Zeropond Cloud Mining

re: Proving identity across multiple online accounts. That seems like just the right application for a cryptographic signature… :wink:

Zeropond official GPG Key. I’ll get it on the website too.

-----BEGIN PGP PUBLIC KEY BLOCK-----
Comment: GPGTools - https://gpgtools.org
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=38MF
-----END PGP PUBLIC KEY BLOCK-----

2 Likes

Onename for blockchain ID

Verifying that +timhosek is my blockchain ID

What are you plans if your datacenter starts mining 51%+ of new blocks?

So the price went up today, is there a reason for this?
Now we have to pay more than a month before live mining, what if a GPU miner comes out by then? will the pricing change, what if my 4 GPUs are equivelent to the 50 cpu cycles ?

20 posts were merged into an existing topic: Errors Mining in VPS

Coindup: I have moved your conversation to another topic. Please keep each thread on the subject it began with.

Hello All,

I plan to live stream the start of Tim’s testnet validation of Zeropond’s mining software. We are hoping to coordinate a 5pm PST start, we will validate the timing as we get closer: Zeropond Zcash testnet validation, Live Stream - YouTube

Note: I’ll be there as a warm body to lend validation of the events as they occur to the forum. I am not a cryptologist and will not be inspecting the software test harness. I will rely on others like Austin-Williams and Zawy to interpret the results and the validity of the test setup. I think Tim was asking for input as to the format of the test and I did see Austin’s response but I really don’t know what the actual test will be.

where will it be at? I’d like to participate, at the data centre in Fremont?

I will interpret the results of the test by the following. Others should critique this methodology and Zeropond needs to state if they agree and if not, how they will interpret the results before the test is started. Complaints and math expressed after the test are not allowed by standard scientific methods that involve statistics or if a bias exists towards a particular outcome.

zawy 2013 desktop CPU equivalents = 1.2 x Difficulty x blocks gained / total blocks issued
zeropond laptop CPU equivalents = 1.426 x Difficulty x blocks gained / total blocks issued

zawy (my) CPUs are 0.044 H/s when benchmarked on 1 core. 4 cores for me are apparently 2.70x single core benchmark when on the network (this is updated from the 2.84x and I reserve the right to update this). By measuring the ratio of blocks I get to the blocks issued, 1.2 x difficulty is the number of CPUs like mine that are on the network. This is independent of any hash rate consideration. It is what I have measured. As more data has come in I have changed this from 1.1 to 1.25 and now 1.2 is the best measure I have.

Using my math above to convert difficulty to hashrate, I get H/s = 1.2 x Diff x 0.044 x 2.70 = 0.1426 x Diff =
network H/s = Difficulty / 7.0
GPU H/s = D / 7.0 x blocks gained / blocks issued / GPUs used

I get their 1.426 factor in second equation from 1.2 x 0.044 x 2.70 / 0.1.

I was told H/s = Diff / 10, but since the network is not a good Poisson process, I can’t let theories override my observations. Someone else may wish to argue I am over-estimating their H/s with this. D / 7.0 ties 4 cores on the network back to H/s for 1 core via observation. Unfortunately, the observation weak link is a dependence on 63 data points which has 95% confidence interval of about plus or minus 30%. It could be H/s = D/5 or H/s = D/10, but again the solves are not a Poisson process so I can’t let my potential statistical error override theories using a known error.

The math above is more reliable than my block collection will be, as it has a lot more data points. So I will turn off my CPUs to try to help the difficulty lower. The more the difficulty is lowered by people turning off their miners, the more data Zeropond will have. 100 blocks is not overkill. 30 is not enough. Either the total blocks collected or the stopping time must be stated before the test.

30 blocks is not enough. This is why I’ve been pushing them to hurry up with it and I don’t see how starting the test would require more than turning it on. 50 blocks are needed. 100 would be better.

[edit: my definition of what they consider a CPU-equivalent is based on their selection of 0.1 H/s and what I am measuring to the the multiplier from 1 core to 4 cores, and my guess that 2013 desktop is close enough to being 12% faster than a 2015 laptop, that I will let them say 0.1 H/s is 1 laptop CPU without complaint. ]

[edit: I may change the 2.70x factor as more data comes in today and this will affect the H/s and their CPU equivalents but not my CPU equivalents. ]

I don’t think the datacenter is setup yet. This will be held at Tim’s
apartment, close to Precita Park in San Francisco. Since this is not my
apartment, I’ll let Tim do the invites.

My setup has just a single GPU, a Radeon R9 390X, running at about 1.1 H/s. I left it on overnight and it got 18 blocks in about 7 hours. 7 hours should have 7 * 60 / 2.5 = 168 blocks. According to zcash-cli the diff over the last 168 blocks is:

./zcash-cli getnetworkhashps 168
8

You cannot just divide difficulty by 7. Please use something like the above command-line. Unfortunately zcash-cli truncates the output to an integer. :frowning:

By zawy’s formula I should have found 1.1 / 8 * 168 = 23 blocks but I found 18. So it looked a little slow but it’s within error bars for sure. Finding blocks follows a Poisson process, and the variance is equal to the mean.

So if you want 30 blocks, someone will need to be filming nonstop for like 24 hours…

I will also do the test @Austin-Williams proposed, which can be done offline over a longer period of time.

As soon as we get our first order from GPUshack and the 6-slot motherboards, we can run a live demo much faster…

@mungwi I’m happy to meet today if you want to see my setup. As I mentioned in the above post, we will need to run Austin Williams’ offline test to prove hashpower, but you are invited to my house to see my setup and watch it solve a block or two. Please contact me at tim@zeropond.com to arrange this.

I seriously doubt we will be mining 51% of the blocks. I know of at least two other big farms, plus there’s probably a large ETH farm that developed a GPU miner in secret. But just in case these other big players do not start right at Genesis block (maybe due to Slow Start), we will throttle to 40% hashpower maximum. But that shouldn’t be necessary after a week or two when the block reward is significant.

Our prices per hash did not actually change. What we changed was the amount of hashpower per “typical CPU”. We responded to zawy’s criticism in this thread that our estimate of a typical desktop hashrate was too low, so we increased our benchmark rate to 0.1 H/s. This makes it look like you are paying more-per-cpu but the price per hashrate is the same as before.

Sorry I forgot to address your other concern: what if another GPU miner comes out? If there is a better GPU miner than ours, we will use it. We are here to provide people the best hashrate from the most efficient mining operation. But I really think our miner is ahead of the competition, and when we get the datacenter set up, we will be able to put time into optimizations, which we will share with our customers. So hopefully by the time a public GPU miner is doing 1.2 H/s, we will be doing more than that.

1 Like

As I said, the network is NOT following a Poisson process and str4d explained why. As I explained before it may give you more blocks than your actual H/s should allow. Instead of something like 26% blocks below 30 seconds, there are about 3%. So if getnetworkhashps assumes a Poisson process, it is wrong.

It sounds like you do not have an average for the difficulty last night. If you give me start and stop times, I can tell you more precisely. It averaged about 10% higher (D=86) than the last 168 (D=79). This gives 86/7 = 12.5 H/s for 7 pm to 2 am EST. And you should be very happy with my method as it calculates your GPU hash rate to be

86 / 7 x 18 / 168 = 1.3 H/s

But your exact start and stop times would be nice to know, along with confirmation that you did not forget to include your listunspent blocks.

1 Like

Man, at least two other farms… I think our CPU dreams are quickly fading… How will any small miner be able to keep up, even with a brand new CPU and DDR4 mem?

If you’re pulling at least 1.1H/s on 1 R9 390x, game over for the small miner. :frowning:

I am agreement, I hope the original mining of the coin will be closer to Satoshi’s vision rather than every other coin. Although I am going to hold off judgement until the test later today. I will be paying attention to the results posted. Hopefully we can see a thorough analysis.

You can argue it’s not exactly a Poisson distribution because there’s a minimum time to create one solution, so during that time your odds are 0. But since we are doing about 1H/s, and the blocktime is 150 seconds, the discrepancy will be very small since 1 << 150. So I will argue that it is very nearly a Poisson distribution. It won’t be Poisson for a CPU miner that takes 20-40 seconds per solution, but it will be almost Poisson for us.

Which two other farms? Any source ?