Computing power does not equal income. Li Bai teaches you how to improve your "luck value" and earn more FIL

(The following text is excerpted from the video live broadcast of 6Block core engineer "Coin Circle Li Bai". The corresponding video can be found at the end of the article)

Recently, many members of the community have talked about the issue of block explosion. In response to the issues raised by the community, our Pomegranate Mining Pool has conducted in-depth testing and research on block explosion in the past two weeks. (Without exploiting any bugs)

It is generally believed in the market that block explosion is a probabilistic problem, and its probability will fluctuate around the proportion of computing power. But this is not the case. There is actually a lot of room for optimization. This article will cover the optimization space for block explosion and some common problems in block explosion.

When it comes to Filecoin mining, people generally pay more attention to the following aspects:

• Having machines does not necessarily mean having computing power

• Having computing power does not necessarily mean having a block explosion TB-> FIL

• Having a block does not necessarily mean unlocking

Today we will start with the last two points. I have talked about the first point a lot before, so I will not say more. So what is this block burst? In fact, it is very simple. The duration of the block burst is 25 seconds. If the calculation assigned by the system can be completed within 25 seconds, the block burst can be achieved.

There are two algorithms in Filecoin, and the block generation algorithm belongs to the PoSt algorithm. In simple terms, there are the following steps:

1. It is necessary to ensure that all blocks at the previous height are obtained

2. Calculate the random number. After getting the random number, you can determine whether you have won the prize, and get the winning sector number and sector data number

3. Reading the sector data will take some time.

4. Calculate the zero-knowledge proof to get the result

5. Build a block with the obtained results and broadcast the block

The official default time for obtaining all blocks of the previous height is 6 seconds, which means that all blocks are synchronized within 6 seconds by default. Therefore, the mining step will begin after 6 seconds, and the calculation of random numbers will begin. This requires that the network latency of our nodes should be low enough, and that we should synchronize with multiple nodes, and sort and filter the synchronized nodes. Therefore, the limitation of obtaining all blocks of the previous height is mainly due to the network.

When you have the right to generate a new block, you need to read the sector data. More importantly, you need a low-latency storage that allows the miner to get data from the sector in the shortest possible time. The Pomegranate Mining Pool separates storage and computing, and sets the Worker to be stateless, so that if there is a problem, it can be restarted or repaired at any time without affecting the entire system. At the same time, our storage is always in a low-load state. In this state, the storage response speed will be very fast. When the storage and miner are connected, proofs can be generated quickly. In addition, Lotus and miner are best placed on a local area network or on a computer. Lotus needs to connect to a large number of other Lotus nodes to ensure that the broadcast is fast enough. The above is the architecture of our Pomegranate Mining Pool.

Here are some common questions:

First, computing and storage are not separated. Under such an architecture, the data extraction speed is likely to be affected and difficult to optimize, and it is difficult to achieve a large number of blocks.

The second is the storage load caused by redundancy and backup. If you blindly do redundancy and backup, it will affect the read and write performance and thus affect the block explosion.

Another problem is that the reading and writing processes will compete for resources. The best solution is to separate reading and writing.

Then there is the issue of multiple miners. This is a very creative idea, the main purpose is to concentrate the computing power under one account, which is easier to promote. However, multiple miners will affect the block explosion, and the reading data may be affected to a certain extent.

Finally, let's talk about calculating zero-knowledge proofs. It also takes a certain amount of time, about a few seconds, but you can save some time by equipping the miner with a graphics card. The calculation of zero-knowledge proofs can also be optimized. In the bellman library, optimizing it can save time.

These are probably the problems and solutions for block explosion. Finally, let’s show the results of our two-week testing efforts:

As a result of the optimization, the time for block generation has been reduced to 1.7 seconds.

In general, if you want to reduce the time of block explosion, you need a low-latency network, a responsive storage system, and a zero-knowledge proof computing system that can quickly calculate Wining PoSt.

Okay, that’s all for today’s sharing, thank you everyone!