MIST user base

[jotego]

Does anyone have any realistic estimation of how many MIST boards have been sold?

For instance, the ZX-UNO is reported to have about 300 users at the moment.

[RobertK]

One who should know this is Lotharek. Just send him a mail and ask him (and briefly tell him who you are and what you are doing, he will be happy to hear about your work).
But ask him if you may publish this information - it is up to him whether this information should be made public or not.

[sebdel]

Unless I misunderstood something I would say that this bit of informations is just as much the property of Till than Lotharek. That being said I would also be interested to know a ball park of just for how many people I'm working my ass of. Thousands? hundreds ? a dozen? Vebxenon?

[Newsdee]

Cores are open source and portable to other boards, so your work is in no way restricted to the MIST...

[guidol]

For instance, the ZX-UNO is reported to have about 300 users at the moment.

Are the 300 users only out of the Crowdfunding?
Because Antonio Villena did have a own batch of ZX-UNOs to sell (around 45) and there also some Prototype-UNOs which have been sold.

Additionally there will be a number of "ZX-UNO Model M" in the near future:
http://www.8bits4ever.com/zx-uno-m.html
a special version where VGA, real Power-Jack and Audio/Video RCA Jack is onboard and a other case is included.

I did buy one from Antonio Villena - and its on the way from spain to turkey (now in Istanbul / PTT Center)

ZX-Uno Facebook-Group: https://www.facebook.com/groups/1571560916478710

[farvardin]

I've already asked Lotharek about this last year, but no answer so far.

[Newsdee]

As of September 2015 last year:

I sold over 80 MIST yet and 50 MCC-216 - without doing any real advertisement or video that shows what you can do with it, yet.

Source: http://atariage.com/forums/topic/242970 ... try3328656

That isn't counting other resellers (Amedia and AmigaStore) nor anything that Lotharek has sold directly, so I'd estimate there are at least 300 to 500 MiST units out there.

Yet that doesn't mean much since any successor device will be able to have the cores ported to it (same as how the MiST is a functional evolution of the Minimig and OneChipMSX, to name a couple). Anybody writing an open source core for the MiST will be reaching many more users than the number of MiST sold.

[jotego]

Yet that doesn't mean much since any successor device will be able to have the cores ported to it (same as how the MiST is a functional evolution of the Minimig and OneChipMSX, to name a couple). Anybody writing an open source core for the MiST will be reaching many more users than the number of MiST sold.

I would actually love to have an improved MiST board with:

• RS232 serial port
  Real 24-bit audio DAC (ideally from Analog Devices)
  Real 8-bit video DAC
  Tape recorder input in order to load Amstrad/Spectrum tapes
  Expansion port (would need a larger FPGA)

[Brt]

I have bought mine ~2 weeks ago and my the serial no. is #374.
Maybe this isn't exact information of units produced, but this might give a clue (I bought it from reseller, and I have no information of when did he make his stocks).

[ericgus]

Yet that doesn't mean much since any successor device will be able to have the cores ported to it (same as how the MiST is a functional evolution of the Minimig and OneChipMSX, to name a couple). Anybody writing an open source core for the MiST will be reaching many more users than the number of MiST sold.

I would actually love to have an improved MiST board with:

• RS232 serial port
  Real 24-bit audio DAC (ideally from Analog Devices)
  Real 8-bit video DAC
  Tape recorder input in order to load Amstrad/Spectrum tapes
  Expansion port (would need a larger FPGA)

Even just a "generic" GPIO "header" would be appreciated .. that way you could make daughter boards or connectors that interface to whatever legacy hardware ports you need (i.e. C64 IEC, Atari SIO, parallel/serial, tape,etc..)

[ryan]

I love my MiST but a successor machine with HDMI (or at least DVI) output would be nice -- it's hard to find a TV that takes VGA these days. I understand there's licensing issues with HDMI, but it's getting to be necessary.

Also the FPGA in there is quite old now. With more gates it should be possible to emulate Falcon. Or even license and run the Apollo core and have a heavy duty machine. Also if you want to do development for MiST you have to download quite old versions of Quartus, newer versions don't support the Cyclone that's in it.

[farvardin]

I have bought mine ~2 weeks ago and my the serial no. is #374.

where did you find this serial number? Probably it's inside... I'll try to open mine one day.

[Newsdee]

As far as I'm concerned, I'd rather keep the current hardware as long as it is practical to do so, to avoid a split in the community of users and devs.

As long as the old Quartus is around for free we can still use it. Also I believe none of the cores has reached 100% utilization of the FPGA. So there is room to push the machine and cores further.

There are already more powerful boards out there, but they are more expensive. While new feaures would be nice, I'd like to see a cheaper MiST that is backwards compatible, which is more likely to attract new users and potential developers.The ZxUno is interesting in this regard, but it's underpowered compared to the MiST.

[Brt]

where did you find this serial number? Probably it's inside... I'll try to open mine one day.

I didn't have to open anything, I have additional sticker with # on bottom of the unit (next to Mists nameplate).

[vebxenon]

As far as I'm concerned, I'd rather keep the current hardware as long as it is practical to do so, to avoid a split in the community of users and devs.

As long as the old Quartus is around for free we can still use it. Also I believe none of the cores has reached 100% utilization of the FPGA. So there is room to push the machine and cores further.

There are already more powerful boards out there, but they are more expensive. While new feaures would be nice, I'd like to see a cheaper MiST that is backwards compatible, which is more likely to attract new users and potential developers.The ZxUno is interesting in this regard, but it's underpowered compared to the MiST.

I Agree

[antoniovillena]

I am Antonio Villena. I have sold about 90 ZX-Unos without count the prototypes in the last 3 years. I know another person who sold a batch of 40, and another one who sold 30. So between all 3 and the crowdfunding ones there are at least 410 sold. It's impossible to do a good estimation. I would say there are about 1000, but can be more. The license is CC by-SA, that allows third parties make their own batches without paying royalties or ask us permission.

[jotego]

As long as the old Quartus is around for free we can still use it. Also I believe none of the cores has reached 100% utilization of the FPGA. So there is room to push the machine and cores further.

A few weeks ago I added YM2151 support to Atari ST in MiST (video). I got the FPGA used up to 99% just with that. When the FPGA got so used up, the lack of a good constraint setup for the ST core showed up and I got many implementations that didn't work. When I tried to add better constraints I found it too hard to get the design meet the timing constraints. A larger device would help with these things. Also, it would then be possible to have more peripherals integrated into the cores, like an FPU for the 68000 cores. I am still not sure an arcade system with M68000+Z80+YM2151+Oki PCM would actually fit in the current FPGA.

So I understand that with a small user base, we are already talking about splitting it up but I think the original design is limited by the aspects mentioned above.

The ZX-UNO has the nice expansion port but they have made some odd decisions, like not having HDMI nor VGA connectors in the original board. Their FPGA is also even smaller and the RAM is also very small. They put the emphasis in having a nice analog video output for TVs in order to replicate the original ZX Spectrum experience so they didn't care about these things.

[Newsdee]

Sure, each board/project has done different compromises... the MiST has done some but somehow hit a sweet spot that got many people onboard. So I suppose before embarking in another MiST design, one should consider existing alternatives carefully.

There is the FPGA Arcade which seems higher spec than the MiST and allows for DVI (so HDMI) and a daughterboard to receive a 68060, which should in principle free up capacity inside the FPGA. Caveat is that it's still not shipping widely (after all these years)... but reading their forums it looks like it might start to happen soon.

The other option I can think of is the newer Terasic Cyclone V boards which have gone down in price. There's a couple of boards on the website that seem very high spec compared to the MiST and FPGA Arcade, but at a price that's not much more than both ($179 for the Cyclone V GX Starter Kit and $350 for the SoC Kit). The FPGA is huge (70K to 110K LE) and it has a lot of stuff people would like (audio input, GPIO, etc). What am I missing? Is there a catch on these boards?

[ijor]

The other option I can think of is the newer Terasic Cyclone V boards which have gone down in price. There's a couple of boards on the website that seem very high spec compared to the MiST and FPGA Arcade, but at a price that's not much more than both ($179 for the Cyclone V GX Starter Kit and $350 for the SoC Kit). The FPGA is huge (70K to 110K LE) and it has a lot of stuff people would like (audio input, GPIO, etc). What am I missing? Is there a catch on these boards?

Agree. I like the DE1-SOC, U$100 less than the SocKit, and more suitable for our porpuses, I think. For about the same price as MIST you get something much more powerful (in terms of RAM and FPGA capacity). It includes its own ARM processor integrated in the FPGA. It is simply impossible to compete with those prices.

Off the shelf boards have some drawbacks. They don't have, of course, legacy joystick connectors. But if somebody insist on using them, it is not big deal to make a cheap, simple daughter board. Some models can't be configured from the SD. Only from flash or from JTAG (USB Blaster), so they are not suited for regular users. Another problem might be the luck of a box. Probably Lotharek won't be interested in selling a box alone. Personally I don't mind, and Terasic boards have some kind of protection, they are not just bare boards. But guess many users would like it boxed.

[ijor]

A few weeks ago I added YM2151 support to Atari ST in MiST (video). I got the FPGA used up to 99% just with that.

Really? How much LEs take your YM2151 core?

Also, not sure I understand the YM2151 integration with the ST. It is logically connected as a MIDI device?

[jotego]

Really? How much LEs take your YM2151 core?

I have attached an implementation summary of the JT51 inside the MiST core. It is comparable to the size of the M68000, basically because of all the memory involved in holding 32 sound operators (which are then mixed to produce 8 channels in stereo). 96% of the logic elements in the FPGA are used.

Also, not sure I understand the YM2151 integration with the ST. It is logically connected as a MIDI device?

Well, the idea was to connect it as a MIDI device so it could be used directly with SIERRA videogames. There are two problems, though: 1. No more space in the FPGA to add the MIDI interface. And 2. Timing violations once the FPGA runs out of real estate so the MiST becomes unstable.

[Newsdee]

Some models can't be configured from the SD. Only from flash or from JTAG (USB Blaster), so they are not suited for regular users

This is probably the biggest problem. Does any of the cheaper boards allow loading core from SD like the MiST? Or is it a premium feature of more expensive ones?

A daughterboard and box are much easier to produce than an entire nee board. Of course making a MiST2 that is mostly backwards compatible would be nice too (trivial port).

[ijor]

This is probably the biggest problem. Does any of the cheaper boards allow loading core from SD like the MiST? Or is it a premium feature of more expensive ones?

It is not exactly a premium feature. It is a limitation of the FPGA non-SOC architecture.

The MIST can configure from SD because it has a separate MCU. The Atmel micro reads the configuration from SD and reconfigures the FPGA. Most Terasic development boards don't have an external micro, so this kind of configuration is not possible.

SOC devices, OTOH, have an internal Arm processor that can configure the FPGA core.

[ryan]

The Zynq series from Xilinx is an SoC combining ARM core and FPGA. Could be used in a similar way.

[antoniovillena]

Some models can't be configured from the SD. Only from flash or from JTAG (USB Blaster), so they are not suited for regular users. Another problem might be the luck of a box

ZX-Uno users has not JTAG programmers (Xilinx cable, USB Blaster or RPi) but can configure from SD in 2 steps. The xilinx cores have multiboot, so you can load up to 9 cores from flash. The first step is upgrade from SD and replace 1 of 9 cores. In a second step just select the upgraded core from boot menu. Both steps last less than 30 seconds to be done.