Exploring Google Coral PCIe via USB 3.0: Feasibility and Tips

Hi everyone, I’ve been diving into the world of smart home enhancements and came across an interesting challenge that I thought I’d share and discuss with the community.

I recently acquired a headless Surface Pro 4, which, despite its swollen battery, has proven to be a treasure trove of potential. My goal is to repurpose it as a more powerful machine learning hub compared to my current Broadwell laptop. The Broadwell simply can’t handle H265 decoding, which is a critical feature for my setup. Enter the Google Coral PCIe card—I’ve been using it with Frigate, and it’s been a game-changer. However, here’s where things get tricky: the Coral is in PCIe form factor, and my Surface Pro 4 only has a PCIe M.2 slot. This means I need to connect the Coral via an NVMe to USB adapter.

But here’s the catch: the Surface Pro 4 is limited to USB 3.0. I’m curious if this setup will provide enough bandwidth to reliably support the Coral. From what I understand, USB 3.0 offers about 5 Gbps of bandwidth, which is less than the theoretical 10 Gbps of USB 3.1. However, I’m not entirely sure how this will impact performance. Will the Coral still function optimally, or will there be noticeable latency or frame drops?

I’ve done some preliminary research and found that others have successfully used USB 3.0 adapters with PCIe cards, but I’m eager to hear more firsthand experiences. Has anyone managed to run a Google Coral via USB 3.0? If so, how did the performance compare to a direct PCIe connection? Were there any specific adapters or configurations that worked better than others?

Another angle I’m considering is alternative solutions. If USB 3.0 proves insufficient, what other methods could I explore to integrate the Coral into this setup? Perhaps a different type of adapter or even a USB 3.1 hub? I’m open to suggestions and would appreciate any insights the community might have.

On a more positive note, I’m excited about the potential this setup holds. The Surface Pro 4’s portability combined with the Coral’s machine learning capabilities could open up a whole new realm of possibilities for my smart home projects. I’m particularly interested in enhancing my object detection and facial recognition systems, which are powered by Frigate.

In summary, I’m looking for advice on the feasibility of using a Google Coral PCIe card via a USB 3.0 adapter, any potential pitfalls to watch out for, and alternative solutions if this approach doesn’t work as hoped. Your experiences and recommendations would be invaluable as I navigate this project.

Looking forward to hearing from you all and sharing more updates as I progress! Cheers,
[Your Name]