Let me make it straight.
His miner is fast indeed, but it's proprietary Windows only with a dev fee. Considering size of the NeoScrypt mining market, it could make him a few bitcents daily. On the other hand, NSGminer is open source and therefore funded entirely by user donations. There haven't been any for the last year despite a dramatic increase in mining profitability. You can check yourself with the donation addresses in the OP. I could assume the people don't care about mining performance, but this thread's activity suggests the opposite. If all those miners don't bother themselves to donate anything, why should I bother myself to write faster kernels to please them?
If I don't get 0.15 BTC in donations to cover the cost of a single decent Vega 64 card for development purposes, I'll be greatly discouraged to upgrade the performance of NSGminer to match Claymore's one.
Some people like to complain about greedy developers if they use devfee, but same people don't want to donate even single day of mining for all efforts that were put into mining software. I donated 0.15BTC, it's devfee from my NeoScrypt miner for about two weeks.
That's very generous, thank you. There is nothing wrong with the dev fee as long as the software is all your work, in fact, a lot of your work. Those who violate the GPL by selling private miners based on open source code don't play fair.
BTW, be ready to spend a lot of time if you want to speed up your miner on AMD, I spent more time than I expected...
I realise that clearly. I've spent much more time than I've expected on my CUDAminer while the gains aren't very impressive compared to other CUDA miners, but I've identified a couple of bottlenecks in my OpenCL code at least. I haven't tested your miner myself because my mining rigs are Linux only except of the oldest one still running Windows XP 64-bit which doesn't support any AMD GCN GPUs obviously.
