
Rock Pi X
The most bang for your buck with embedded computing! This is probably the most impressive single board computers (SBC's) that we've experienced for the price. This is similar to the Up Board and LattePanda, but more affordable.
Because this board is running Microsoft Windows, it can also run ARC directly. There is an audio jack for speakers and an HDMI for video.
ROCK Pi X is the first X86 SBC(Single Board Computer) by Radxa. It can run Windows and ARC. ROCK Pi X features...
- Intel Cherry Trail quad core processor Z8350
- 64bit dual channel 1866Mb/s LPDDR3
- up to 4K@30 HDMI Video
- 3.5mm audio jack with mic
- 802.11 ac Wi-Fi
- Bluetooth 4.2
- USB Port
- GbE LAN
- 40-pin color expansion header
- Realtime clock
- USB PD and QC powering
ROCK Pi X comes in two models, Model A and Model B, each model has 1GB, 2GB or 4GB ram options. for detailed difference of Model A and Model B, please check Specifications.
This comparison chart, done by our friends at Explaining Computers says it all...

Installation Tips & Performance
We have a guide in the Support section that includes steps on freeing storage space and increasing performance of robot computers, such as single board computers. View the instructions HERE.
Related Content

Live Hack Ideas/Suggestions

Grippy Live Hack Part #1

Have fun!!!!!
I should set that up as a live hack one day . Use the inmoov as a telepresence robot and control it remotely from dinner or something haha
You can view it here: https://synthiam.com/Support/Install/free-space-with-windows-10
One area we are lacking is GPU support for ML at edge and the dependance on X86 Architecture is a challenge. Migrating to .NET core with multi architecture support for Lin/Win/Mac opens up a lot more cross platform capabilities. With ARC supporting ARM and CUDA on NVIDIA (Nano, Xavier etc) would be ideal for robotics platforms. Obviously that requires a fair amount of work on Synthiam part. Another option is to look at partnerships with companies like AMD. ROCm is maturing, they are an X86 platform and are launching edge devices enabled by chips like the V2000 so this could be an option to explore in future.
I also don’t really believe in the ml running local. It doesn’t make sense to consume that much power in the robot locally - specifically since the global internet communication infrastructure is barely at capacity. It was significantly over engineered expecting less data compression. But with media compression the way it is, bandwidth and low latency is readily available for cloud computing.
offloading ml to the cloud is the right thing to do. I don’t even want to consider what batteries would be needed to power a gpu for useful ml.
Like, come on... tensor flow is a joke to run locally even on the most powerful hardware. There’s a lot of technologies we kinda skip at Synthiam / mostly because we wait for it to mature and stabilize.
just look at google dialog flow. Ugh that’s the worst. It’s changed so much that it’s impossible to integrate. I should have waited longer before making a skill. Now it’s just a mess.
The open source and education/exploration space is really fragmented and unreliable. By the time you figure out how to implement something, the technology had changed.
so back to core... until it matures AND has a GUI, we can’t use it. However I should add that our entire cloud infrastructure and website is core
Moores law kicked in and now you can put an SBC like Rock Pi X directly on the robot so I don't need an external computer anymore. 7nm has brought in 10w SBC's with CPU/GPU/TPU cores onboard. This opens up a whole new world for Autonomous robots. Network goes down, robots keep running. Sure you can still offload complex ML to the cloud you don't need to process local when network is up, but there is enough smarts on board to allow the robot to continue to function off line. You really don't want a robot that a 12 year old kid can shut down with a $10 jammer he bought on Ebay.
Hopefully as MAUI matures ARC moves over, it sounds like you have given it a lot of thought and it's on your roadmap. I understand that avalonia probably isn't a strategic move when an official UI is in development.
so the decision was to be aware that one of two things is going to happen soon
1) core will get a GUI that’s cross platform. We can easily fork and adapt to it. Because it’s something we consider to be core compatible as we continue developing
2) windows moves to the cloud and runs in a Remote Desktop type portal and is available from any os
one of those two things will happen - depends on what’s first
in the meantime, the love from paying subscribers has gotten us pretty hyped and validated our efforts. So there’s a bunch of work going on for new skills. Such as yolo object detection, intel realsense navigation, and ros node integration. Oh and a unity add on. Oh oh and a two way telepresence option for exosphere that turns absolutely every robot into a telepresence.
some pretty cool stuff! Lots of ML items planned to. We’ve been collecting a ton of data from exosphere, and now starting to experiment with the robots self navigating.
I’m really enjoying the ui and performance improvements of recent ARC releases - hope you are to. It’s made a drastic improvement on sbc’s. I’ve been using the up board a lot lately and ARC is so much more response than it has ever been!
I don’t register windows on my sbc’s
The Rock Pi X heat sink is back in stock at Seeed Studio (China Warehouse)! The stock is pretty limited though.
They still have plenty of Rock Pi X boards too:
https://www.seeedstudio.com/catalogsearch/result/?q=Rock+pi+x
Oh and if I hadn't said it yet, Happy New Year to you guys and gals!