I don't know about AMD, but NVIDIA GPUs also have internal protection. Tesla and Quadro GPUs have had internal error detection on registers and cache since Ferm

I don't know about AMD, but NVIDIA GPUs also have internal protection. Tesla and Quadro GPUs have had internal error detection on registers and cache since Fermi at least. But this is totally irrelevant anyway, the comparison we started with is to an analog alternative. Anything analog will have strictly worse noise and error problems. In neural networks, errors are probably not even a problem for analog implementation, so they definitely aren't a problem for GPU implementations.