Hi, thanks for your report !
I’m zepan, the shopkeeper of Sipeed Tech. Taobao shop, Sipeed M1 is our first AI moulde ,and the second RISC-V module.
I need correct some mistake in your article, and supplement some information.
1. The chip name is K210, not K201; and it is RV64GC.
2. It have 8MB high speed SRAM, not 6MB.
3. There are 5.9MB SRAM can be used for convolutional neural network acceleration, so, it is possible to run small model like tiny-yolo v2,MobileNet, as you see in face detection routine video.
4. It isn’t preoder mode before 10.7, but it is too popular that it is sold out in 10.7, and I change it to preorder mode.
More information:
1. It can be up to 800MHz if you want, and give you more than 0.5TOPS, 240FPS detection speed @ QVGA.
2. It is low power consume, unlike other ARM board; It only consume 0.35W when running face detection routine.
3. Its IO speed in simple test is up to 300MHz
4. It is possible to run openwrt on it as it have MMU, but too tight in ram.
5. We’re adding micropython to the board, and soon you can simply run python on it, you know python is the best lang for DL.
6. We also have Mic Array board, and another fancy board in development.
7. We have best price for those solution, Sipeed M1 module in $6, M1w(with wifi) in $7.5, the simple dev board with LCD, 2M pixel Camera, I2S MIC, Speaker PA, on board downloader, in $15.
8. We have model shop soon, you can sell your model on our online modelshop.
Possibly a very nice chip. A couple of questions:
- any ADC on board?
- I presume it uses QSPI flash memory?
- will the items be available on aliexpress?
I don't think it has a GP ADC on board, and the GPIO section only seems to talk about interrupts bus width configuration, so probably no ADC on those pins either; but the datasheet does mention an Audio Processing Unit "APU".
Here is a bit detail of the neural network chip. [1]
Anything aside, as a former embedded dev, I would never use there kinds of voodoo chinese chips. Don’t get me wrong, it’s not because it’s from China. It is because of __zero__ well-written public documentation.
Just listing specs in datasheets is meaningless. The datasheet is translated but it does nothing to devs because it has absolutely zero about low level peripherals and its core. You cannot imagine how painful writing hardware-level code is when the documentation doesn’t help. I’m 100% sure even the chinese version documentation has almost zero information too.
Look, they have a sort of SDKs and the only example is one-line-long Hello World. [2]
It seems like it quantizes FP16 into INT8. That's a waste, because many models can work perfectly fine with binary/ternary weights, and some can even tolerate low precision activations.
”Documentation is available, although the datasheet will need to be translated”
That’s one way to see it. An alternative is ”If you’re a young hacker, you better start learning Chinese, just as young hackers fifty-ish years ago didn’t wait for translations, but learnt English”
Lol, I'm literally already doing that (duolingo + Bunnie's Shenzhen book is a pretty good combo).
FWIW, there's not much in the datasheet. Pin assignments and broad features, but no ISA for the KPU (the CNN accelerator), no registers or addresses except some of the SRAM as far as I can see. More 'datasheet' than 'reference manual' since those terms get thrown around meaning different things.
If you are a hacker you probably would likely take a tech powered short-cut. Google has a translator toolkit that lets you upload documents and also create a glossary for technical terms and consistent terminology. Haven't tried it yet but it would be great to start a crowdsourced glossary for electrical engineering documents if one doesn't already exist.
Yes, I decided that Mandarin is a skill I need a few years ago - China is 1/5 of the world's population, and 1/5 of the world's smart people, and with so much energy - not engaging from the west means we're missing out
Since RISC-V is being mentioned here, thought maybe I took the opportunity to ask about it. I would like to learn more about instruction set, I know very little about it so I was planning on purchasing a book called Computer Organization and Design, it comes in three flavors: MIPS, RISC-V, and ARM. Which one would be the best option to read of those given, or which one would be the most relevant where I would get the most value out of it. Can anyone provide a minimal suggestion? Thank you in advance!
They're all RISC architectures (ARM maybe isn't any longer given all the extensions), so the treatment will be largely the same. RISC-V is the newest of those instruction set architectures and is the most cleanly designed, so the book probably won't need to explain legacy cruft that MIPS and ARM include but that aren't essential to understanding how computers work.
Unless you have a reason to work with MIPS or ARM I would go with the RISC-V edition.
Once it becomes available again I would really like to order one (or more), but the web page linked from the article for ordering is in Chinese, and I didn't see any way to get to an English language ordering page.
start with google translate .... TaoBao is kind of the best place to buy cool Chinese stuff, and often with the best prices, but it's not designed for sales to the West - it's where Chinese people sell stuff to each other
Hi, thanks for your report ! I’m zepan, the shopkeeper of Sipeed Tech. Taobao shop, Sipeed M1 is our first AI moulde ,and the second RISC-V module.
I need correct some mistake in your article, and supplement some information.
1. The chip name is K210, not K201; and it is RV64GC.
2. It have 8MB high speed SRAM, not 6MB.
3. There are 5.9MB SRAM can be used for convolutional neural network acceleration, so, it is possible to run small model like tiny-yolo v2,MobileNet, as you see in face detection routine video.
4. It isn’t preoder mode before 10.7, but it is too popular that it is sold out in 10.7, and I change it to preorder mode.
More information:
1. It can be up to 800MHz if you want, and give you more than 0.5TOPS, 240FPS detection speed @ QVGA.
2. It is low power consume, unlike other ARM board; It only consume 0.35W when running face detection routine.
3. Its IO speed in simple test is up to 300MHz
4. It is possible to run openwrt on it as it have MMU, but too tight in ram.
5. We’re adding micropython to the board, and soon you can simply run python on it, you know python is the best lang for DL.
6. We also have Mic Array board, and another fancy board in development.
7. We have best price for those solution, Sipeed M1 module in $6, M1w(with wifi) in $7.5, the simple dev board with LCD, 2M pixel Camera, I2S MIC, Speaker PA, on board downloader, in $15.
8. We have model shop soon, you can sell your model on our online modelshop.
9. Our telgram link: https://t.me/joinchat/IoJz2UoLTnlscC0WrCCNrg
Our qq group: 878189804 for AI, 826307240 (3k members group) for common discussion.