As Bitcoin is about to usher in its next production halving, its network is going through multiple transitions. In addition to the major changes in the mining economy due to reward adjustments, Bitmain ’s Antminer S17 mining machine is also replacing the long-standing S9 series mining machine and becoming the network-leading mining hardware.
It is reported that Bitmain was the Antminer S9 released in 2016. Since then, it has quickly become the most popular SHA-256 mining machine in the market. After several years of development, S9 still occupies a large proportion of the market.
Due to the lack of public data on the types of mining hardware used by various miners, it is difficult to measure the speed of this transition. However, one signal source does reveal the changing trend of mining hardware: the nonce random number distribution of the network. The arrangement of these arbitrary numbers (miners include them in the hash of each block) suggests how the usage of mining hardware has changed over the years.
- Carbon emissions are comparable to first-tier cities in Europe and America, and bitcoin is produced as a “black hole” in energy consumption.
- Bitcoin mining difficulty has risen sharply, or the price of coins has picked up
- Counting mining 2019 | Difficulty of mining increased by 97.67% throughout the year. Which mining pool is on the right pace?
- About bitcoin mining, things you don’t know
- ChainNode live room, starting from Avalon's strongest mining machine A1166, is this a mining machine or a printing machine?
- Explore the Stratum V2 (Stratum V2), how to achieve the decentralization of Bitcoin mining?
In the last report on the state of the network, we studied how to use the random number distribution pattern to discover the rise of ASICs. In this report, we will further explore the particularity of the bitcoin nonce random number distribution and the source behind the distribution stripes to investigate the latest changes in mining hardware. Then, we will break down these data by mining pool, so that we can get a deeper understanding of the hardware used by a specific mining pool.
Learn about the process of Bitcoin mining through golden marbles
Mining is a key part of the Bitcoin security model. It can be said that it is the most important improvement to previous attempts to create digital currencies. Although the meaning of mining is quite complex, the concepts behind it are relatively easy to understand.
From a miner's perspective, mining a block is similar to repeatedly selecting marbles from a bag without replacement. It is important to know that the number of marbles in the bag is very large, and the proportion of blue marbles is very large, while the proportion of golden marbles is very small. When the miner takes the golden marbles out of the bag, they receive a reward.
To explain in more technical terms: Bitcoin miners are racing to find a golden random number. After the random number is added to the proposed block header, it will be hashed below a certain value determined by the network difficulty parameter . The miner searches for this nonce or any number that can only be used once by guessing the value and checking whether the resulting hash value is below a certain threshold. The first one to find this value for a valid block and broadcast it to the network has the right to select and sort the transactions in the block, which is the necessary step to finally make these transactions valid.
In return, the miners can get a block reward and charge a processing fee from any transactions contained in the block, both of which are obtained through special coinbase transactions. Assuming that the attributes of the SHA-256 hash function remain unchanged, the distribution of golden nonces in any given block is random. Unless calculated by brute force, golden random numbers cannot be found.
Since the reference to the coinbase transaction is included in the block header, each mining entity samples from a different distribution. In other words, each entity takes marbles from different bags, and the bags contain the same number of marbles, and it is expected that the ratio of blue and golden marbles is the same.
The proportion of golden marbles is determined by the network difficulty parameter (automatically adjusted by the network) and is fixed within the relevant period. Today, due to higher block difficulty and random variance, there is usually no golden nonce random number for a specific block header. In other words, there are no golden marbles in some bags.
Miners who use up the random number space of the proposed block will usually increase the time stamp of the block to generate a new set of random numbers. In other words, when the miners have taken the marbles, they will grab a new bag full of marbles. If the timestamp has reached the point where further adjustments are made invalid, the miner must adjust the transaction set contained in the block. Similarly, if a miner runs out of bags in a room, they need to grab more from another room, which is time-intensive.
In order to increase the probability of finding golden marbles within a fixed time, miners can parallelize their calculations, which is similar to grabbing a marble at a time instead of grabbing one at a time. By using hardware suitable for the task (especially the GPU and a dedicated chip called ASIC), you can find nonce random numbers in parallel. The parallelization efficiency of ASIC is higher than any other method.
In another form of parallel computing, several miners coordinate their nonce discovery and agree to split any mining returns, and the group of miners acting in this way is called a mining pool, and its operators usually A certain fee will be charged, and individual mining unions accept this fee to reduce the volatility of their income.
Bitcoin's Nonce random number distribution
The difficulty parameter of Bitcoin is adjusted every two weeks, so that if the amount of calculation performed on the network remains unchanged, a new block will be generated every 10 minutes on average. This function ensures that even if the computing power may change significantly, the network will continue to operate. In a fully competitive mining market dominated by parallel computing miners, we can expect that the distribution of golden nonce random numbers should appear to be evenly distributed over time. But surprisingly, this is not the case.
The non-random distribution near the left side of the graph can be attributed to mining through iterative test values starting at 0. If a miner passes the CPU and does not perform parallel mining as an individual, so it is unlikely to conflict with other members of the pool, then this strategy is as effective as other strategies because the nonce distribution of each new block is independent. The disappearance of this model coincides with the introduction of GPU miners, because GPU miners parallelize computing.
Near the right side of the figure, there is a striped area where the nonce random numbers are few. To our knowledge, this anomaly was first discovered by Twitter user @ 100TrillionUSD in January 2019. The area is marked below.
Shortly thereafter, BitMEX's research paper explored this strange pattern, speculating that this anomaly was caused by the controversial mining optimization technology AsicBoost.
There are two variants of AsicBoost: 1. Recessive AsicBoost (which cannot be definitely observed on the chain), 2. Public AsicBoost (which can be clearly observed on the chain). The BitMEX research team discussed these two variants, but was particularly interested in the effect of hidden AsicBoost. With the activation of SegWit in August 2017, non-empty blocks are almost impossible to use hidden AsicBoost. Of course, researchers cannot confirm their guesses.
In the 23rd report in October 2019, we delved into the nonce random number distribution of Bitcoin and pointed out the striped pattern. Since then, the striped pattern has gradually disappeared, and the random numbers of recently mined blocks seem to be distributed more randomly.
However, the abnormality in the random number distribution does not seem to be directly related to AsicBoost. The hidden AsicBoost has become unavailable in 2017, and the first firmware update to support the public AsicBoost was released in October 2018, but the nonce distribution stripes between these two dates are clearly visible. In addition, although the utilization rate of disclosed AsicBoost is still very high, no matter whether there is obvious AsicBoost, this abnormal pattern is no longer visible in the newly produced blocks.
Another possibility is that the pattern in the distribution of nonce random numbers may be caused by the way the Antminer S7 and S9 miners in Bitmain sample the nonce random numbers. This image may be caused by the side effects of optimization, and it is ultimately harmless to miners and the network.
When observing all nonce random values on the network, the stripe pattern first became clear at the end of 2015, which is consistent with the time when Bitmain released S7 in late August and completed the order in late September.
Antminer S9 was released in late May 2016, and the first batch of buyers received orders in mid-June of that year. Soon after, S9 replaced S7 as the dominant miner of the Bitcoin network at that time, the stripes became narrower and narrower.
The recent collapse of this model coincides with the transition from S9 to Antminer S17 (the main miner on the network). Although S17 was released in April 2019, due to the mining economy, until recently, miners are still using S9.
By layering the miner data set of each block, we can view the nonce distribution at a finer granularity.
We know that block miners are usually identified by marks in the coinbase data field of the block, and these signs are provided voluntarily. There may be forgeries. Miners do not need to leave information, they can choose to leave The label of another mining pool replaces its own. Therefore, in some cases, there are even incentives for these misleading behaviors, so we should recognize the shortcomings of this method. However, this technology has now become the industry standard. Although many miners choose not to leave an identification code, we do not believe that large-scale forgery is happening.
Once we have classified the blocks by miners, we can merge this information into our Bitcoin nonce random number distribution graph.
We can also view the nonce random number distribution of each mining pool. Even in this case, the abnormal pattern is still visible. Please see the chart below, which shows the blocks mined by Antpool and BTC.com (both of which are owned by Bitmain) and ViaBTC.
In the nonce distribution of Bitmain's affiliated mining pools, the stripe pattern needs to be more clear. This shows that during the relevant period, the proportion of S7 and S9 mining machines in these mining pools is relatively high, which is in line with expectations given the association between the mining pool and the manufacturers of these mining machines.
In 2015, the proportion of blocks mined by unknown entities declined sharply. This is the result of the war on block expansion. During this period, many previously anonymous miners began to identify themselves on the chain to express support or opposition to the zone. Increase in block size. Today, with hashing power, we can identify the vast majority of miners. In the block nonce random numbers mined by unknown miners, striped patterns are faintly visible, as is their gradual disappearance.
Since its release in 2016, Antminer S9 has been the most used mining machine model on the Bitcoin network. Although Bitmain released S17 last year, S9 has maintained economic operation for a period of time, but in view of the continuous improvement of the computing power of the entire network and the changing market conditions, this type of mining machine is being gradually phased out.
While the miners transitioned from S9 to S17, the striped pattern that used to define the characteristics of bitcoin's random number distribution has disappeared. These mysterious stripes appear in a random-looking space, the source of which has been the subject of speculation. The time point of the visibility of the stripes proves the theory that these lines are the products of mining hardware, especially the previous S9 and S7.
Nonce data allows us to measure the scale and speed of this transformation in an otherwise impossible way using only public information. By using the traces left by S9 in nonce sampling, we can estimate the proportion of these miners on the network. The separation of these data by mining pool provides unique information about the efficiency of the mining industry, which we will introduce in future reports.
One week insight into network data
In the past week, bitcoin (BTC) mining has shown signs of a healthy recovery, and network computing power has increased by approximately 7.3%. Inefficient mining machines may have begun to surrender and have been replaced by more efficient mining machines, which is beneficial to the long-term health of the network. For more information about the economics of miners and the impact of the recent difficulty reduction, please refer to the 44th report on the state of the network.
In addition, the number of BTC active addresses also showed positive signs of recovery this week, with data increasing by 6.3%. In contrast, the active address of Ethereum (ETH) has developed in the opposite direction, with a decrease of 13.4% year-on-year last week.
Network data highlights
Since the bitcoin market experienced a plunge on March 12, the number of addresses holding a small amount of BTC has been increasing.
Over the past 90 days, the number of addresses holding between 1 billion to 1 billion of the total supply of BTC has increased by approximately 6%. Similarly, the number of addresses between one hundredth and one millionth of the total supply increased by about 4%.
Starting around March 12, the growth rate of both has increased significantly. As new users begin to acquire a relatively small amount of BTC, this may indicate that the adoption rate is growing.
Data from Coin Metrics
Within the scope of our research, the amount of ETH held by the exchange has increased, while the amount of BTC held by the exchange has decreased.
Among them, the number of ETH increased by about 5%, while the number of BTC decreased by about 3%. The decline of BTC is mainly related to the rapid decrease in the supply held by BitMEX.
The chart below shows the 30-day change in the total amount of ETH and BTC held by the following exchanges (Bitfinex, Binance, Bitstamp, Bittrex, Gemini, Huobi, Kraken, BitMEX and Poloniex).
Of all the exchanges we studied, Bitfinex's ETH supply grew the most. In the past 30 days, the amount of ETH held by Bitfinex has increased by about 17%, while the growth of other exchanges has not exceeded 10%.
Market data analysis
Over the past week, the cryptocurrency market began to climb and began to make up for the losses suffered on March 12. Our previous research found that the sell-off on March 12 was driven by short-term holders and was affected by the BitMEX liquidation spiral. There is a broad consensus that this sell-off reflects the sell-off seen in traditional markets, and the increase in Coinbase user activity indicates that retail investor sentiment does not appear to be affected.
The debate about Bitcoin being a "risk asset" and "safe-haven asset" continues. Some market participants have observed that in the past month, Bitcoin's "safe haven" narrative has been disrupted, which has had a negative impact on institutional investors' willingness to enter the field.
And by studying Bitcoin and gold, some evidence can be provided, which shows that this narrative is reasonable and may be stronger than ever.
According to the data, the correlation between Bitcoin and gold has reached a record high in the past 30 days.