How big is gridcoin mining hardware network?
I estimate that to match computational capability of the network someone would need around 2000 (+/-50%) computers based on Ryzen 5 processor and Nvidia GTX 1060 6Gb graphic card, that run 24h a day. Such set-up costs close to $1000 and is good to be considered as a reference computer, at least for the needs of this article and as a representative of an average amateur miners hardware.
There are several benchmarks and ways to measure computer performance. Some computations are based on integers, some single or double precision numbers. A processors with many relatively slow cores might be faster in some tasks than one with few but fast cores, and this one might be faster in some other types of tasks. Proper benchmark needs to be multidimensional. There are miners who run 24/24 high end computers, multi card systems, but some use budget computers or even smartphones for short periods of time. Above reference computer is a big compromise, but useful here and I think quite representative.
Currently (September 2017), each day there is around 18800 coins mined through PoS and around 28000 minted through PoR (even though target is ~50000, but for reasons that I don’t known magnitude unit is too low). This means there is only ~ 10 GRC per our reference computer a day. At current exchange rate of 1 GRC ~ $0.025 it means $0.25 revenue a day or ~$90 a year.
Load power consumption is at least 250W, this leads to 1 year energy consumption of 2200 kWh. Cost of electricity varies from
- $0.08 - $0.12 per kWh in China, Canada, USA and Russia
- through $0.20 in France and UK
- to as much as $0.40 in Denmark.
Cost of 2200 kWh at average price of $0.20 is $440 or $220 in cheapest countries. Thus where you live has big impact on chances to mine profitably. To cover electricity costs at $0.20 per kWh miner would need to earn 48 GRC per hour. Well, electricity costs during cold winter months could be compensated to some extent would you consider mining rig as a heater!
As computer is under high load for long periods of time, I assume it is not worth much after 2 years and miner - to stay competitive - needs to upgrade to newer hardware. Thus he needs additional revenue to cover hardware costs at half the price of hardware a year, assuming (naively) that he will steadily make 10 GRC a day for 2 years and GRC price stays constant. To cover hardware costs he would need to mint another 55 GRC a day.
In total, just to break even, miner needs to earn 100 GRC per day with a reference computer (Ryzen 5 plus GTX 1060). Mining focused rigs, like 4 x GTX 1070 cards on one motherboard would be of course a bit more cost effective.
Rising price (exchange rate) fallacy.
‘Mining might not be profitable at current prices, but IF prices will go up, profit will be huge!’ It seems true, but there is other, more profitable way, if IF would happen. Instead of mining, it’s easier, hassle free in the meantime, and more profitable to just buy some gridcoins.
Lets assume GRC will go up four times within a year from $0.025 to $0.1. With our reference computer you spend $1000 on hardware and $440 on electricity and earn up to 20 GRC (assuming daily PoR bounty will go up to 50 000 a day) a day or 7300 within a year, worth $730 (at 1 GRC = $0.1). Your initial investment was $1440, now you have gridcoins worth at most $730 and a computer used heavy for one year, worth maybe $500 if you would wish to resell it. Effectively you have lost at least $210. Plus all the work you’ve done was unpaid. Would you invest $1440 in gridcoins, after a year you would be worth $5760. With GRC price up tenfold you would make some profit on mining, but still much less than with just buying and holding strategy. Well, would GRC price fall to zero, in first case you would still have a computer and be proud of scientific research you’ve done!
Amazon offers p2.large service that costs ~$0.1 per hour - one of two GPUs of K80 card, which is, what I’ve read, equivalent to K40 card. Surprisingly, K40 card capabilities are for some tasks comparable with GTX 1060 performance. While worse on graphic card memory size, FP64 calculations, our reference computer would match FP32 capabilities of K40 and deliver probably more CPU power than p2.large offers.
Cost per year:
- $876 for p2.large (1 GPU)
- $912 for reference computer (at $0.2 per kWh electricity cost)
As we see cost is comparable. With cost optimized mining rigs running on lower priced energy, with some miners mining more for fun than profit, gridcoin network could offer even lower prices than AWS for similar services.
Gridcoin – just another BOINC credit, even though blockchain based and transferable?
Currently, bitcoin and altcoins prices are almost purely speculative. This why 20%, even 80% price changes during just one day are common.
Frankly, I don’t see much (economical) value flowing into the Gridcoin system. We can argue that research is valuable, what is true, but it doesn’t make BOINC credits to have market value. Gridcoins seem to be not much more than alternative BOINC credits, that are blockchain based and transferable. Some could buy gridcoins to support science, some to start solo mining and receive gridcoin badges for BOINC calculations. As tokens are transferable and part of booming blockchain phenomenon, others might buy some coins in hope of making good profit in the future. These reasons do not sound like a solid base for long term demand for gridcoins, while there is a constant supply.
For the network of our 2000 reference computers, to make it sustainable, we need an influx of value equal to around $3 million a year. Let’s dream a bit and let’s assume gridcoin hardware network would reach computing capabilities equal to one million of our reference computers. To support such a network we would need over one billion USD flowing into the system each year. Frankly, I don’t know how big or small market is for services that could be offered by Gridcoin network. However, big data, AI (like deep learning) are already generating dozens of billions of USD and demand for computational power is rapidly growing.
To support scientific projects we already have a bounty fund of minted coins. Commercial multi project sites (like Citizen Science Grid or yoyo@home, where several sub project are run) should be implemented. Clients could buy gridcoins and set up a bounty for computations they need. This way miners income would come from two sources – Scientific-PoR bounty that already exists and Commercial-PoR bounty.
$1 bllion market cap.
M2 money supply and GDP values for various countries are quite close to each other. Assuming gridcoin network turnover is an equivalent to countries GDP, at $1 billion revenue, gridcoin fundamental market value would also be close to $1 billion. Assuming miners make 10% profit, using capital market P/E =10 ratio, we also come to $1 billion. For several reasons (like growth expectations) actual market cap could differ greatly.
Would scientific projects suffer with introduction of commercial projects?
Someone might argue that commercial projects would draw computing power from scientific projects, which are in the heart of Gridcoin. I think, on the contrary, they would only benefit. There is already thousands of people supporting BOINC projects and covering costs of hardware and electricity from their own private pockets. They would stay with BOINC. At Scientific-PoR bounty at 50 thousand GRC a day or 18 million a year, at market cap of $1 billion and assuming, for simplicity, 500 mln gridcoins in existence, Scientific-PoR bounty would be worth $36 million. This would sustainably support 36 thousand of our reference computers. Also, some profit oriented miners could sometimes point their rigs to scientific projects.
I know commercial projects idea is nothing new and was already discussed in detail, but it seems nothing has been implemented so far.
Many years ago my computer was running seti@home. Two weeks ago I’ve bought some gridcoins, set up my desktop computer for BOINC projects and started mining gridcoin / supporting network.