I find frustrating that this type of product (like the MSX VR some years ago), don't specify clearly what are they selling. Is it an ARM SoC? Does it have a Z80? Is it implemented with a FPGA?
Unless the FPGA is very limited, at least MSX2+ compatibility should be available. I would not be surprised if it could do Turbo R as well.
BTW, I wonder if there are easily available notebook shells for sensible prices that could fit a small ARM board and a reasonable screen. An RPi 5 or some of its competitors can power a lot of interesting emulators, not only for games, but for historical research and preservation.
I thought this concept looked familiar and indeed it's by the same company which makes those portable 386 and 8086 machines. LGR covered them a while ago: https://youtu.be/6bODiZ5bP84
The sad thing is, they kindof gave up and officially added support for the things, although on the other hand, trying to sue anyone in China is probably impossible.
My question is: why we have to be stuck in nostalgia?
Why isn’t there today a computer with modern processors and GPU which can boot to a REPL, where you can easily interface with all the HW, for example audio and video?
I wanted to do something like that for the RP400. But is more difficult that it seems…
Because software complexity is truly out of control. Everything needs their own firmware and software and periodic housekeeping tasks to prevent the system crashing. No one knows or has enough time to account for all of those problems, let alone just to build an interpreter environment.
I think Apple and Google both using their phone OS for smartwatches is one indication of lack of such capacity in the industry: Apple Watches come with ~300mAh batteries, old Nokia had 500-700mAh. Watches are rated for 36 hours in normal modes or 72 hours in low-power, whereas Nokia would ran for minimum 1 week, despite ~15 years of improvements in semiconductors.
We'd have to wait for maybe GPT-8y or DeepSeek-R4, probably on wheels, each given a cubicle to get it back down to human scale.
We don’t have to be. It’s just that people like the nostalgia aspect and it gives a de-facto standard with some degree of preexisting hardware to coalesce around.
To have a complete grasp of the software stack, complexity has to be controlled, of course… so the typical offerings are 32 bit architectures with a few MB of memory and an SD card for storage, and if a “GPU” is included it’s usually just another MCU or core acting as the graphics processor.
There are a lot of good cyberdeck options built around the ESP32 and RP2040, for example. While some of these are used primarily as emulators, others run arbitrary code or projects like meshtastic.
I would love to see, though, a simple OS that actually provides a shell and multitasking ability that is designed to run on these platforms, idk if anything like that exists.
It would be super cool to have a python REPL, a text editor, file browser, gopher client, MUD client, things like that running with alt-t to switch context or something like that.
There has to be a small Unix-like OS that could fit an RP2040 or ESP32. No MMU might be a deal-breaker for a more complete (and secure) OS, but we don’t need all that functionality in every form factor.
I’ve been hoping for this as well, but so far I haven’t seen anything that isn’t so limited as to be unusable for anything other than saying “look, a shell!” Can’t run programs from the shell etc.
Is not that aspect that bothers me. I’m on the same boat. I mean: why I feel I was capable of doing things in a C64, that I cannot dream of doing in my new 2000 dollar computer.
There’s nothing you could do on a C64 that you couldn’t do on a modern computer. It’s just that doing that on a vastly more complex and capable system is intrinsically more complicated.
An interesting experience is trying to run a program on an emulated 60’s mainframe OS. 1960’s and 1970’s mainframes were about on the same league of 1980’s and 1990’s home computers, but a lot more complicated and with much more control over the low level details. It gives an interesting perspective on how software complexity tracks hardware capabilities.
One aspect is likely that you get a "free" software ecosystem. They can presumably grab a bag full of MSX demos and freeware and drop it on the disc, their job is done. They don't even have to document the REPL (BASIC) since books already exist.
A from-scratch platform has to at least be given some level of documentation to make it accessible, and ideally some software you can show for marketing purposes. It's a completely different skill set.
It's entirely possible these guys are not software people (they did the Pocket 8088 earlier, and you might remember the backlash when it was shown they just grabbed an open-source BIOS and filed the serial numbers off) so even if they made a super-cool machine, it might wither due to missing that first solid software baseline.
Modern hardware is too complex to understand and bootstrap for a single person even more so as a hobby project. These known unknowns can be a powerful motivation for someone who (partially) understands modern hardware to tinker with understandable systems. Some go down the retro route while others use medium to largish microcontrollers.
I started my career as a programmer in 1972, writing real-time systems in assembly language on DEC PDP-8s and PDP-11s. Back then all software was written in assembly and distributed as machine readable source. (This being DEC, not IBM, that meant paper tape, not punched cards nor magnetic tape :-)
Pre-microcode and mostly pre-CISC instruction sets were readily mastered.
The constraints of storage, distribution media, lack of any high-level system programming languages, the small number of programmers, etc... meant that there just never was not that much system code for any given architecture. Similarly, I/O devices were exteremely simple and limited in number. MMU hardware was still a rarity. All this meant that any reasonable system programmer mastered a machine entirely, its instruction set, all its I/O and all of its (system) software.
At the end of 1977 DEC shipped its VAX-11/780 and VMS V1. Less than a year later, I join DEC's compiler group to write VAX Pascal V2. I can only compare moving to VAX/VMS to being kicked out of the garden of Eden. Quickly I realized that never again would I enjoy that sense of mastery that I had felt on earlier machines :-(
I miss the entire Maker Movement moment. It isn’t actually super complex to make a simple circuit board that does some basic things for you.
It’s cheaper to just outsource making the board, but I even made my own simple boards (etched using hobbyist CNC machines). I even used a bunch of surface mount parts with them.
I can program this thing fully
without access to internet. You can understand 100% of the hardware down to transistor level. Modern stuff doesn't have that.
Yes. Absolutely. But let me dream, there is an operating system, very good integrated with a decent IDE, where youncan create applications very easily, without downloading terabytes of cuestionable libraries. I mean like in the leftpad case.
Unless the FPGA is smaller than the 1ChipMSX, it could load the uMSX and go up to MSX2+. Certainly enough for a SymbOS environment.
I'm curious, however, how it manages disks. MSX-DOS didn't have support for subdirectories, so I'm assuming a lot of different hard disk and floppy images would end up being visible.
My MP/M emulated Altair Docker image has 8 1MB floppies and 2 5MB hard disks. And, of course, no subdirectories (unless you count USER areas).
You need to provide the firmware, because they aren't free to distribute. There's an open source one, but it doesn't boot to BASIC and can only be used to run cartridges.
And it is because is not an MSX, it emulates one with an FPGA.
You can replace pouch cells like-for-like, IIUC, the only real parameter is charging current... just don't tell your fire insurance that I told you so(or that you buy 18650 online, either)
> is this actually useful without a cartridge slot?
It has 2 of them actually. One standard one on the back and an internal one that I think doesn't use the standard connector that is intended for hardware expansion. Despite what the product page says, the internal slot is mapped to slot 2 and not slot 1 in the MSX mapping system, so it doesn't conflict with the external cartridge slot.
That said, most computers of this era including the MSX had a fair bit of software released on cassette tape as well. There were also disk drives and software released on disk as well.
Looking at some of the more famous games for the system: Metal Gear was released on cartridge, but Snatcher was released on floppy disk with an expansion cartridge that contained extra RAM and a sound chip. It seems this system has support for that expansion builtin though.
It has slots. But anyway you do bot need slots at all. MSX was mostly tapes at that time. I’m sure there is either a simulator, or you can generate the sound from binary files.
Yes, it says "OneChipMSX" https://www.msx.org/wiki/1chipMS but IMHO it could be clearer.
If you go to http://www.8086cpu.com/MSX/103.html there's more information, but it definitely puts me off when considering one of these devices.