A tiny RISC-V MCU board selling for $3 | Hacker News Reader

A tiny RISC-V MCU board selling for $3

cnx-software.com

129 pointschildintime41 comments

As already noted in the comments under the linked article: This device is currently seriously lacking software support, bordering on making it unusable.

It's interesting to note that the company that developed the chip (Gigadevice) initially got big by copying microcontrollers by ST microelectronics. This went as far as copying the entire product portfolio including naming scheme and marketing strategy. (Personally I find this absolutely appalling)

I guess not having to develop a software toolchain for their devices provided a huge benefit for them. Now that they have a device that cannot rely on external tools, things look a bit more difficult...

I'd been wondering if the GD32F107 was a clone of the STM32F107! Thank you for confirming that! But surely you can't honestly think that ST microcontrollers are a reasonable alternative for most electronics manufacturers? As far as I can tell, ST doesn't even offer datasheets in Chinese!

I think RISC-V software support will probably take shape pretty quickly, in the next $smallinteger years.

In terms of the ethics of cloning products from other companies, such as Linux, Chrome, Intel’s 64-bit chips, and Google Search: do you want the future to be dominated by engineers who create new things, or by owners of government-granted monopolies like copyrights, patents, and RF spectrum? Bram Cohen and Richard Stallman, or the RIAA and Verizon?

ST's microcontrollers were really popular in Chinese-designed electronics before GigaDevices cloned them, and the STM32F103 - which is the one they initially cloned - was especially so. Its popularity in China is almost certainly why they decided to clone it in the first place.

Sure, when you're desperate for decent parts, it makes sense to use overseas suppliers with long lead times and customer service and datasheets all in English. And the STM32 line is really great. But if you can instead use a domestic supplier who speaks Chinese and can get you parts in a few hours instead of a month, why wouldn't you?

The GD marketing materials say they support the standard RISCV tool chain.

https://www.gigadevice.com/press-release/gigadevice-unveils-...

who knows, sometimes plagiarists start for the low effort and then start to make good work

For a microcontroller, the chip is better supported than I had expected. All peripherals are well documented with registers clearly defined, the headers all seem to be correct, and the documentation has been translated well enough.

There are shortcomings, such as the USB example code in the SDK not building, but I'm grateful for the chance to hack on a RISC-V SoC for less than the price of a beer.

Isn't that how the world works?

Hollywood copied Edison's patents which you could consider just as bad but it resulted in a lot of new tech which we would not have.

It is interesting that they got their start that way - I've been wondering if the peripherals for SPI, USART, etc. work the same way in their RISC-V boards as they do on STM32/GD32 chips.

If so, one side effect of their mimicry might be that their RISC-V MCUs are much easier to get started with than their competitors'. I don't know how similar the two lines are, but it sure would be nice if you could just swap arm-none-eabi-gcc for the appropriate 32-bit RISC-V toolchain. I've been meaning to get some of the $5 Longan Nano boards carried by Seeed Studios, but the holidays are always distracting.

' GigaDevice similarly claims “complete compatibility” between its new GD32V RISC-V microcontrollers and the classic GD32 series of MCUs based on Arm. '

https://www.eetimes.com/gigadevice-intros-general-purpose-ri...

Sweet!! Can't wait to try them as drop-in-place Cortex-M replacements.

Gigadevice is indeed a very funny company.

Yiming Zhu is said to be a youth league cadre. A very opportunistic man I'd say.

He is very different from Chinese dotcom bosses. He has a ton of side businesses which he runs himself, and investments in all kinds of things, but unlike his peers he is not using his company's money, but his own cash.

Given that he is still in the business 20 years down the road while doing things his way, he is really something.

There's LOTS of hardware that is interesting but useless due to poor software support.

Conversely there's also hardware that is less than impressive but the best option because it has good software.

These boards are very cool, and I would happily play with them if I had time. However one thing is a bit mystery to me, what are the use cases more professionally? Obviously there must be lots of non-hobbyist demand for boards like this.

The boards are a great reference point for developing a product with a given component. Any time I make something with an MCU I pick up a couple of dev boards just like this to learn from. It's basically a ``known good'' configuration and really handy for prototyping and debugging hardware issues.

Typically, boards like this would only be used for prototyping. It's a dead simple system and there is basically nothing of value in the board itself for production use. Any design using the same microcontroller would just incorporate it directly rather than integrating an extra board. That differs from more complex boards like an RPi Compute Module or an RF module. On those, you might take advantage of the existing high-layer-count board attached to a simpler, low-layer-count board, avoid needing to redesign systems that have signal integrity or RF challenges, or utilize existing certifications around emissions.

> or utilize existing certifications around emissions.

This is one of the biggest reasons for an extra board for RF modules. RF is black magic where you REALLY need to know what you're doing to develop your own board, and if it's not in an already certified module then you have to have your product pass FCC certifications (costing a lot of money)

Very often even the layer count is an issue. For perfect example see almost any brand name japanese AV DVD/DVR combo which has giant single layer board with only passives, connectors and links which holds small subboard with bunchs of (fc)BGAs and somewhat incredible layer count.

On the other hand for small scale devices not only certification comes into play but also availability and cost. There are niche devices that simply use some Arduino-ish board as a component, because the board is easier to source and, more importantly, cheaper than the MCU on it, personaly I view that as bad design and good reason to use some non-AVR MCU, but that is how it is.

There are a couple of issues. PCB and and Assembly for an RF circuit while it's not really dark arts requires higher quality pcb fab and assembly. Using an RF module thus lowers the manufacturing requirements for the main board. If the RF is on the main board switching board houses often causes problems.

You can also treat an RF module as a component and test it separately. Parametric testing RF as part of a full design is problematic. And RF systems appear to fail QA at a high rate than digital/analog circuits.

And also can share the same module across different products.

It's early in its life cycle and needs a JTAG interface to work with now, but it _could_ be a sweet little Arduino board. Comparable to the TeensyLC but $3 instead of $12. 3.3v 33mA 108Mhz 32k RAM 128k flash.

It's probably more closely comparable to the STM32F103-based boards which cost about the same amount as it but have better software support. This is sort of a derivative of a clone of that exact microcontrller.

There's some pretty cheap blue boards that cost less than the CPU on them, due to a manufacturing flaw. (one resistor is wrong, so it won't program on most PCs)

I've been interested in homebrew computer projects that use a Z80 as the CPU, I wonder how easily it could be done with a RISC-V cpu.

It would be interesting when ChibiOS supports RISC-V.

Black on green silk screen text is an unlikely choice. Still early days however so probably fixed in next revision.

We just had an article on HN yesterday saying there were no cheap RISC-V board available... and here we are!

He was looking for microprocessors that support the privileged RISC-V specification, not microcontrollers.

The issue was that there wasn't a cheap riscv board that supported the privileged section of the ISA (so you can run an OS on it). This, being a microcontroller, doesn't either.

The Reference Manual claims "Machine (M) and User (U) Privilege levels support" on page 22.

Probably not the kind of privileges discussed yesterday.

yup, for linux you want M/S/U modes at least (kernel runs in S) and page table table support of course

I was under the impression you could get away with using Machine mode for both the SBI and the Kernel?

Sv39 paging is the major requirement as you point out.

Machine mode runs without the MMU, linux really needs an MMU to do kernel stuff too (including stuff like copyin/copyout)

Ah, I didn't actually know this, thanks for pointing it out :)

There is actually quite a bit of work going on in Linux without MMU. Check Linux Plummers conference RISC-V track for example.

Yeah that's why I said "really needs", there are cut down versions that will run without an MMU, it's not really the mainline Linux though - I've worked with MMUless kernels in the past, it's not a lot of fun (and I started porting V6/V7 for base and bounds swapping machines)

Such boards can be used for emulation retro games?

No. It's a microcontroller board.

At least the STM32 and LPCxxxx boards have plenty of horsepower for 80s game emulation.

ESP32 too: http://www.fabglib.org/

You're not going to be doing much emulation with 32 KB of RAM.

The Atari had 128 bytes of RAM. Nintendo had 4.3K.

Reply on news.ycombinator.com