Single board computers are making a huge splash in the world. Not too long ago they were a darling of the tech sector. But these days they’re often the unseen powerhouse between any number of technological achievements. They’re often the underlying technology of any number of media centers or gaming devices. But do they have a deeper secret?

Yes, in fact there’s a whole new realm of research going on with single board computers. Single board computers, or SBC, are amazing devices. But they’re typically fairly limited in terms of raw power. The upper edge of SBCs is currently around 1.7 Ghz per processor core. To be sure, this is rather amazing in terms of technological progression. A cutting edge computer in the early 2000s was typically single core and only around 1 Ghz. And these devices were also huge tower based machines. A single board computer today takes all that power and puts it into something smaller than one’s hand.

But it’s still lacking in overall power for many tasks. And in particular one will run into problems trying to run any kind of AI on a single board computer. The fact that a quad core SBC can run multiple processes between the cores is helpful. But it’s still going to be hampered by that upper speed limit. However, this is where people have found some interesting workarounds.

The easiest method to get around speed limits is by simply offsetting most processing to an external device. This is how most so-called smart assistants work. The majority of smart devices are actually quite a bit slower than the standard single board computer. The only thing these devices really do is shuffle data between a user and a larger device. The big issue here is security and reliability. If the internet connection lags then so will the device. And an external device will also have access to 100% of a user’s data.

This might suggest that AI isn’t possible with a SBC. But there is a way around this by using edge computing. Cloud computing offsets almost all computing needs to a computer accessed over the internet. Basically, a computer that one has no control over. Edge computing uses a processor on the edge of a device. This is usually something physically connected to the system.

This is sometimes done with a dedicated co-processor. A co-processor is a specialized system that’s specially targeted to a particular task. This is especially popular with AI techniques involving simulated computational neural networks. A neural network is quite feasible on any device. But it typically needs high speed processors on devices with a limited number of cores. Even a quad core SBC with heavy processing power will often have trouble with neural networks. This is due to the branching nature of these mathematical simulations. But a specialized coprocessor when used with an SBC is better able to handle the task.

Still, there aren’t many single board computers built to this spec. What’s far more common is for a system like this to attach to an SBC via USB or special sleeve system. This is even more viable now that many popular SBC brands have moved up to USB 3. The extra speed from this standard makes it even more practical to set up additional systems in an edge network.

In theory one can even chain these configurations together to form something known as a cluster. And a single board computer is especially well suited to this thanks to their lower price point. Combining two single board systems with a dedicated co-processor can create overall exponential growth in overall processing speed.