A tech expert cracked the M1 MacBook Air for mining, and Apple was beaten badly

Graphics cards are in short supply, and NVIDIA has restricted mining functions, so miners have to look for other mining equipment.

Recently, software developer Yifan Gu found a way to implement Ethereum mining in the M1 MacBook Air and open sourced it on GitHub:

https://github.com/gyf304/ethminer-m1

Mining on M1

What is the mining performance like on M1 Mac?

The first thing I did was to run the stock ethminer and the following error occurred:

Apparently, Apple GPUs are not whitelisted in ethminer. This should be easy to fix though. The relevant lines are in libethash-cl/CLMiner.cpp, where I add the Apple GPU to the whitelist, “pretending” it’s an Intel GPU.

Then boost would not compile because it was trying to compile with the -fcoalesce-templates parameter which does not exist in the latest clang version. So I had to update boost to the latest version and fix the related asio code because ethminer is using a deprecated asio api.

I also needed to upgrade OpenSSL to the latest version to make it support darwin+arm64.

After everything is compiled, the result is as follows:

According to Gu Yifan's experiment, the mining efficiency of M1 MBA is average, only 2MH/s, and the power consumption is about 17~20 watts.

Compared with NVIDIA's CMP dedicated mining core, the daily income is about US$0.14, which is less than RMB 1.

These numbers pale in comparison to the performance offered by discrete graphics cards. Especially since they are probably optimized for mining scenarios, even Nvidia’s weakest cryptocurrency mining processor has a hashrate of 26 MH/s.

As you can see in the chart above, the M1 also can’t match Ethereum’s gaming GPU hashrate. Of course, this difference comes down to Nvidia’s graphics having a higher power budget, and it’s hard to tell what Apple’s GPU architecture will look like if the eight-core design is scaled to larger devices with more computing power. There are reports that Apple is working on a 128-core GPU to replace AMD’s graphics cards, so we may not have to wait too long to find out.

Gu Yifan is not the first person to try mining with the M1 CPU. In December last year, XMRig developers used M1 Macs to mine Monero.

XMRig used a new Mac mini with a new M1 ARM processor to mine at a peak of 3.725 kh/s for an entire night, and without paying for electricity, the profit was very small. Compared with professional mining machines, this solution also seems to be only suitable for novices.

In the past few years, as the price of Bitcoin has skyrocketed, the market demand for mining chips has increased, and many chip giants have entered the mining chip market to make money in the virtual currency field.

At the end of 2018, Intel developed a new mining chip and claimed that it would go a step further than the standard application-specific integrated circuit (ASIC) mining technology. This move sparked heated discussions in the industry and became a landmark event for the chip giant to enter the mining chip field. It also promoted the further popularity of the virtual currency market.

Intel's entry into the market has sparked heated discussions in the industry. In fact, chip giants such as Nvidia, TSMC, ASUS, and AMD have already been players in the field.

In December 2017, ASUS launched its first professional mining motherboard, the B250 Mining Expert. The mining motherboard supports 19 cards at the same time and features ultra-durability and stability. ASUS has also developed a mining BIOS mode for it, with a large number of CSM-rich configurations that can be customized.

In July 2018, AMD claimed in an interview that APU processors specifically for mining would be available for sale in the near future. Then AMD's 5600XT and 5700XT became one of the last popular mining graphics cards on the market.

On February 18 this year, Nvidia announced that it would release a series of chip processors specifically for mining cryptocurrencies. Nvidia's new chip series is named CMP (Cryptocurrency Mining Processor), which is dedicated to Ethereum mining. Nvidia said that the first batch of CMP graphics cards will be available in March this year, but the specific pricing has not yet been announced.

On February 24, Nvidia released its financial report for the latest fiscal year (ending at the end of January this year), and data showed that Nvidia's revenue for the fiscal year was $16.7 billion, up 53% year-on-year, with gaming GPU sales growing in the fourth quarter exceeding market expectations. Nvidia CFO Colette Kress estimated that $100 million to $300 million of sales in the fourth quarter came from "miners."

The release of the M1 chip has attracted much attention. In addition to mining, what is the office experience of a Mac equipped with M1?

Apple officially released the M1 processor in November 2020. By the end of that year, Apple had launched a total of three products equipped with M1 chips, namely the MacBook Air, 13-inch MacBook Pro and Mac Mini. These devices have been widely praised in terms of benchmark performance and battery life.

The owner of the public account "MacTalk", senior developer Mr. Chi, gave an objective evaluation result after evaluation (the test model is MacBook Pro 13.3-inch M1 chip):

1. Super long battery life, 20 hours is not a boast. I went on a business trip on Tuesday and used it for three days without charging. Now it still has 45%. This finally allows me to take my Mac on a business trip without worrying about the battery life, which is very reassuring.

2. This winter, M1 is very cold. When I use this computer normally, M1 is always cold and quiet, like a proud girl. No matter what you do, she will look at you coldly, cruelly and silently.

3. Previously, when Mac Book was connected to an external monitor, the screen would always shake. My 16-inch external monitor would not light up after the screen was off for a long time. M1 does not have these problems at all. When connected, it lights up, and the previous App will automatically move to the external monitor. When disconnected, the App in the external monitor will smoothly move to the M1 monitor without any flickering. Why does it flicker when connected to the monitor? Maybe it's because it flickered for too long.

4. Image and video processing is really fast, amazingly fast. It has powerful machine learning capabilities, which are very friendly to professional users.

5. The screen never turns off, it lights up as soon as you open the cover, just like iPhone and iPad. This feeling is very good. Every time you turn on the computer, there is no need to wait for a moment, and the fingerprint recognition directly enters the computer to process the work, and the screen seems to have never turned off.

6. Daily App compatibility is very good. If it is not a native ARM architecture App, M1 will translate and start it through Rosetta 2. It will be a little slow for the first time, but you won’t feel it later. Apple’s own native Apps, such as XCode, Safari, etc., should have been optimized, and opening and using them are as fast as lightning.

However, because this laptop has only 16 GB of memory, if you open enough apps, such as dozens of apps. I have always used 64 GB of memory, and I basically never close apps. If I open too many apps, the memory will be exhausted, which will cause the system to freeze when switching. This problem is also easy to solve, such as adding more memory. Unfortunately, the top configuration of the M1 13-inch is 16 GB.

7. For the first time, you can use iPhone and iPad apps directly on macOS. This is a new starting point. A trend of integration is evident. Considering the large number of apps on the iOS App Store, it is still very exciting. However, I have used apps such as WeChat Reading, Geek Time, and Douban. Each of them has its own problems. There are many areas that need to be iterated in terms of fonts, interactions, layout, and architecture. Both the technical architecture of macOS and iOS App need to be continuously improved. The current user experience is not good.

8. M1 is not so friendly to backend developers. For example, if you want to use Homebrew, you need to install two sets:

1. arch -x86_64 /bin/bash -c

2. $(curl -fsSL

https://raw.githubusercontent.com/Homebrew/install/master/install.sh )

This is for x86, and you need a native set. When installing system services, some only support x86, and some can already support Arm. You need to try before you decide which method to use for installation.

For example, Docker and virtual machines will have various problems running on M1, and Multipass will not work. However, migration takes time, and everything will be fine. At least the commonly used IDEs have no problems using them, or the problems can be solved.