GPU Miner to be released? Optimizations?


#1

I'm just curious as to how the Z.Cash team plans on releasing mining. Thus far in every single cryptocurrency that I'm aware of - CPU only has turned into a GPU coin. Primecoin, Monero/Bytecoin, Ethereum, etc. In addition most of these coins have had private miners running on GPUs at one point in time that made the mining field unlevel.

From a moral standpoint I don't have an issue with it - but security wise it seems like a potential problem. And it limits (a little) new entry. More people involved is better than a few proprietary miners soaking up all the speculation revenue.

Anyway. I guess my questions are

A - How many optimizations do you plan on putting in mining? Have you thought about outsourcing to someone like Wolf0 or Claymore (who've both built optimized miners for the coins I mentioned above).

B - Do you plan on releasing a GPU miner for the coin or letting the private market build it? Doesn't that kinda create a security hole? If someone has a miner that's 2X as fast it's half as cheap to attack the blockchain. 5X as cheap etc. Without some amount of optimizations there's not a good way to gauge what they will be.


GPU miner software
#2

It's hard to tell at the moment how much of our development and/or financial resources we'll be able to assign to mining optimization. Yes, ideally there will be well-optimized public, open-source implementations, including GPU implementations, in order for everyone to be on a level playing field. I doubt that small constant-factor variations in efficiency are likely to be a major security threat in the short term, though. The differential between CPU and GPU implementations will probably be smaller than other coins due to the memory-hard proof of work.


#3

Note that Equihash is not intended to be "GPU-resistant"; only to limit the relative advantage of a GPU over a CPU. There is some discussion of the relative efficiency of CPUs and GPUs for parallel sorting, which is designed to be the main performance bottleneck, on page 10 (section VI part b)) of the Equihash paper.


#4

Hi Daira, your post is quite old but none the less only few months old, the paper you are referring to mention a GTX480 GPU as being the fastest which is a card released back in 2010, now obsolete. I don't think they have tried any card released in 2014/15 which is when the GPU cards game changed. I would be careful referencing to papers without the proper knowledge. Whilst there is a slight but not exponential improvement from a I7 3.2 Ghz tested on that paper (4.2 Ghz being the fastest commercial product and only 30% to 50% faster than the one tested) there have been exponential progresses in terms of GPU, which means the x4 factor between CPU and GPU mentioned on that Equihash paper have easily become a 10x factor and considering that multiple CPU deployment on the same board is harder and a lot more costly than multiple GPU (the max CPU speed in a multi CPU system is 3.5 GHz and the cost is $6000 each CPU) and that most GPU mining systems have 6 GPU (60x factor) suggest that once again CPU should not be referred as a viable mining solution.


#5

I'm wondering if it's possible to tweak the Equihash parameters to decrease the GPU advantage. Or are they fixed at this stage?