Yes, I've seen it. I think it needs some adjustments. The sha256 algorithm requires you to "pad" your input to a multiple of 512 and I don't think this implementation does that. I'll need to write a circuit that does that, then I can pass the result to this function.