According to the recent data recorded by GeekBench 5,M1 processor在單核效能表現贏過Intel採10核心設計、運作時脈為2.7GHz的Core i9-10910處理器，而在多核效能表現則比採12核心設計的Xeon E5-2697處理器還高，同時也直接扳倒用於現有16吋MacBook Pro的Core i9-9880HK處理器，但事實上真是如此嗎？
Account data does not necessarily reflect the actual user experience
Regardless of whether the GeekBench 5 data was deliberately fabricated, as no official test data has been released yet, we must first dispel a common misconception about performance comparisons: you can't infer overall performance from a single result. This is especially true when comparing the M1 processor's operating mode to Intel's x86-based processors, which differ in instruction set execution modes, not to mention software compatibility and design application differences.
According to the details released by Apple, the M1 processor built on the Arm architecture is confirmed to use 160 billion transistors, configured with 4 large-core performance CPUs and 4 small-core energy-saving CPUs. In addition, it also integrates a 16-core NPU design similar to the A-series processors and is equipped with an 8-core GPU, corresponding to a total of 2.6TFLOP display performance, and can also cooperate with the NPU to accelerate the execution efficiency of artificial intelligence computing.
Compared to Intel's latest laptop processors, Apple says the M1 processor can deliver approximately 10x performance at 2W of power, using only about a quarter of the power for the same performance, and more than 3x performance per watt. The M1 processor's GPU can deliver more than 10x display performance at 2W of power, using only about a third of the power for the same performance. This means that compared to existing Intel processor designs, the M1 processor will be able to achieve the same performance with lower power consumption and perform higher performance while operating at the same power.
At the same time, Apple also clearly stated that the comparison benchmark is to compare the new MacBook Air equipped with an M1 processor with the existing MacBook Air equipped with an Intel Core i1.2 processor with a running clock of up to 4GHz and a quad-core design. In addition, it also includes the new 7-inch MacBook Pro with 16B memory and also equipped with an M1 processor, for comparison with existing laptop products on the market.
However, Apple did not disclose the operating clock of the M1 processor, but clearly stated that the current memory can only support a maximum of 16GB, and it only supports a maximum of two Thunderbolt and USB 2 ports. Although it is backward compatible with Thunderbolt 4.0 functions, it only supports a single 3K resolution screen display specification in the display output part, but does not support lowering the display resolution to correspond to the two-screen connection output mode. This means that after the laptop's own screen and the Mac Mini's built-in HDMI occupy the output resources, it will not be able to correspond to the second screen connection output, which means it cannot correspond to the eGPU connection usage mode.
Does it really beat the Core i9-10910 processor?
Since the M1 processor adopts a SoC chip design, the memory naturally cannot be expanded. However, existing Intel processor application models also solder the memory modules to the motherboard, so there is no difference in actual usage. The only difference is that the design of the M1 processor allows the motherboard to take up less space, so a larger battery can be placed, or the new Mac models can be made thinner and lighter.
As for the performance crushing of the Core i9-10910 processor, I'm actually more inclined to consider it a "win" based on the data alone. After all, actual performance and user experience are more than just the numbers on paper. Furthermore, the Intel processor used for comparison is actually a downclocked, low-voltage version designed for laptops. So while it may appear to outperform Intel's high-end processors on paper, the actual performance and user experience may differ.
However, since the NPU component application is inherited from the previous A-series processor design, a lot of the computing performance of the M1 processor is generated through the collaborative execution of the NPU and GPU acceleration. In addition, the processor itself is highly integrated with the new version of macOS 11 Big Sur, so it can also give play to the benefits of software and hardware integration, thereby allowing the processor execution efficiency to have higher room for development.
Although Intel processors later added DL Boost design, allowing the processor to execute instructions in artificial intelligence computing mode, it still requires a high degree of software integration. Therefore, even applications in the Windows operating system still need to use the corresponding DL Boost API resources to achieve actual execution benefits, not to mention the situation in the macOS operating system environment where this processor function design is rarely used.
Therefore, in reality, the M1 processor's performance crushes the Core i9-10910 processor because it outperforms the down-clocked, low-voltage version used in laptops. At the same time, the operating clock should also adopt a higher-frequency design, combined with the NPU operating mode, so it can have higher performance in single-core and multi-core computing performance.
Is Apple lying?
In reality, it's not true that Apple is deceiving anyone. After all, Apple didn't directly compare the performance of its Core i9 processors to Intel's, but rather to models using lower-voltage Core i7 processors. Even the Core i15.6 processors previously used in 16-inch and 9-inch models were actually downclocked and used at lower voltages to accommodate the thinness of their designs.
The M13.3 processor, featured in the new MacBook Air, Mac Mini, and the new 1-inch MacBook Pro, is primarily targeted at entry-level users across each product line. While the MacBook Air will only feature the M1 processor, the Mac Mini and 13.3-inch MacBook Pro will continue to offer higher-performance Intel processors with more ports, indicating that the M1 processor-based versions are primarily targeted at entry-level models.
In addition, among the three models launched this time, only the new Mac Mini and the 13.3-inch MacBook Pro are equipped with active cooling fans. It can be speculated that this is based on the design of the M1 processor used in the new MacBook Air. By increasing the frequency and other methods to improve computing performance, the new Mac Mini and 13.3-inch MacBook Pro can have higher performance and meet the performance positioning of different products.
M1 is just one of the Apple Silicon processors
Apple's statement indicates that the M1 is just one of the Apple Silicon processors, which means that there will be more series of products in the future. In addition, the author has further confirmed with Arm that the Cortex-A instruction set under the Arm architecture can already support 7-bit memory address space (i.e. 4040 Byte size), and from version v8A onwards it can correspond to a 64-bit memory address space (i.e. 264 Byte size), which means that Apple Silicon processors will indeed be able to cope with larger memory usage capacity in the future. Therefore, it is possible that in the future, it will provide higher processor performance for the 16-inch MacBook Pro, or iMac, Mac Pro and other models with higher performance requirements, while also being able to handle more complex computing content through larger memory capacity.
At present, considering factors such as heat dissipation and battery capacity, the M1 processor can only balance the space occupied by components such as the CPU, GPU, NPU, and memory within the relatively narrow SoC chip space design. However, in future processor product designs, it is indeed possible to further increase memory capacity according to demand.
As for the performance of Arm architecture processors, it has actually continued to grow significantly in recent years. In particular, Arm itself has proposed the Cortex-A78C CPU design for use in large-size devices such as laptops. Currently, many manufacturers have also built server-grade processors based on the Arm architecture.
Furthermore, Arm-based processors have already progressed to the 5nm process, while other x86-based processors remain at the 7nm and 10nm process stages. Given the number of transistors that can be accommodated within the same size and the efficiency of execution driven by the same power consumption, the Arm architecture surpasses the x86 architecture in performance and power consumption, with the exception of relatively limited software compatibility on PCs and servers.
Apple's plans to improve software compatibility and the amount of software available through its Universal 2, Rosetta 2, and Catalyst projects may resolve the challenges Microsoft has faced in developing Windows on Arm, thereby expanding the popularity of Mac models designed with Arm architecture. Through their lightness, long battery life, and high-performance, they could rewrite the long-standing x86-dominated PC market.
Related links: ▶︎
▶ ︎ The first Apple Silicon processor for Mac will be called M1
