Version 1 Postmortem & Proposal for future

[Wavicle]

First things first: everybody stop beating up on arachnivore. How we integrate new-comers into the community is an important component of being a healthy community. When possible criticize ideas, not individuals.

I'm sure some of those requirements have shifted over time, and it looks like the project has managed to garner a bit of a community. Still, one of the most glaring problems with the current version is that it ended up with a $350 price tag, which puts it well north of a MiSTER system which can essentially emulate the X16 along with many other systems.

If one is willing to assemble it yourself, there is a $250 kit available also: https://www.vectorheadarcade.com/produc ... mputer-kit

Obviously, the no FPGA rule was eventually dropped, but I don't think the potential of the FPGA was fully realized. For example, as far as I understand, the VERA board is capable of implementing a sound chip making the Yamaha chip a bit redundant.

There has been ongoing discussion about the future of the YM2151. One argument put forth is that VERA audio is good enough.

The two are very different. VERA cannot produce the same sort of FM audio that the YM2151 produces. The FM-synthesized audio is very characteristic of high-end retro audio and in my opinion adds to the feeling of "quality" of the product. My opinion isn't shared by all, and I understand that reasonable people can disagree on this point.

The thing about FPGAs is: They shouldn’t require any deeper understanding, however; they do allow for a deeper understanding of the system. I could put an FPGA with a config ROM and voltage regulator on a PCB with 40 through-hole pins and cover the whole thing with black plastic and you wouldn't even know it's not an authentic 6502 or whatever. However, I could also expose the config ROM so that onece the user learns BASIC then assembly, they can move on to HDL if they want to. It lets them go further down the rabbit hole, but it doesn't force them to.

The kit above does not ship with a YM2151 but instead uses an FPGA-based drop-in replacement. This idea is not anathema within the community. I believe jumping from BASIC to HDL is a little extreme though. What the X16 does bring to the table is a very simple bus design that can be implemented in a handful of logic gates. On the kit version, the individual blocks implemented with the gates are highlighted.

Here are some nubers:
A 6502 core takes up <700 logic elements running over 10 MHz (source)
A 32-bit RISC-V core (a more relevant architecture to learn) can use about 900 logic elements running several hundred MHz (source)
Some crazy people have even managed to squeeze a RISC-V CPU down to <200 Logic elements.

The 6502 is actually pretty hard to implement effeiciently in an FPGA. Other 8-bit designs from Lattice, Xilinx, and Intel use in the 200 Logic element range and run in the 50+ MHz range even on low end hardware.

I'm not quite sure what the point here is. Is a change to the CPU to one that is more FPGA-friendly being proposed?

Anyway, a single FPGA replacing all that logic is 100% the way to go. I think a RISC-V CPU would make more sense than a 6502. I also think there might be better languages than BASIC to start out with, but I haven't really done a survey.

An FPGA-based low-cost system is roughly what is already planned. Swapping the CPU out is not. Replacing the CPU with a modern 32-bit replacement is about as far from the retro aesthetic being sought as one could get. There may be better languages than BASIC, but I think it is important to keep in mind that these old machines did not require one to become intimately familiar with filesystems and file management prior to pounding out code. There aren't many languages that are friendly towards that model. As infuriating as BASIC line numbers are/were, they facilitated this objective.

[BruceRMcF]

If you "know that" the "requirements" weren't ever actually the requirements of the Commander X16, why talk about them as if they were?

It's interesting to know what got the ball rolling, but why do so many people treat the parts of the blog post that were already superseded by the time of the video as if they are were still design targets at the time of the first Dream Computer video, when David explicitly lays out that the Dream Computer pt.1 video was made after months of discussion by hobbyists with a range of interests and experience.

When David explains how the project got started, and quotes the original blog post that has the points that you describe as "requirements", this is in the context of explaining what got the conversion started.

However, at that point in time, "no FPGA" is quite plainly not a "project requirement", because later in that video he describes the Gameduino V1.0 as the proposed heart of the video generator, and the Gameduino is implemented with an FPGA.

...the system described had an FPGA-based video support chip.

Yes, I'm aware of all of this.

If you are aware of some of the facts on the ground, why write as if you were unaware of the facts on the ground?

I still think the project was plagued by what I would call "FPGA-phobia" and specious logic for avoiding MCUs.

Logic?

Group A wants something like X. Group B wants something like Y. Group C wants something like Z. Both X and Y seems like it can be done with the same system, Z seems like it can't be. The market logic is for Groups A and B to join forces and aim for a larger, more robust software ecosystem, and for Group C to go it alone.

The system done by Groups A & B will entail more compromises, and also be supported by a broader range of software. The system done by Group C will have fewer compromises, and a more limited range of software.

But when some people have different preferences than you have, it would be silly to put that down to "specious logic", it is just that different people have different preferences.

As far as "FPGA phobia", the original board design, before the great scope creep pruning, had multiple FPGA's/CPLDs. That was sliced back to one when the design creep was pruned back.

The realistic version of the "no FPGA, no MCU" point from the rough draft of the concept seems to be, "no FPGA when a workable 8bit ASIC alternative exists, no MCU more powerful than the 8bit system that it resides in".

But along the way, people were on Facebook at the time proposing various MCU's, and none of them were easily integrated directly onto an 8MHz 6502 bus. The Propeller was proposed, but the propeller could reside in an I/O slot and be read on the 6502 bus in a single machine cycle. Other MCU's were proposed which had similar issues. Vera sits in an "I/O slot" on an 8MHz 6502 bus with no difficulty.

Now, maybe an RP2040 with its I/O state machines would work for that, but at the point that they settled on the Vera video chip solution, the RP2040 was not yet on the market.

For people who want to emulate their 8bit video display on a 32bit MCU accessed over a serial interface, and to give up on the full range of types of 1980 style sound chips on the X16, the Agon light already exists. It's open source hardware. There's no good reason for the Commander X16 project to change course to pursue that design strategy.

I posted early on in the Facebook group that I thought an entire system could fit on an (at the time) $5 Lattice Ice40 UP5K FPGA with maybe ~$2 ESP32 for expanded I/O and WiFi/BT if desired, but that was shot down by a bunch of people who had never touched an FPGA yet swore such a system was impossible (and were quite rude and arrogant dispite their ignorance on the subject).

How could the Facebook group "shoot down" anything? The Facebook group wasn't designing the system.

In the project itself, a system similar to that was prototyped -- that's the X8. Many people, myself included, were not keen on the fragmentation of the target hardware development environment that would result from having the Gen3 SBC be unable to run many binaries written for the X16.

Also, obviously, getting squeezed into 128KB combined Video and System RAM meant that the X8 was not a very appealing target for a large slice of the software already in development for the X16.

The decision had to be made whether to launch the X8 and then go ahead with Gen 1 of the X16, or go straight to Gen 1 of the X16, and it was the latter decision that was made.

[Wavicle]

I'd be interested to hear you articulate why. Honestly! It seems like David gathered a bunch of requirements and asked people what they miss about retro computing then the community threw that all out the window and decided what's really important is that you can open the computer and point at different black rectangles and say, "That's the 6502 CPU! That's the video chip! That's the RAM! There's the audio chip!".

This project seems like a textbook case of scope creep.

Even if that is so important to you, the Agon project already did that at a fraction of the cost!

I disagree. Where's the ROM on that board? IO controller? Bus decode circuit? How do I build something that expands it and connects directly to the bus? E.g. how do I build a memory expansion card for it? What about a bus-mastering network peripheral?

That system is a 32-bit dual-core 240MHz microcontroller system with a modern Z80 compatible peripheral. My favorite video for it is the one where the designer plays space invaders entirely on the ESP32 on it. If that scratches a person's retro itch, more power to them. It doesn't mine.

[Ser Olmy]

If one is willing to assemble it yourself, there is a $250 kit available also: https://www.vectorheadarcade.com/produc ... mputer-kit

One thing I find slightly annoying (and has made me put the purchase on the back burner, to be honest) is that even if you buy the devkit, you're still not really getting "a computer." You get a motherboard and (in my opinion) a quite mediocre keyboard.

Now, a half-decent mATX case isn't hard to source, but then the system ends up looking like a (boring) PC. I'd very much want it to look like the modern retro computer the X16 arguably is. And as for the keyboard... well, a very good alternative does exist, but because of the PETSCII requirement it's quite expensive.

There has been ongoing discussion about the future of the YM2151.

Well, that one is easy: It has none.

For the life of me, I can't fathom why this chip was chosen. It's been EOL for quite a while, and it's not even that good. And it's certainly not beginner-friendly when it comes to programming!

One argument put forth is that VERA audio is good enough.

I'd reject that outright. The PSG doesn't even compare all that well against the MOS SID, lacking as it does both envelopes and filters. OK, so VERA sound with its multiple waveforms is arguably better than the AY-3-8910, but that's a really low bar.

The FM-synthesized audio is very characteristic of high-end retro audio and in my opinion adds to the feeling of "quality" of the product.

FM sound lacks all bass and cannot be made to emulate percussion. There are plenty of videos on YouTube with YM2161 music from various systems, and frankly it's all pretty rubbish compared to a four-channel MOD with half-decent samples.

There were two reasons why most 8-bit users ditched the C64 in favour of the Amiga: The Amiga's graphics capabilities left the C64's lackluster palette and low resolution in the dust, and the PCM sound of the Amiga was so much better than the SID it wasn't even funny. The VERA doesn't look bad compared to the Amiga in the graphics department, but the PSG is really rather lackluster. And honestly, I find the presence of a single PCM channel with a tiny buffer almost insulting, as we're SO CLOSE to having proper sound.

If someone were to make a daughterboard that fit in the 2161 socket and enhanced the system's sound capabilities with multi-channel PCM and dedicated sample memory, I'd be ordering tomorrow.

I believe jumping from BASIC to HDL is a little extreme though.

I'd agree with that, and I'd also reject the suggestion that RISC-V would be preferable to a 6502. A big selling point of the latter is that it's almost ridiculously easy to get started with, and that fits the X16's "edutainment" profile to a T.

What doesn't really fit though, is Commodore BASIC, and PETSCII for that matter. This was chosen purely for nostalgic reasons, and now we have a system that requires a custom keyboard, doesn't support most international character sets out or the box (forget marketing this to most European countries as an educational tool), and has a BASIC dialect that's quirky and quite slow. The X16 team actually paid a licensing fee for this, so they could have chosen a better option.

[Ser Olmy]

I think your putting the word "Postmortem" in your post title has kind of rubbed a few people here the wrong way.

He probably thought it was common knowledge that a post-completion evaluation of a project is called a "postmortem." I would have made the same mistake, to be honest.

[BruceRMcF]

...

There has been ongoing discussion about the future of the YM2151.

Well, that one is easy: It has none.

For the life of me, I can't fathom why this chip was chosen. It's been EOL for quite a while, and it's not even that good. And it's certainly not beginner-friendly when it comes to programming!

For chiptunes fans of FM chips, the advantage of the YM2151 over the YM2610 are twice as many 4-op channels and IIUC, the second detune, while the advantage of the YM2610 over the YM2151 are the six fixed and one variable ADPCM channels and the noise channel.

But I don't think it was a choice of one of the better FM chips for a system that already had PSG and PCM, but not understanding how the used chip market works ... in other words, chosen by mistake, thinking that the experience that it was easy to get in numbers from the used chip market without problems with fakes was supported by multiple sources, rather than supported by exactly one source with a number of people willing to buy from that single source and sell them to you, so that one bigger project buying up that stockpile drained the inventory.

[DragWx]

One argument put forth is that VERA audio is good enough.

I'd reject that outright. The PSG doesn't even compare all that well against the MOS SID, lacking as it does both envelopes and filters. OK, so VERA sound with its multiple waveforms is arguably better than the AY-3-8910, but that's a really low bar.

One of the things that made me want to register here was the amount of people who were disappointed that the VERA's PSG wasn't just the SID chip again.

The VERA sounds very similar to Konami's VRC6 on the Famicom, and while it's fine to prefer something other than square waves, I suspect most of the "it's not the SID" disappointment comes from there not being good examples of what the VERA's PSG actually sounds like when it isn't just simple beeps. I was actually planning on using only the VERA's PSG in my first few titles, just to see if it as capable as I think it is.

[arachnivore]

Oh my! There is a lot to follow up on!

Thank you, DragWx for filling me in on the Phases and sharing your perspective on the Phase 1 version.
Thank you, Wavicle. I didn't know about the $250 kit. That actually looks a little more appealing. I'm still not sure it's competitive. I think a lot of this could have been implemented on a single Raspberry Pi pico. One core could handle a "Temple"-like OS and the other could handle audio and video.

I also didn't know the YM2151 is an FPGA-based replacement. Is it open source? Could it be implemented on the VERA? Is the VERA open-source HDL? Do we know what FPGA is used on either? Where can I find this info? I couldn't find it on the GitHub page for X16.

Is a change to the CPU to one that is more FPGA-friendly being proposed?

Yes, but it's not essential.

Replacing the CPU with a modern 32-bit replacement is about as far from the retro aesthetic being sought as one could get.

Is it, though? Does the bit-width of the data bus to system RAM really rank so highly on the list of important aspects of the retro computing experience that we're trying to recreate? That's kind-of the point of my OP. It's hard to separate what was actually great about retrocomputing from a sort-of stockholm syndrome syndrome swirled in with the nostalgia.

If you "know that" the "requirements" weren't ever actually the requirements of the Commander X16, why talk about them as if they were?

I thought I did a pretty good job of distilling the original goals of the project. I laid them out pretty clearly. A huge emphasis of the original proposal was recapturing the learning experience of growing up on retro computers. Now it seems to be re-living the glory days of when 6502 assemby was a useful skill. Which is fine, but I don't think the emphasis is on education anymore.

It's interesting to know what got the ball rolling, but why do so many people treat the parts of the blog post that were already superseded by the time of the video as if they are were still design targets

I don't think I fit this straw man. I know the "requirements" stated at 2:44-3:20 of the video were not fully relevant. He says so in the video. I know that. I pointed out as much in my original post.

If you are aware of some of the facts on the ground, why write as if you were unaware of the facts on the ground?

I didn't. Re-read my post if you have to.

Obviously, the no FPGA rule was eventually dropped, but I don't think the potential of the FPGA was fully realized.

Did you miss that bit? I even explain why I brought it up in the first place.

Logic?

Yes, the logic repeated by Wavicle later in this thread:

That system is a 32-bit dual-core 240MHz microcontroller system with a modern Z80 compatible peripheral. My favorite video for it is the one where the designer plays space invaders entirely on the ESP32 on it. If that scratches a person's retro itch, more power to them. It doesn't mine.

The logic is something like: if we use MCUs for something like an I/O controller, the I/O controller ends up being more powerful than the computer itself. That's absurd therefore we need to avoid it.

I don't think that's sound logic. Instead, you make the entire system absurdly expensive so that you can pretend to avoid the absurdity of building an 80s-era computer out of modern parts. It seems like you're targeting a certain level of suspension of disbelief that you're really working on an 80s computer and the more fragile that suspension of disbelief is, the more expensive the system gets.

From my POV, the "Groups A & B" you talk about decided to go maximally fragile suspension of disbelief = maximally expensive hardware so you could point out on a circuitboard, as Wavicle puts it:

Where's the ROM on that board? IO controller? Bus decode circuit? How do I build something that expands it and connects directly to the bus? E.g. how do I build a memory expansion card for it? What about a bus-mastering network peripheral?

As far as "FPGA phobia", the original board design, before the great scope creep pruning, had multiple FPGA's/CPLDs. That was sliced back to one when the design creep was pruned back.

This sounds like you're conflating the problem of scope creep with the use of FPGAs. David mentions this in his videos and all I have to say in response is: This sounds like bad design. I mean, you still ended up with two FPGAs in the design: VERA and the YM2151 clone. My position is that you could have saved a lot of cost by using a single FPGA than multiple and a bunch of socketed components.

How could the Facebook group "shoot down" anything? The Facebook group wasn't designing the system.

Early on it was supposed to be a place to discuss the X16 project. I didn't know of a better place. I posted that I thought a single cheap FPGA could suffice and empower learners beyond assembly if they wanted to continue the learning journey. A lot of people were very rude and hostile to my input, so I left. That's how.

I'm completely unaware of the X8 project. It seems like you're very upset about it. I'd like it if you didn't direct that anger at me.

Also, 128KB combined video and System RAM sounds like it's more in the spirit of restriction than 512kB-2MB of system RAM with video RAM ontop of that. If David can write a whole RTS in less than 64kB, then 128kB sounds luxurious.

Wavicle wrote: ↑Thu Feb 08, 2024 2:20 pm
I believe jumping from BASIC to HDL is a little extreme though.
I'd agree with that, and I'd also reject the suggestion that RISC-V would be preferable to a 6502. A big selling point of the latter is that it's almost ridiculously easy to get started with, and that fits the X16's "edutainment" profile to a T.

RISC-V is ridiculously easy to get started with too. It is a Reduced Instruction Set Computer (RISC) after all. I suppose it's subjective which is easier to learn.

Also, I wasn't advocating jumping from BASIC to HDL. I was suggesting kids learn BASIC (or C or whatever) then assembly, then they could take their journey further by learning HDL. It's a big jump, yes, but it's possible and valuable.

The main shift I've noticed in the original vision vs. what is being made is a focus on education to a focus on adults returning to a nostalgic comfort zone.

[Wavicle]

If one is willing to assemble it yourself, there is a $250 kit available also: https://www.vectorheadarcade.com/produc ... mputer-kit

One thing I find slightly annoying (and has made me put the purchase on the back burner, to be honest) is that even if you buy the devkit, you're still not really getting "a computer." You get a motherboard and (in my opinion) a quite mediocre keyboard.

Now, a half-decent mATX case isn't hard to source, but then the system ends up looking like a (boring) PC. I'd very much want it to look like the modern retro computer the X16 arguably is. And as for the keyboard... well, a very good alternative does exist, but because of the PETSCII requirement it's quite expensive.

Be specific - what would you propose? I see the complaint about a boring PC and also the keyboard being quite expensive. Unless a few hundred thousand fans are ready to buy an X16, it will be difficult to get cheap and non-boring.

You can have this case 3D printed for the OtterX, but it isn't cheap:

OtterX_Wedge_Front.png (703.35 KiB) Viewed 1920 times

OtterX_Wedge_Back.png (980 KiB) Viewed 1920 times

OtterX_Wedge_Inside.png (694.05 KiB) Viewed 1920 times

Or you can have this case cut, but it isn't that different from a "boring" PC:

Otter_Acrylic.jpg (360.46 KiB) Viewed 1920 times

There has been ongoing discussion about the future of the YM2151.

Well, that one is easy: It has none.

For the life of me, I can't fathom why this chip was chosen. It's been EOL for quite a while, and it's not even that good. And it's certainly not beginner-friendly when it comes to programming!

One argument put forth is that VERA audio is good enough.

I'd reject that outright. The PSG doesn't even compare all that well against the MOS SID, lacking as it does both envelopes and filters. OK, so VERA sound with its multiple waveforms is arguably better than the AY-3-8910, but that's a really low bar.

It doesn't feel as though an honest attempt to understand what was being said was made here. It doesn't matter that it is EOL because there is a drop in replacement. The discussion around it is whether to remove it from future generations or use the replacement. The VERA PSG has a lot of voices; it is missing envelopes and filters, but it can play 16 note polyphony without any weird tricks.

The FM-synthesized audio is very characteristic of high-end retro audio and in my opinion adds to the feeling of "quality" of the product.

FM sound lacks all bass and cannot be made to emulate percussion. There are plenty of videos on YouTube with YM2161 music from various systems, and frankly it's all pretty rubbish compared to a four-channel MOD with half-decent samples.

Be precise - what about bass is it lacking? Bass is sometimes defined as 60-250Hz. Is the argument that the YM2151 cannot output in that range? I have spectrographs that would disagree. Is it that there are capabilities it has in the higher frequencies that do not reproduce in the lower frequencies? I would argue that is special pleading.

There were two reasons why most 8-bit users ditched the C64 in favour of the Amiga: The Amiga's graphics capabilities left the C64's lackluster palette and low resolution in the dust, and the PCM sound of the Amiga was so much better than the SID it wasn't even funny. The VERA doesn't look bad compared to the Amiga in the graphics department, but the PSG is really rather lackluster. And honestly, I find the presence of a single PCM channel with a tiny buffer almost insulting, as we're SO CLOSE to having proper sound.

It sounds somewhat like the complaint here is that the our 8-bit machine isn't better than a 16-bit machine. I don't think most of us hold that as a requirement. How would you propose proper sound be done? Audio samples in VRAM? VERA only has 128Kbytes of VRAM (slightly less, I know - details).

If someone were to make a daughterboard that fit in the 2161 socket and enhanced the system's sound capabilities with multi-channel PCM and dedicated sample memory, I'd be ordering tomorrow.

A hardware modification that would make it incompatible with any X16 software using 2151 audio? How many are going to jump on this new hardware bandwagon? Software sells hardware. You need to get an ecosystem up and running to justify that.

What doesn't really fit though, is Commodore BASIC, and PETSCII for that matter. This was chosen purely for nostalgic reasons, and now we have a system that requires a custom keyboard, doesn't support most international character sets out or the box (forget marketing this to most European countries as an educational tool), and has a BASIC dialect that's quirky and quite slow. The X16 team actually paid a licensing fee for this, so they could have chosen a better option.

I think nostalgia is being undersold here. I've been watching people come up and experience the X16 for nearly 2 years now and people with a background on Commodore systems find the familiar feel of it charming - even though it is quirky and slow. Most PS/2 keyboards work fine. We've only recently been getting reports of some that don't. One does not *require* the PETSCII characters on the keyboard.

[novemix]

With all due respect to @arachnivore. This seems like another in a long litany of "you're doing it wrong, it should have been done this way!" posts/arguments. We've been down this road quite a number of times, haven't we?

(By the way, your technical knowledge and passion on the subject, in my opinion, is a welcome addition to the community. I hope you stick around, and I hope you acquire and help extend the machine through hardware, software or both!)

Without quoting him directly, just expressing my own understanding, David set out to build a "spiritual successor" to Commodore's 8 bit line, that, as best as could be done, wasn't plagued by the problems of the existing rapidly aging 40ish year old installed base. Something that fit in between the VIC/64/128 and the Amiga. And I think arguably that goal has been achieved. Does it have its quirks? Yes it does, but show me a Commodore 8 bit machine that doesn't.

Discrete hardware was a major factor in the design. As was a machine which a single person could reasonably understand and build software and/or hardware for. To my mind, an FPGA machine where you can just load a core and be running pretty much any architecture of your choosing isn't that far removed than just having a capable laptop and running an emulator of the same architecture.

And not to put too fine a point on it: this is a hobby. One that primarily is going to appeal to old(er) men. Quite a bit of nostalgia mixed with enough "newness" to keep it sexy. Anyone that wants to play and learn on it is of course welcome, but that's the heart of the matter, in my opinion.

edit: just clarifying the demographic point: it has niche appeal. It's a subset of people that have any interest. Even among retro enthusiasts, "new retro" is not a majority interest, in my observation.