So a SNES game would likely be unusable on a MD unit without a controller with at least 8 buttons. A Super NT using SNES/SFC controllers has enough buttons to play most consoles except those that require the coleco/atari number pad. A Mega NT could actually use 9-pin controllers from the Atari, and it's not an issue, though finding original controllers might be difficult. It might be viable to find after-market controllers, or aftermarket jaguar controllers that could be remapped from 15 pin to 9 pin.

It really depends on the business model. Nothing to do with technology. Incidentally, "NT" is commonly known to mean "New Technologies", eg, Windows NT, Zeppelin NT, which basically made the most sense with the NT Mini, because the previous "Analogue NT" used the original NES CPU and PPU. If the business model of Analogue is to flog as many as possible, then using the same PCB with different cartridge slots would be the logical thing to do. However this doesn't make a whole lot of sense from a EE level since all you need to do is implement the cartridge slot with the most pins (eg the NES's 72 pin slot) , and build pin converters for everything else. This is why the crummy Retron and clone devices with more than one slot are so terrible, because they basically solder 3 PCB's together (NOAC, SNES, SEGA MD/MS) with little regard for build quality. The build quality of a lot of the NOAC type clones look like something out of a "daddy has a hobby selling counterfeit nintendos in the basement". Sure outside they look a bit polished, but the inside is sloppy at best. How else do you sell 10$ worth of parts as an $80 console? So you actually want to go in both directions: a) Sell BYOCAG (Bring Your Own Controllers and Games) PnP (Plug and Play) models, that feature the original cartridge and controller ports. b) Sell two-piece Value-Added pin converters. Instead of having to make 30 different pin converters for each BYOCAG, you only make one common connector (eg a keyed PCIe connector) and thus all the pin converters plug into this PCIe-styled connector, and allows you to make much cheaper lower profile pin converters that are all passive. The part that plugs into the base unit goes "hey, device attached to me is a X, switch to that core" So the the base pin converter would know if it's plugged into a A4 or A6 FPGA, and if someone plugged in a SNES or MD cart into a NT Mini, it could just do an "unsupported pin converter D: " message. At any rate that's how I could see that work. I don't see a physical 32X for a Mega NT ever being supported by virtue of that requiring adding an entire set of analog capture inputs solely for this one part to composite it through HDMI (for reference it costs around $300+ for upscaler FPGA hardware.) It might be within the capability of a FPGA add-on (a la SD2SNES) that works in tandem with the base unit, but it might literately require an add on with three FPGA's, one for each CPU and one for the the rest of the hardware. Previously stated here and elsewhere, the upper-limit of present FPGA's would be emulating parts in the 33-50Mhz range, which is within the range needed for the 32/psx/saturn, but as kevtris said, memory bandwidth becomes an issue.