UPDATE
I’m disappointed to report that this Crowdfund has been unsuccessful as it was originally planned. The main developer who was working on the project has decided to take a private offer.
The support for such a project is immense, there must be a developer out there who is willing to take it on.
I’m re-opening this thread for suggestions on how to attract developers who may be interested in taking advantage of this unique opportunity.
So @ampy has gone with someone else?
This isn’t what I was thinking at the time but it is another good reason for the bounty to be open to whoever is willing to deliver the goods.
I am also committing 1BTC for the GPU miner crowdfund.
So sorry to hear that Ampy took a private offer - not good for Zcash 
Hopefully somebody else will take our offer. But time is running out!
Must have been a good offer.
Can we agree on a list of specific qualities required of the devs?
I don’t want to tap my resources, get someone to invest a ton of time on this, and then have him not get paid upon completion. That would really burn a connection I value. I know someone who is expressing interest but wants more details.
Can we make a bullet-pointed list of the minimum qualities needed to have the funds released from escrow?
Some thoughts from one with only a few graphics cards (4 st RX470)…
Why not offer two different options for the same miner?
1: 1 BTC one-time fee (0% mining fee)
2: Free (5 % mining fee fore 1 year)
(After one year the miner software shall be open sorce)
Then you who have large mining rigs can pay a one time fee and give the developer a substantial amount direct.
And all with just a few graphics card can use the free version (5% mining fee) and it gives the developer a variable compensation also.
The advantage is that the developer then has a reason to continue to develop the software for one year, and not only meet the minimum requirements.
and last but not least , it would benefit z-cash by increasing the number of possible miners and reduce centralization.
/ Nimos
@Austin-Williams I agree,
One of the main reasons this crowdfund fell through was that you have way too many cooks in the kitchen. Everyone has a different idea of what they want. The only way this can work, keep everyone happy, and keep the developer sane is for the developer to lay out what he is willing/able to do from the start.
So I propose that the developer who wants to claim the bounty:
This way everyone knows what they are getting into from the beginning.
Would it be possible to gather the funds in escrow? If so, Wolf0 and I can dedicate some actual time to it, then. Even though there is interest, the issue is that usually these things go with, “Yeah! I’ll donate!” but donations never actually come through.
I need to ensure that Wolf0 can get a steady income before we can really throw our full weight behind this.
Also, regardless of what offers we do get (if we develop one), we’re not selling it to a third party - so rest assured that if it does happen, it’ll be released to the community.
This is a great point. If we re-boot this crowdfund then we need to gather the funds into a trusted multi-sig escrow before the work begins.
This also has the caveat that the developer has to set a target amount that they want for the work performed. That way everyone can see that an acceptable amount has /has not been met.
Depending on contribution amount, If it’s not enough, the crowdfund may be over before it begins.
Put it this way:
It is a lot of stress and work to develop. We barely make enough to make ends meet. So, say we do spend anywhere from eight to twelve hours a day developing the miner (assuming we don’t burn out): that’s eight to twelve hours a day we’re not making guaranteed money for bills, for medicine, for our families, etc.
If there is a bounty - a proper bounty, with the funds in escrow - people are more likely to jump on it, us included. But as it stands, even I can’t convince Wolf0 to work on this fulltime with me.
I completely understand your and Wolf0’s situation, it is not easy to do this work and it is not fair to start work and have it fall through.
Let’s start with a few basic questions the community may be interested in:
Since Zcash / Equihash is so new do you think it’s possible to create a GPU miner for?
What kind of gains do you feel you could achieve vs the basic CPU miner?
How long would it take you guys if you were able to convince Wolf0? Deliver Before/After launch?
What hardware would it be for? AMD/ Nvidia?
Do you know of a trusted Escrow service we can use?
How much would it take in Escrow to accept the bounty and begin work?
*It also should be noted that I am soley acting as an mediator for this effort because I feel it would benefit the Zcash community. I am only a forum moderator, I am not paid by Zcash Company nor do I represent Zcash Company
Yes. We already have a c++ implementation in zcashd, and that can be optimized (modestly) for GPUs.
Thats the thing. The Equihash algorithm doesn’t get much of a speedup over CPUs via parallelization.
The performance improvements over CPU mining are going to be very modest (as in, less than a factor of 5 between a CPU accessing DDR4 RAM and a GPU with access to GDDR5). They discuss this on page 6 of the Equihash paper under Corollary 1.
Even with a screaming fast modern GPU with 4,000 cores, you’re only going to be running 4 or 5 of those cores. If you try to use more than that you’ll actually be slowing yourself down.
I’m just bringing this up because I want people’s expectations to be realistic. We are looking at a very modest performance increase here. And it will come at a higher cost.
For example, if you are thinking of GPU mining on AWS, just stop now. GPU instances on AWS cost more than 5 times the standard compute instances. So you’d be paying a higher price per coin mining on 1 GPU instance than you would mining on 5 CPU instances.
This GPU miner will only be cost effective for miners who own their own GPUs. And we haven’t event talked about the difference in electricity cost yet.
Tl;dr: Keep your performance expectations modest.
I’d trust @str4d.
Would it be possible to run the CPU and GPU miners simultaneously on one machine? So I could use the CPU and lets say one GPU at the same time?
As long as this project is open source (and written well enough) I would expect it or a fork of it to quickly become as sophisticated as the miners used on other networks.
Um… Go away? Did it ever occur to you that the point of open source software is to encourage people to build from it and grow support from multiple efforts?
I think that should be possible.
I will throw in 1 BTC as well.
I don’t believe we should complicate things by adding in additional qualifiers. I feel the main goal is to get working gpu miner out so as to level the playing field against those that are paying for a privately developed gpu miner. To that end, terms should be as simple as possible, such as
(1) working (gets us around 4x cpu mining result)
(2) open source