There hasn't been enough incentive to do it. If someone offered a large enough payment, I could write a Linux ZEC miner that is at least 10% faster than Claymore 9.2, and probably have it finished in under 30 days.
OK, how much for a 10% faster miner?
10BTC would be enough.
Sure you can
I actually think I could get 25% faster performance, but would only promise 10% to play it safe. It would be even easier for Wolf to do it given his expertise with GCN assembler, but he seems to be having more fun doing other stuff.