5

I'm looking to mine crypto on a Raspberry Pi Zero, now I don't mean coins like Bitcoin or Ethereum, I mean ones that are easier to mine.

I have read a Pi Zero can do about 10 H/s which would make not much profit at all, but if I created a cluster of let's say 10 Pi Zeros would the hashrate be more? Or would it just be 10 x 10 H/s the Pi Zero already does?

closed as primarily opinion-based by joan, tlhIngan, Steve Robillard, Darth Vader Jul 24 '18 at 9:32

Many good questions generate some degree of opinion based on expert experience, but answers to this question will tend to be almost entirely based on opinions, rather than facts, references, or specific expertise. If this question can be reworded to fit the rules in the help center, please edit the question.

  • Why would you use a device that does not have a decent GPU? GPUs win over CPUs when it comes to pararelized math tasks which cryptocurency mining is. – Tomáš Zato Jul 19 '18 at 22:23
  • 1
    @TomášZato there are a number of CPU only coins right now. – Tim Jul 20 '18 at 11:26
  • @Tim How would you make it CPU only? Mathematical concept is not limited to any means of calculation. – Tomáš Zato Jul 21 '18 at 14:19
  • @TomášZato I’m not an expert, but I gather by making not easily to parallelise? GPUs only win when it’s lots of the same. I don’t know how one would prevent that, but I believe there are examples. – Tim Jul 21 '18 at 15:56
9

Clusters generally work well for tasks that can be parallelised—i.e. you can split the task into lots of small ones that can run at the same time. Fortunately, cryptocurrencies are ideal for that, and a mining pool is essentially a really big cluster for mining cryptocurrencies.

Since proof of work essentially boils down to trying many combinations and hoping to find the correct one, all your cluster would need to do is split these combinations between each Pi so that they can work on them separately. Once the tasks have been split, the Pis can work at full speed testing. I don't imagine in this case the overhead for clustering would be at all high, so in this case I would expect ~100 H/s. If there was more overhead and the delegation of tasks was more complex, then you might get less than 100 H/s. But you could never expect to get more than 100 H/s, really, because each Pi can still only work at a certain rate.

The implications of the above are that if you would make a loss with one Pi, you would still make a loss with a cluster of Pis. You can't scale up the cluster and expect to get more efficient in this case as the limiting factor is the hash rate of each Pi.


As is often the case with the Pi, it's not really optimised for any one task very well.

I plugged some extremely rough numbers into this calculator for Ethereum (though you said you weren't interested in it as much, it is still useful for comparison):

  • Hash rate of 100 H/s (10 Pis at 10 H/s, which according to various sources for Ethereum, is about right)
  • Power consumption of 12.5 W (based on 1.25 W, roughly derived from the linked pd and current figures)
  • $0.08/kWh (based on the lowest state's cost from here)

  • 0% fees (assuming you manage the cluster yourself)

and you can expect a cool profit of... -$0.70 per month. (That's a loss—or as the site kindly puts it, a profit ratio of -100%)

Not great, as you can tell; the electricity cost will certainly outweigh any profits if that hash rate quoted is true. You would break even at 751 kH/s... much more than your 100 H/s.

Bear in mind also that your local electricity cost is probably higher than $0.08/kWh (I would expect to pay double that where I live, at least).

Apparently there are some coins like Magi Coin that can yield $0.20 - $0.25 per day per Pi 3, in which case clustering may be profitable.

A fun project, if you're interested in that sort of thing, but realistically don't combine "Pi mining cluster" and "profit" in the same sentence. Even if you did make some profit with Magi Coin or something similar, the payback time is likely measured in years before you start making any real money yourself.

  • The actual question was about what hash rate you would expect from 10 Pis at 10H/s; not about the profit or anything like that... to me this answer is off-topic apart from one line stating without explanation 100H/s from 10 pi's – Joshua Lin Jul 20 '18 at 0:05
  • 1
    the problem with crypto is that there are a lot of n00bs who don't look into processing and electricity costs – ina Jul 20 '18 at 0:36
7

You wouldn't gain any extra magical power, as you can't create more energy than you put in. It would just be a maximum of the power of a single Pi multiplied by the number of nodes in the cluster. So, assuming one Pi produces 10 H/s, as per your example, overall the maximum for 10 Pis would be 10 x 10 H/s.

However, you could/would never attain that maximum, as clusters require some power to manage, and communicate between, the nodes. Generally speaking, depending upon your cluster set up, you either:

  • dedicate a single Pi for management (and not for mining) or;
  • distribute the management across the Pis.

The first example, would be 9 x 10 H/s, i.e. 90 H/s, if you had 10 Pis and set one aside for management.

Taking the second example, you would obtain about 80-90 % of the computing power for mining, the other 10-20 % would be used on management, so, the maximum hashrate would be something in the range of 8 or 9 x 10 H/s, i.e. ~80-90 H/s.

To reiterate, these example figures are approximations and you would probably not even obtain those speeds.

However, give it a go, as it sounds like a fun project.

Not the answer you're looking for? Browse other questions tagged or ask your own question.