Will a large number of miners be eliminated after the Filecoin mainnet goes online? Learn how to identify cloud computing power

Written by Calvin Zhang

Under the popularity of Filecoin, mining machine manufacturers piled up machines overnight to compete for the "computing power list". Today, this company won the throne and touted its company's technology; tomorrow, another mining machine manufacturer announced that it had entered the top ten of the ranking list and began to brag about its "cloud computing power subscription opening". The Filecoin browser, which was originally launched for large-scale evaluation and testing before the mainnet went online, suddenly became a marketing tool for major mining machine manufacturers.

The number of vendors selling cloud computing power on the market has increased sharply, and there is a mixed bag of good and bad. Many vendors are using one-sided information and data to deceive investors and engage in false propaganda.

However, the biggest drawback of cloud computing power is its lack of transparency. Many scams have occurred under the banner of "cloud computing power" before. For this reason, identifying the reliability of cloud computing power providers has become an "essential skill" for survival in the cloud computing power market.

Faced with opacity and information asymmetry, cloud computing investors are inevitably caught in various doubts:

• Is it reliable to choose cloud computing power based on single indicators such as "price per T" and "effective computing power"?

• Are the miners on the test network computing power list reliable?

• Are the rate of return and payback period promised by the merchant credible?

Today, let us clear away the fog of industry confusion and talk about what we are actually looking for when we examine a cloud computing provider, and what parameters should we compare when evaluating cloud computing vendors?

Using the pool model to explain Filecoin mining

On Saturday, I was lying in bed and browsing my phone when I saw a discussion in the circle communication group about purchasing IPFS mining machines/cloud computing power.

In fact, we have seen various communities discussing Filecoin mining for a long time. Indeed, due to the relatively complex mechanism of Filecoin mining, the community is also full of misunderstandings about the project. In the final analysis, these are caused by people who do not have a deep understanding of Filecoin. It feels that the industry urgently needs someone to come out and give cloud computing power a good name, so this article was written.

So, at the beginning of today, let’s use a pool model to briefly simulate the process of Filecoin storage mining. If we compare Filecoin mining to filling a pool with water:

• The reservoir corresponds to the storage part (hard disk) behind each Filecoin mining machine;

• The water injection pipeline corresponds to the computing part of each mining machine (CPU/GPU/AMD motherboard);

• The amount of water that has been injected is the effective storage of the mining machine;

• The water injection speed corresponds to the effective packaging speed (GB/H/W);

Note: Pay close attention to the relationship between water injection speed and water injection volume. These two parameters are easily confused. This is why people in the IPFS cloud computing market often confuse the effective packaging speed of mining machines (the growth rate of effective computing power) and the effective computing power of mining machines.

What really determines the revenue of a mining machine is not the size of its effective computing power, but the proportion of the mining machine’s effective computing power to the effective computing power of the entire network. The effective packaging speed (that is, the water injection speed) determines the proportion of its effective computing power to the total network’s computing power.

Another question is why we should buy cloud computing power, because the practice of the test network has already explained the problem. The cluster mining method is much more efficient than the single mining machine mining. At present, cloud computing power mostly adopts the cluster mining method, which is much more efficient than the single mining machine. Many machines are used to form a node to separate the mining process. Generally speaking, a master node, that is, a miner, and N remote workers, that is, workers, will be set up. The master node is generally responsible for the deposit task, and receives the sectors returned by the worker after processing, and then submits it to the chain for verification and becomes valid storage. It also runs the time-space proof and the election proof, and is responsible for the most important block explosion task. The worker is responsible for encrypting and encapsulating the initial sector sent by the master node, and then returning the processed sector to the host point.

Cloud computing mining generally has the advantages of building large-scale mining machine clusters, coordinating the management of power supply, professional bandwidth and mining machine operation and maintenance.

It’s the same 1T, why is the price different?

We investigated the cloud computing products available on the market and compared and counted the prices.

The prices of cloud computing products on the market vary widely. The highest price is nearly 3,000 yuan per terabyte of effective computing power, while the lowest price is 1,106 yuan for 1 terabyte of effective computing power. The difference is nearly double.

Why are there such big price differences between merchants?

Here we have to mention the pricing model of IPFS cloud computing power. Buying cloud computing power is different from buying vegetables. It is an investment behavior. Don’t just look at the price. Ultimately, you still have to look at the return on investment. After all, we invest in buying mining machines, and the ultimate goal is to produce coins, for output, and for returns. So the final evaluation is not the price, but the input-output ratio.

Unfortunately, the main network has not yet been launched, and no one can verify the input-output, so the manufacturers selling computing power are more like selling "commodities". Therefore, the current pricing model of each manufacturer is based on cost plus marketing expenses, plus the expected profit to form the current selling price of each T of cloud computing power.

However, in addition to the unreasonableness of the current cloud computing power pricing model, there are two other "pitfalls":

(1) The logic of FIL mining machine investment is to pursue the highest possible currency production efficiency per unit investment. Under this logic, the FIL output per T will be an important reference indicator for cloud computing power manufacturers. The optimal solution for effective computing power is only a means, not a goal. If you only know the price per T of effective computing power, but do not know the FIL output per T of effective computing power, the price will appear pale and meaningless.

The imbalance between effective computing power and coin production efficiency is a common phenomenon in the Filecoin network. Please keep this "pitfall" in mind.

For example: in the current test network ranking, the effective computing power of Spacetime Cloud & Xi'an Lingdong is 6.29PiB higher than that of Xianhe System, but the block reward within 24 hours is more than that of Spacetime Cloud & Xi'an Lingdong. The reason for this is that the average amount of FIL mined by miners per T of computing power is different.

This also proves once again that a single effective computing power cannot reflect the final amount of coins obtained from the investment. If you want to clearly judge the block-producing ability of a mining machine, you must know the indicator of the number of FILs mined by a single T of effective computing power.

Current testnet effective computing power ranking

Current testnet block reward ranking

(2) Currently, all cloud computing products require users to pay for the effective storage filling process. If you buy 1T of cloud computing power, will you be given 1T of effective computing power on the first day? Or will you be given 1T on the last day? The difference in benefits is huge.

Back to the reservoir model mentioned in the first part, there is no water in the pool at the beginning. That is, the 1T storage space purchased by the user is not filled with effective computing power, and the user's income is greatly discounted. Only when the effective storage of the mining machine reaches 1T can miners enjoy a higher mining rate of return. The previous filling time is equivalent to a "cold start".

Therefore, if a cloud computing vendor only tells you the price but does not tell you the filling time or the effective packaging speed (the filling time can be calculated), then there is a high probability that there is something fishy about its price and it is not transparent.

 Here we introduce a concept: effective packaging speed (GB/h)

Knowing the effective packaging speed of cloud computing power can not only calculate the time required to fill 1T of storage, but also calculate the daily income before filling up.

 The time required to fill 1T storage: 1T full storage time = 1024GB / effective packaging speed (h)
Daily revenue: Daily effective storage = packaging speed (GB/h)\*24h
Theoretical daily mining income = cumulative effective storage / effective storage of the entire network \* (daily coin production - mortgage amount) \* coin price

After the mainnet is launched, since everyone can see their returns, Filecoin cloud computing power will be more like an investment product, and its pricing model will become the return on investment. The price of this product will be calculated according to my expected annualized return. At that time, the price will be comparable to see whether it is relatively expensive or cheap.

Therefore, we have listed the cloud computing power parameter table currently on the market, and you can also compare and judge it yourself.

"Our mining machines are the strongest because we are on the list"

Everyone is familiar with this picture. It has been overused by various mining machine manufacturers or cloud computing providers. It appears in various marketing copy and various soft articles. So why is it so important?

Ask first if, then ask why.

So, is the effective computing power ranking of this test network very important?

The "effective computing power ranking" is meaningful to mining machine manufacturers, as it can help companies understand their own strength and that of their competitors, and facilitate the adjustment of their technical solutions and market strategies. However, this single parameter is meaningless to ordinary users, because users do not know how many machines the mining machine manufacturers used and how much they spent to obtain this effective computing power ranking. There is no way to compare the advantages and disadvantages of using 100 machines to rank first on the test network and using 10 machines to rank tenth on the test network.

A high ranking on the list may mean that the node has a large number of mining machines, but if the block generation rate is not high and a lot of effective storage is used, it may also lead to a high ranking. The brand cannot be equated with the absolute strength of the manufacturer.

There are a few points I would like to remind users about the “Effective Computing Power Ranking”:

(1) The test network ranking manipulation is a marketing behavior that has far more significance than technical testing. The wool comes from the sheep, and the testing costs are ultimately borne by the users. Currently, the Filecoin test network cannot mine "real coins". The FIL rewards mined by various mining nodes seen on the test network browser are just "data" used for testing and experiments. In other words, it is "mining air."

(2) Before the "space race" on the test network begins, the test data on the test network should be mainly for technical testing, and there is no need to deploy machines on a large scale to increase computing power. The main reason is that when the main network is launched, all the effective computing power stored in the test network will be cleared, and everyone will start competing for computing power from the same starting line. There is no need to deploy a large number of machines in the test network stage, because once electronic products and machines are opened, they will be discounted by 1/3, and there will also be performance degradation.

(3) Some mining machine manufacturers run nodes together. I know that several mining machine manufacturers have merged to run a node number to improve their rankings, and use the networking method of larger cluster mining machines to improve each other's stability. This is very worthy of encouragement. First, forming a cluster mining machine with other mines can greatly improve the mining efficiency of the mining machine and maximize the hardware performance. Multiple pairs of computers can be packaged in parallel to increase the effective packaging speed per unit time; secondly, heterogeneous cluster mining machines require multiple mining machines to be matched, but they can provide more stable FIL output and reduce the impact of orphan blocks on the income of subscribing users; thirdly, cluster mining machines formed by multiple manufacturers can keep each other warm, increase their share in the entire network, and combine forces to increase FIL production.

As we have been emphasizing, the standard for evaluating the mining efficiency of a Filecoin mining machine should not be the effective computing power ranking, the effective computing power growth rate, or the price, but the input-output ratio. Mining machine manufacturers can configure high-performance hardware regardless of cost, complete high-speed packaging at the expense of memory, and expand mining machine clusters to improve computing power rankings, but these single factors cannot be used as evaluation criteria for the strength of mining machine manufacturers.

Effective computing power should be measured by combining multiple factors such as cluster, hardware, and algorithm. The ranking of "effective computing power" is more of a market promotion need. If miners want to obtain high returns, they should combine the cost of effective computing power growth per T of mining machines, block rate, service time, and algorithm technology to get the right answer.

After the mainnet is launched, most mining machine manufacturers will be gradually eliminated

This is the focus and climax of this article.

(1) Let me first state my opinion: According to the official Filecoin mining rules, if the growth rate of a mining machine manufacturer's effective computing power is lower than the average growth rate of the effective computing power of the entire network, then the FIL income of the mining machine manufacturer will continue to decrease. In addition, the official requirement for FIL pledge will cause most of the current long-tail mining machine manufacturers to gradually lose their competitiveness and may even fail to mine FIL.

(2) Imagine a rapidly expanding balloon. This is the state when all mining farms are running at full capacity after the mainnet is launched, and the "effective storage" of the entire network is growing rapidly. The speed at which the balloon expands is the growth rate of the effective computing power of the entire network. If you are lower than this average, your "effective computing power" will continue to decline as a percentage of the entire network, and the FIL you earn daily will also continue to decline.

(3) In order to explain this issue clearly, I established a mathematical model to simulate the difference in FIL revenue caused by the difference in packaging speed after the main network is launched.

There are four mining farms, namely ABCD. They started mining together after the mainnet was launched. The initial effective computing power is 0. The data is as follows:

After running for 6 days (1d=24h), the distribution of the total effective computing power of the entire network is shown in the figure:

We first exclude the impact of the cyclical decrease in official FIL release (even if it is taken into account, it can only make the conclusion more valid), and the situation of pledge depreciation, simplify the calculation, and the profit change curve of the four mining farms ABCD.

After the mainnet is launched, the investment attributes of mining machines will be highlighted, and the market will become more sensitive to the return on investment, which means all-round requirements for products, technologies, etc. All of this plus the Filecoin mining logic will further amplify the "Matthew effect", and a large number of mining machine manufacturers will quickly lose their market competitiveness.

Investment is the realization of cognition

The process of investing in Filecoin mining machines is essentially a process that requires continuous improvement of one's own knowledge and then choosing a reliable mining service provider. This is very different from Bitcoin and Ethereum mining in the field of Filecoin mining investment. Whether it is Bitcoin mining machines or Ethereum mining machines, the measurement indicators are relatively simple and direct: the price per unit of computing power and the power consumption per unit of computing power. Filecoin mining does not have such indicators, but through this article, there should be a clear direction and standard for judgment.

For Filecoin to form an ecosystem and a mature decentralized storage market, it requires the efforts of every participant. Everyone who participates in the Filecoin mining wave needs to continuously improve their understanding of the industry, independent thinking and decision-making.

When contacting mining machine manufacturers, you need to listen carefully to their marketing content, grasp the essence of investment - risk control and rate of return, do not be blinded by false reputations, clear away the fog of merchants, let every investment play its greatest value, do not be a leek that can be slaughtered at will, and help the industry continue to correct itself.