edgetpu-yolo VS edgetpu

I'm on the fence. It's a very nice device if you can get your models working on it - basically untouched at the price/power point. Drivers for me have been OK. I have an M.2 card connected to a Jetson devkit (makes for a nice embedded test bench) and it runs fine, no worse than the NCS for setup anyway. There were a couple of PCI settings to tweak but I documented the setup here [0]. For common use cases it's a decent option, I think. For custom models you really need to know what you're doing.

The main issue I've had is that the compiler behaviour differs between versions (and it's very difficult to find older releases), so where previously you could run a big model and delegate things to the CPU, now it sometimes won't compile at all. There were also problems where we trained a model in AutoML - using free credits but the real cost would have been over $100 - but edgetpu compiled model lost a lot of performance. The developers have been very helpful when I've contacted them, and generally you can get through to real devs (not generic support) who can look at your model for you. Mostly I think you need to take care when training models for these devices, but quantisation-aware training is not trivial to use in Tensorflow and there are only a few off-the-shelf models which are supported in the various toolkits. Model maker looks promising, but it's also finnicky in my experience [1].

I'm not super worried about hardware availability. They're suffering from the chip shortage like everyone else, so it's not surprising that lead times are long. I was able to buy my device in late 2020 without any trouble.

[0] https://github.com/jveitchmichaelis/edgetpu-yolo/blob/main/h...

I know the tensor power pixelbook was shutdown and I never heard the actual reason just a bunch of speculation about costs/profitability which is probably true.

It's a shame that there isn't more competition and development in the neural asic world to harness the power of llms/generative AI on a low power, cheap hardware platform like the pixelbook line. For someone that invented the TPU they have done a not so great job of ensuring it's commercialization and support. Both on the hardware and software side.

The coral edge tpu seemed to be the right high level idea but without proper execution.

https://github.com/google-coral/edgetpu/issues/668

> even in the 80s I wanted an FPGA accelerators in every machine

Mostly unrelated, but I recently discovered that you can buy TPUs, right now, as a consumer product, from https://coral.ai.

The stock firmware already allows you to run these things so hard they overheat, which is amazing.

But yes, I also want FPGA accelerators.

I'm looking at the coral.ai cards and they have a M.2 A+E card, same form factor as the wifi slot in the m900 tiny. Has anyone tried another card in that slot other than wifi?

Chips optimized to perform the type of calculations used for NN inference at high parallelism. A good example would be the google spinoff https://coral.ai/ (though their usecase is highly limited by sub-par software constraints)

By no means an expert, but I have seen prototypes using a raspberry pi and a dongle from Coral Ai. They have PCIE and USB based modules.

Last news on https://coral.ai/ was on May 5 2022

Activity on the github project seems to have stopped. https://github.com/google-coral

    https://coral.ai/

Your problem statement still needs more clarification. If the above applies, the best way is to evaluate your ML-based pattern matcher on high-level scenarios. One approach to speed up the evaluation is to lift and shift the execution of scenarios into cloud. Another approach is to use an AI accelerator, such as http://coral.ai or other.

You probably could but not with raspis. Maybe the TPUs they sell. https://coral.ai/