Gridcoin Guide for Odroid (Performance & Profitability)

in gridcoin •  last year


Gridcoin is a cryptocurrency that rewards users for running the Berkeley Open Infrastructure for Network Computing (BOINC). With BOINC you are not burning electricity in order to find a fitting nonce for a hash, you are actually contributing to scientific research.
To get started I recommend using this guide by @dutch.

In my first Guide I described all the little steps involved in setting up BOINC on an Odroid XU4. It's finally running stable, so in Part 2 we can now look at performance and profitability.

Switching to Arch

In part 1 I was using Ubuntu 16.04 minimal. Unfortunately I ran into some problems with it when it comes to stability. From time to time the whole device stopped working and every command I threw at it responded with an “input/output error” indicating a corrupted SD-Card. I formatted the SD-Card and installed everything from scratch, but the problem reaccured. I bought a new SD-Card and when it was running stable for a few days I published my first guide. A few days later I was back to the “input/output errors”. I knew that it was running smoothly for other users, but that was the point in time where I just wanted to set it on fire or to start something different from scratch.
I heard that some other users had stable configurations on Arch Linux, so I wanted to give it a try as well. If you have heard about Arch you probably know, that it can be a pain to set everything up. Luckyly Tholo pointed me to an installation script that helped a lot!

A few words about the installation script

The github page has some nice documentation about how to prepare everything. A little something I would like to add:

  • You need a Linux environment in order to prepare everything. If you are not running Windows like me, you can use VirtualBox. After creating a Linux VM you will have to install an Operating System (I used Ubuntu). Just download the image and add it to an optical drive in the VM and go through the installation process.
  • Your VM will need access to your SD-Card-Writer. To do this go to the USB settings of your VM and add the device there (for me it was a mass-storage-device in USB 2.0 Controllers).

The benefits of using Arch:

  • It's stable. It has been running smoothly for over 3 weeks now.
  • It's very slim.
  • Identidy is no longer a problem. In my first guide we did a lot of things to make the Odroid unique for the pool and the project websites. With this setup you are not using an image and will always receive an unique MAC address.

Performance & Profitability

A lot of people were interested in the actual Magnitude this thing could achieve. Now that we are stable, I can give you some numbers. Remember that MAG and RAC will vary in the future and if you are running other projects. There numbers are from running Universe@Home only.

RAC: 11000
MAG: 9.73

With the current Mag_Unit of 0.2 this gives us 1.946 GRC/Day.
Let's take a look at the current prize levels.

GRC price: 0.00000902 BTC
BTC price: 3745$ / 3131€

Earnings $: 1.946 * 0.00000902 * 3745 = 0.0657$
Earnings €: 1.946 * 0.00000902 * 3131 = 0.0549€

Power consumption full load: 13.5W (0.0135kWh)
Price kWh: 24.61 Cent/kWh (Germany)
Cost €: 0.0135 * 24 * 0.2461 = 0.0797€
Profit €: 0.0549€ – 0.0797€ = -0.0248€

As you can see the Odroid is currently not able to fully cover its costs in Germany. But all of these values are constantly changing. Competition, project selection, Mag-Boost, Mag_Unit, Electricity-Price, GRC-Price and BTC-Price are all having an effect on this calculation. Let's see at what electricity-price the whole thing would be profitabe: 0.0549 / (0.0135*24) = 0.1694€ or 0.2027$
Which means if your price per kWh is below 0.1694€ or 0.2027$ you would currently be making profit (ROI not calculated in).

You can look at it from a lot of different perspectives. You could say that it's just not profitable and drop it. You could also crunch like a mad man and hope for huge profits when GRC price climbs back to previous levels. Or you could just see all of it as a great technological adventure with the outcome of actually contributing to science for just 2.48 Cents a day. :)

Happy crunching!

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  

Wow, your energy is quite expensive, my energy is £0.122/kWh (about 14Eurocents, 17UScents), I think most users would find they would be in profit.
I run Raspberry Pis and you may have seen some of my guides if you look back far enough, I use Raspian-Lite, but DietPI is another good, lightweight distro thats easy to setup. I dont pool mine though, so I dont know about the uniqueness problem.


Yes, our electricity costs here are huge. That's why I calculated the price per kWh to be profitable (because most other countries are below that).
I tried to find a solution with solar in the past (running 24/7 and during winter as well), but that setup would have been way to big. Maybe I could try to find something to run it during daylight in summer or something :)


I have this crazy idea about a solar panel and a sensor that starts and stops a SBC when the power exceeds or drops below a specific level.

So if you have a 5v solar panel and your SBC needs 1Amp, you start the SBC at 1.1Amp and stop it at 1.01Amp, or something.
I'm not an electronics person, but I may get to it one day.


I have thought about that as well. But you would need something to measure the Amps as well. If you ever explore further, please let me know :)

Quite mind blowing.

With just 10 of these boards you would reach 6th position in the whole team Gridcoin in this project! With just ~ 135W power consumption... These boards are on par with modern desktop processors for a fraction of price and running cost - at least for some tasks.


My first reaction was that you must be exaggerating and then I had a look at boincstats. You are absolutely right :D


6th in team Gridcoin, 14th overall.... I wonder what machines Syracuse University is running, to match them close to 700 Odroids is needed...

Thanks for sharing. I have been planning for a while to do the same thing but too busy doing other things. I also see it as a fun experiment contributing to science. Once I get to it your experiences will greatly help me getting started. Thanks a lot!


Glad I could help you out. And you are right: Experimenting and figuring everything out is half the fun :)

Arch Linux is anything but stable, haha. However it shines in providing updated software, including kernel, as fast as possible.

Are the previous SD-cards corrupted to unsalability? Can you re-format and surface scan them?

In my case it was about luck. First three cards lasted only a month in my Pi. Then I bought a bunch of the cheapest, 30GB cards I could find. I used only one of them and it is running Arch on BTRFS for two years non-stop.


I feel like Arch is fine as long as you never touch a running system :D
The previous SD-Cards are both fine. I could re-format both of them. One is currently in the Odroid and running without any problem at all. I just checked the other one and did a surface scan - no errors. :-/


Great. So it is only a file-system corruption. It could be caused by unclean shutdown, insufficient power input, noise, overheating or kernel bug.

Awesome ... given the cost of solar or wind energy in some places this is actually a valid option for considerable profit!


Im on solar, so for me all mining is free in the daytime :)


Cool, check out claiming your free SolarCoin if you didn't already

Woah it has the same magnitude like my i7 6700hq running 24/7 at 50%, how much did you pay for it? I've taken an interest ;)


It has 8 Cores (4x 2000Mhz and 4x 1400Mhz), so it's pretty beefy :)
You can buy directly from Hardkernel from South Korea for 59$.
I paid 79€ at Pollin in Germany.

Good post, thanks for sharing your experience and profit calculations!

I have three of these crunching this project. Currently, their RACs are 12.3k/11.8k/11.5k. You still might see yours go a little higher.

Luckily, I pay $0.075 USD for power :-)


Yes, my mag is still climbing very slowly :)
I would love to have 7.5 Cents per kWh ... And are you actually running 27 SBCs on Universe@Home or am I misinterpreting something?! That's crazy. I guess I have to move somewhere with cheaper energy... :D


Yes, it's true. 23 SBCs, a laptop, and a workstation on universe@home. 3 SBCs on tn-grid. A laptop and my phone on WCG. 30 machines total.

BOINC is eating around 600 watts and my mag is about 150. It's about break-even at current prices (given my power cost). The SBCs make a profit, but the laptops and the workstation can't pay for themselves.

Congratulations @thorondor! You have completed some achievement on Steemit and have been rewarded with new badge(s) :

Award for the number of upvotes

Click on any badge to view your own Board of Honor on SteemitBoard.
For more information about SteemitBoard, click here

If you no longer want to receive notifications, reply to this comment with the word STOP

By upvoting this notification, you can help all Steemit users. Learn how here!

Hmm, your RAC is way higher than mine on yoyo, but the measured floating point and integer speeds are similar.

Seems like you get a bit more mag from universe than I from yoyo too, I might have to switch projects!

Most the xu4 I see on projects stats pages get around 1100 million ops/sec floating point speed and 5500 million ops/sec integar speed . I dont really know what this means lol, but it seems good in comparison to what my pc gets


You can not compare RAC between projects because they reward it differently.
Project selection for maximum profit also depends on what kind of CPU and GPU you have.
If you would like to read more about it, my link-collection might help you. There are a lot of articles that help you optimize your CPU for example :)

I would personally advise to use android. You will get more, and more profitable projects that way.


Yes, there are some projects that are for Android but not for Linux-ARM (
WCC, SkynetPogs, SETI, Rosetta, Moo, LHC, GPUGrid). But in most of them you will compete against GPU anyway, so they won't be more profitable.
Is there any specific project you are referring to?


Rosseta is quite profitable in arm. It doesn't use any SIMD so arm has a very good performance when compared.


Thanks for pointing that out, I might try it in the future :)