In this way, the strongest supercomputing Summit in the United States can earn 79 thousand dollars a day and earn 2 million 360 thousand dollars a month if mining is done, but it will return to this cycle for 9 years.
It's not just a joke. It's actually a fun thing. Especially after the most powerful American Summit came out, it used 95 percent of the CPU GPU heterogeneous system to speed up the NVIDIA Tesla V100 card. And graphics cards are the best tools for mining, and pcgamesn really does calculate how much Summit can make.
Summit supercomputing each node has 6 Tesla V100 acceleration cards.
The process of calculating the income of supercomputing mining is the same as the principle of ordinary graphics and mining. First, it is necessary to make clear the performance of the supercomputer. Summit uses 27648 Tesla V100 acceleration cards, and the performance of the single card ether workshop is about 94MH/s, the total calculation force is about 2599GH/s, which accounts for 1% of the current total calculation force of Tai Fang.
In addition, the price at the ether square is now $534.
Then the cost is calculated. The power consumption of Summit is 13MW, which is 13000000 watts, and it is installed at the Oak Ridge National Laboratory in Tennessee, USA, with an average electricity cost of 0.06 cents / kWh. Considering that this supercomputing is sponsored by the US Department of energy, the electricity bills are likely to be discounted, but they are still on average.
Now we need all the data. We can figure out that Summit supercomputing can earn US $79 thousand a day and earn US $2 million 360 thousand a month. Does it sound like a good deal? It makes about 15 million 110 thousand RMB a month.
But the key to the problem is the cost. Summit uses a large number of IBM Power processors, NVIDIA graphics cards and memory, hard disk. The 27 thousand Tesla V100 graphics cards are $221 million, more than 9000 Powero processors are $10 million, and ECC has more than 30 million dollars in memory. These main components will be $265 million, and Su to use Su MMIT supercalculation of the one-time cost of mining will take 9 years to get back.
NVIDIA, IBM, and memory vendors are said to offer discounts to Oak Ridge National Laboratories - Buying 27,647 Tesla V100 Accelerators will be a free gift, NVIDIA CEO Huang Renxun said "the more you buy, the more you save money" - The original black-and-yellow mortal heart is here! Lao Huang said this sentence at the previous DGX-2 conference.
In short, mining with supercomputers sounds good, but it takes too long to recover costs. Who knows what the price of Ethereum will be in the next few years?