Analogue announces the Super Nt for $189 (SNES)

[bigbadboaz]

My understanding: when Nintendo committed to the SNES classic, 8bitdo's original SNES controller design was deemed too close to be sold, and they were issued a cease-and-desist. The new look of the controller was necessary for them to continue selling such a model. I don't believe the receiver has been changed at all - you mention it being "powered by the system" but that was always the case. The new pricing for the controller + receiver is much better than the original separate pricing, which I can only attribute to downward pressure from their own 2.4G offering and the growing retro market in general.

The 2.4G version is exclusively for the Classic machine, and I think they did this simply so they could offer it cheaper (they're only $25, also including the receiver!) given that market. You wouldn't expect buyers of an $80 novelty system to shell out $40 apiece for pads, but the $25 version is a freaking steal. The nice thing about that set is that the receivers are pre-paired, you never have to sync them or touch the receiver again once it's in the machine. Their Bt receivers, while being compatible with many more controllers, are slightly less convenient in the constant sycing/resyncing this leads to. Anecdotally, the 2.4Gs are a hair faster than their original Bt controllers, as well.

They needed to keep making a Bt version as their original and continuing pitch was for these pads to work on much more than just the SNES. If indeed they have improved the lag with the new ones, even better. As I had mentioned a bit earlier, they did in fact improve the D-pad on the 2.4G version versus their original "infringing" Bluetooth pad. Based on the look, I'm hoping it's the same on the new Bts.

[Wolf_]

Byuu lashing out against polygon's article:

https://byuu.org/articles/fpgas-arent-magic/

I kind of suspected some of things Byuu is talking about here, but I don't really know much about FPGAs.

This is my response to Byuu:

1) fpgas emulate hardware, emulating hardware is completely different than software because software runs over incompatible hardware, and an operating system which means it has to account for those differences at the end point. This is what causes software emulation to have lag and require systems many magnitudes of power greater than the originals to have accurate/close to accurate performance meanwhile hardware emulation can run with no performance differences on hardware that is significantly less powerful than what software emulation would require. For example even on a 1080ti with an i7 7700k his emulator doesn't run at the exact timings of a snes and a $180 fpga console does.

2) The fact that you can emulate hardware for better performance and using less resources drastically reducing the cost of emulating any system supported via fpga does not make it more accurate, but it does make it better than software emulation because software emulation still hasn't accomplished perfect reproduction of even 8bit systems. (Which makes fpgas more accurate today even though emulation has the potential to reach that level of accuracy in the future)

3) Regardless of what causes it, software emulation still has added latency and hardware emulation does not.

4) Analogue is selling a product, but if they just gave out their code that only ran on specific hardware and told everyone they could have the code for free but needed to buy all the individual parts and assemble their own consoles how many people do you think would actually prefer that?

5) Emulation - Using fpga technology to emulate hardware is still more accurate than the emulators he listed (higan, BlastEm, gambatte, and mGBA) because all software emulation doesn't have the processing power to exactly replicate a system yet and when it does the hardware needed to accomplish it will be incredibly more powerful and more expensive than what it would take to accomplish the same feat (which for most of those systems is possible today) by fpga hardware emulation.

6) Fpgas - Here he literally says that software emulation can be as good as fpgas but they just need a crap ton more resources and that fpgas are inherently better and use less resources. He also says that "just because the Super Nt is more accurate doesn't mean it is more important than all the effort I put into my emulator". Sorry but just because you are sentimental about your own project doesn't mean people should prefer it over a better alternative.

7) Latency - Here he concedes that fpgas are faster but his counter point is that software emulation could be just as fast if you installed a custom operating system designed to emulate each individual console (which doesn't exist). So basically he is simply saying fpgas are better but emulators could be just as good if they had things that don't exist.

8) Preservation - Here he gets upset and says that Analogue isn't in it for preservation but to make money because they are selling the consoles. My response would be: So what? Do you think everyone can be like you and literally spend their retirement funds and throw away their future for the sake of preservation? In the real world people need to make money to survive and if making a superior product means they need to turn a profit shut up and take my money. As for it being closed source and thus dooming all their work to fade into obscurity just because it is closed source today, doesn't mean they won't release it later. There is nothing wrong with putting food on your table so you can work at the things you love and offering everyone a better experience with things they love as a result.

9) Game Preservation - Here he basically says it is meaningless to preserve the hardware games run on because they will eventually fail. Evidently no one has told him that the sd2snes is still in production, or what the odds are that the Super Nt will be able to play backups just like the Nt Mini.

10) Closing - After he in this same post claims that fpgas have better response times using less resources and complains that just because the Super Nt has less bugs doesn't mean people should ignore the effort he put into his emulator finishes by saying that the Super Nt isn't more accurate. This completely goes against the things he literally just said.

[orange808]

Byuu lashing out against polygon's article:

https://byuu.org/articles/fpgas-arent-magic/

I kind of suspected some of things Byuu is talking about here, but I don't really know much about FPGAs.

This is my response to Byuu:

1) fpgas emulate hardware, emulating hardware is completely different than software because software runs over incompatible hardware, and an operating system which means it has to account for those differences at the end point. This is what causes software emulation to have lag and require systems many magnitudes of power greater than the originals to have accurate/close to accurate performance meanwhile hardware emulation can run with no performance differences on hardware that is significantly less powerful than what software emulation would require. For example even on a 1080ti with an i7 7700k his emulator doesn't run at the exact timings of a snes and a $180 fpga console does.

2) The fact that you can emulate hardware for better performance and using less resources drastically reducing the cost of emulating any system supported via fpga does not make it more accurate, but it does make it better than software emulation because software emulation still hasn't accomplished perfect reproduction of even 8bit systems. (Which makes fpgas more accurate today even though emulation has the potential to reach that level of accuracy in the future)

3) Regardless of what causes it, software emulation still has added latency and hardware emulation does not.

4) Analogue is selling a product, but if they just gave out their code that only ran on specific hardware and told everyone they could have the code for free but needed to buy all the individual parts and assemble their own consoles how many people do you think would actually prefer that?

5) Emulation - Using fpga technology to emulate hardware is still more accurate than the emulators he listed (higan, BlastEm, gambatte, and mGBA) because all software emulation doesn't have the processing power to exactly replicate a system yet and when it does the hardware needed to accomplish it will be incredibly more powerful and more expensive than what it would take to accomplish the same feat (which for most of those systems is possible today) by fpga hardware emulation.

6) Fpgas - Here he literally says that software emulation can be as good as fpgas but they just need a crap ton more resources and that fpgas are inherently better and use less resources. He also says that "just because the Super Nt is more accurate doesn't mean it is more important than all the effort I put into my emulator". Sorry but just because you are sentimental about your own project doesn't mean people should prefer it over a better alternative.

7) Latency - Here he concedes that fpgas are faster but his counter point is that software emulation could be just as fast if you installed a custom operating system designed to emulate each individual console (which doesn't exist). So basically he is simply saying fpgas are better but emulators could be just as good if they had things that don't exist.

Preservation - Here he gets upset and says that Analogue isn't in it for preservation but to make money because they are selling the consoles. My response would be: So what? Do you think everyone can be like you and literally spend their retirement funds and throw away their future for the sake of preservation? In the real world people need to make money to survive and if making a superior product means they need to turn a profit shut up and take my money. As for it being closed source and thus dooming all their work to fade into obscurity just because it is closed source today, doesn't mean they won't release it later. There is nothing wrong with putting food on your table so you can work at the things you love and offering everyone a better experience with things they love as a result.

9) Game Preservation - Here he basically says it is meaningless to preserve the hardware games run on because they will eventually fail. Evidently no one has told him that the sd2snes is still in production, or what the odds are that the Super Nt will be able to play backups just like the Nt Mini.

10) Closing - After he in this same post claims that fpgas have better response times using less resources and complains that just because the Super Nt has less bugs doesn't mean people should ignore the effort he put into his emulator finishes by saying that the Super Nt isn't more accurate. This completely goes against the things he literally just said.

I like turtles.

[bobrocks95]

You read or interpreted some of byuu's post incorrectly, but yeah I'm not going to break down that wall of text.

[FBX]

Byuu is a bit salty over the subject, but it's understandable. He put in 13 years into his software emulation, and I was there from the beginning cheering him on back in the days when we were hanging around the ZSNES forums. However, I think this is making him exceptionally negative over shit that really doesn't matter. Just from my limited knowledge from Kevtris's explanations of how FPGA programming works, it seems like byuu should want to try FPGA programming himself and embrace the concept. You know, make Higan into an FPGA machine that can be touted as being even better than the Super Nt. But to worry about comparisons to his emulation software is pointless. You have a small device that hooks up in HDMI and plays Super NES games lagless with excellent accuracy, and then you've got a software program that you can load on a powerful computer to get 100% accuracy, though with maybe some lag. That's all there really is to say about it.

[Wolf_]

You read or interpreted some of byuu's post incorrectly, but yeah I'm not going to break down that wall of text.

Then you have nothing to add.

[DirkSwizzler]

I like turtles.

I'm glad someone is saying what needs to be said.

[GigaBoots]

I agree 100% with FirebrandX on this one. It's a shame that he's this upset over it, but the representation of emulation in that Polygon write-up is not completely accurate (disingenuous from his perspective.)

I think it's bullshit to insinuate that the CEO of Analogue doesn't care about posterity because he charges for a product. That's a very radical notion. It's not like Analogue is churning out Retron's or anything.

[tjstogy]

You read or interpreted some of byuu's post incorrectly, but yeah I'm not going to break down that wall of text.

Then you have nothing to add.

Some people post on forums just to keep their post count up, just ignore him.

[bigbadboaz]

The overarching point of Byuu's entire post is that "professional" sites such as Polygon (bleargh) are glossing over a distinction they clearly don't understand, and that the simple use of an FPGA != automatically mean a superior/more accurate experience. Both implementations are 100% dependent on the quality of their authorship.

There's really nothing to argue against with this.

[tjstogy]

The overarching point of Byuu's entire post is that "professional" sites such as Polygon (bleargh) are glossing over a distinction they clearly don't understand, and that the simple use of an FPGA != automatically mean a superior/more accurate experience. Both implementations are 100% dependent on the quality of their authorship.

There's really nothing to argue against with this.

Meh...

No one is arguing that FPGA is inherently better than emulation, instead... that it's individual chip emulation & played on real hardware... so it equates to a same experience as when playing a real cartridge on real hardware as long as the chips are properly reproduced (emulated).

With PC based emulation, it's 100% dependent on the computer's hardware...so it's always: "YMMV"

[bigbadboaz]

...as long as the chips are properly reproduced (emulated).

You just restated exactly what I said, and Byuu was saying.

[Wolf_]

...as long as the chips are properly reproduced (emulated).

You just restated exactly what I said, and Byuu was saying.

Byuu's point is incorrect. While the accuracy of the fpga or emulator obviously matters a near perfect fpga is still better than a near perfect emulator because of the timings and resources it takes to run a fpga. That is how you can have a fpga that runs the snes and only costs $180 but to run Byuu's accuracy core at full performance you are looking at a $800+ computer.

His point that emulation has the potential to be as good as fpgas is correct but also irrelevant because that has no impact on the best way to enjoy snes games today, or for the next several years at least (by my estimation).

[Chocograph]

He's so salty. It's just funny. The point NEEDED to be made that whatever has just been released isn't better than whatever he put out.

[tomwhite2004]

That is how you can have a fpga that runs the snes and only costs $180 but to run Byuu's accuracy core at full performance you are looking at a $800+ computer.

A relative wanted to play some snes games on an old second hand dell optiplex that cost £100. The accuracy core that was just added to retroarch ran full speed for everything thrown at it.

[Wolf_]

That is how you can have a fpga that runs the snes and only costs $180 but to run Byuu's accuracy core at full performance you are looking at a $800+ computer.

A relative wanted to play some snes games on an old second hand dell optiplex that cost £100. The accuracy core that was just added to retroarch ran full speed for everything thrown at it.

Huh, I last tested out the accuracy core about a year ago on a system that was 3-4 years old but was actually about $800 when I bought it and it hardly ran at all. Besides even if you did run it at full performance it still didn't run as well as the fpga console would because the fpga console has exact timings of the original console and is lagless. At best an emulator can get close enough to that where it is difficult to notice the differences. It is simply superior replication technology and that really is the bottom line.

[tomwhite2004]

His point that emulation has the potential to be as good as fpgas is correct

At best an emulator can get close enough to that where it is difficult to notice the differences.

Hmmmmm.

Huh, I last tested out the accuracy core about a year ago on a system that was 3-4 years old but was actually about $800 when I bought it and it hardly ran at all.

I can't believe any desktop Intel CPU from a the last 3-4 years would have any problem with the accuracy core. I have an i3-6320 which is over two years old now and it barely uses half a core on taxing games. If people bought an AMD CPU in that time period expecting it to be competent for emulation then that's their fault for not doing their homework.

I put these quotes that Byuu posted on Reddit on other forums but think they need to been seen more, may dilute that Analogue marketing cool aid people have clearly been drinking by the bucket load.

"There are things I know aren't perfect, and I have first-hand knowledge that those things aren't emulated by the Super Nt, either. Kevtris and I talked about them and agreed we lacked the knowledge and capability to emulate them."

"They're roughly equally accurate. Kevin will have better PPU emulation, I'll have better CPU, SMP, DSP emulation."

And on that bombshell.......

[mufunyo]

4) Analogue is selling a product, but if they just gave out their code that only ran on specific hardware and told everyone they could have the code for free but needed to buy all the individual parts and assemble their own consoles how many people do you think would actually prefer that?

1. OSSC.
2. OSSC Wolf Edition.

QED.

[bobrocks95]

You read or interpreted some of byuu's post incorrectly, but yeah I'm not going to break down that wall of text.

Then you have nothing to add.

Some people post on forums just to keep their post count up, just ignore him.

[Fudoh]

My Life in Gaming review has been released: https://www.youtube.com/watch?v=d_OW_t9RXEM

Sounds great. Kevtris on video is sometimes bizarre because he makes it sound so easy

[Wolf_]

4) Analogue is selling a product, but if they just gave out their code that only ran on specific hardware and told everyone they could have the code for free but needed to buy all the individual parts and assemble their own consoles how many people do you think would actually prefer that?

1. OSSC.
2. OSSC Wolf Edition.

QED.

I have no idea what you are trying to say. Are you trying to say the ossc did that because the ossc wolf edition doesn't even exist yet and tbh I kinda doubt it will ever mature since the regular ossc cannibalized the entire reason development on the wolf started. Also no one buys the parts to an ossc and builds their own, it comes premade/gets purchased from people that assemble it. Either way you are buying it from /somewhere/ which just made it an incredible pain in the ass to obtain that took 6 months on a waiting list. No one enjoys that process and it is not beneficial to the community.

His point that emulation has the potential to be as good as fpgas is correct

At best an emulator can get close enough to that where it is difficult to notice the differences.

Hmmmmm.

Huh, I last tested out the accuracy core about a year ago on a system that was 3-4 years old but was actually about $800 when I bought it and it hardly ran at all.

I can't believe any desktop Intel CPU from a the last 3-4 years would have any problem with the accuracy core. I have an i3-6320 which is over two years old now and it barely uses half a core on taxing games. If people bought an AMD CPU in that time period expecting it to be competent for emulation then that's their fault for not doing their homework.

I put these quotes that Byuu posted on Reddit on other forums but think they need to been seen more, may dilute that Analogue marketing cool aid people have clearly been drinking by the bucket load.

"There are things I know aren't perfect, and I have first-hand knowledge that those things aren't emulated by the Super Nt, either. Kevtris and I talked about them and agreed we lacked the knowledge and capability to emulate them."

"They're roughly equally accurate. Kevin will have better PPU emulation, I'll have better CPU, SMP, DSP emulation."

And on that bombshell.......

You quoted me talking about the future, it is quite clear. Currently emulation does not perform as well as fpgas.

The accuracy core must have had major improvements because literally a year ago it was stuttering horrible on a gtx 960.

And even if the accuracy of the emulation is roughly equal the fact is fpgas perform better. Would you rather be accurate with a pistol or a railgun? It isn't a difficult concept to grasp.

[FBX]

A relative wanted to play some snes games on an old second hand dell optiplex that cost £100. The accuracy core that was just added to retroarch ran full speed for everything thrown at it.

I have a hard time believing this is the same exact implementation in Higan. I remember the accuracy core couldn't reached 60fps on my overclocked Q6600. Granted that's an old CPU by now, but it wasn't exactly weak. Try playing SMW2 and check the title screen on that computer and see if it holds 60FPS without dipping.

[Joelepain]

The accuracy core must have had major improvements because literally a year ago it was stuttering horrible on a gtx 960.

The power of your graphic card is absolutely irrelevant in this case scenario...

Try playing SMW2 and check the title screen on that computer and see if it holds 60FPS without dipping.

For SMW2 Higan has to emulate the SuperFX in its "ultimate version". The super nt use cartridge so it doesn't have to.
The accuracy core runs fine on not so powerful cpu for games that don't use co-processors (I can play F-Zero or Super Metroid just fine on my FX8350@stockClocks)

[Wolf_]

The accuracy core must have had major improvements because literally a year ago it was stuttering horrible on a gtx 960.

The power of your graphic card is absolutely irrelevant in this case scenario...

Try playing SMW2 and check the title screen on that computer and see if it holds 60FPS without dipping.

For SMW2 Higan has to emulate the SuperFX in its "ultimate version". The super nt use cartridge so it doesn't have to.
The accuracy core runs fine on not so powerful cpu for games that don't use co-processors (I can play F-Zero or Super Metroid just fine on my FX8350@stockClocks)

I tested it on a 3 year old $800 laptop a year ago with A Link to the Past and it was constantly stuttering to the point where the intro alone was painful to watch. I have a hard time believing the resources were reduced so drastically in a single year that it could run on a $100 junker.

[tomwhite2004]

The accuracy core must have had major improvements because literally a year ago it was stuttering horrible on a gtx 960.

Hate to break it to you but Higan doesnt utilise the GPU, its all about the CPU.

I tested it on a 3 year old $800 laptop a year ago with A Link to the Past and it was constantly stuttering to the point where the intro alone was painful to watch. I have a hard time believing the resources were reduced so drastically in a single year that it could run on a $100 junker.

Because Higan syncs to the audio by default (to be line with the original consoles 60.08hz output) with no vsync, so stuttering is completely normal. This is why having the Higan core in retroarch was such a big deal, it has dynamic rate control so you can sync the emulation (audio and video) to the exact refresh of your screen eliminating the issue.

https://higan.readthedocs.io/en/v104/fa ... -for-higan

I have a hard time believing this is the same exact implementation in Higan. I remember the accuracy core couldn't reached 60fps on my overclocked Q6600. Granted that's an old CPU by now, but it wasn't exactly weak. Try playing SMW2 and check the title screen on that computer and see if it holds 60FPS without dipping.

I texted him to confirm and it has a i5-3470 which has more than double the single thread performance of a Q6600 at stock clocks.

https://www.cpubenchmark.net/singleThread.html

Yoshi's island was amongst the first games I tested as I was aware of the extra overhead that game requires, locked 60fps at all times.

[Joelepain]

I tested it on a 3 year old $800 laptop a year ago

You mean that load of crap Intel is calling "i5/i7" but is at best half of what a desktop cpu at the same price is?

[Joelepain]

Because Higan syncs to the audio by default (to be line with the original consoles 60.08hz output) with no vsync, so stuttering is completely normal. This is why having the Higan core in retroarch was such a big deal, it has dynamic rate control so you can sync the emulation (audio and video) to the exact refresh of your screen eliminating the issue.

https://higan.readthedocs.io/en/v104/fa ... -for-higan

And this +1000.
Strange I didn't think about it and I was playing with that last week...(and all the other options of retroarch to optimize performance/lag like v-sync, hard sync, framedelay, etc...)

[Wolf_]

Because Higan syncs to the audio by default (to be line with the original consoles 60.08hz output) with no vsync, so stuttering is completely normal. This is why having the Higan core in retroarch was such a big deal, it has dynamic rate control so you can sync the emulation (audio and video) to the exact refresh of your screen eliminating the issue.

https://higan.readthedocs.io/en/v104/fa ... -for-higan

And this +1000.
Strange I didn't think about it and I was playing with that last week...(and all the other options of retroarch to optimize performance/lag like v-sync, hard sync, framedelay, etc...)

When I said stuttering I didn't mean a few slips happening from time to time, I meant the audio sounded like someone put the track in an actual blender and I was getting maybe 10 frames at odd intervals a second. It was a completely intolerable experience. If Byuu has made that functional in just a year then good on him but it still doesn't change the fact that the fastest responses he can hope to achieve occur as a software fix ontop of an operating system and the ability to do that seamlessly has not yet matured. Even if it can be "damn good" an fpga is still faster.

[Joelepain]

Then you have a big problem with your configuration because I was already able to run the accuracy core like 4 years ago on a phenom 2 with basic games (these that don't use co-processors).
The problem with the accuracy core for me has always been games with demanding co-processors (superFX, cx4).

[Ikaruga11]

Just watched MLiG's video review of this and it looks really cool. So you can output SNES and GB games with square pixels?

Would you say this is the definitive way of playing SNES and/or GB games? Is it more accurate than the SNES Mini/Jr.?