Why is the M1 Chip so Fast?

Apple has long been a pioneer in the field of computing and computing technology. With every product they have launched in the market, it is easy to see why. The brand has always had foresight into the future with design ideas to facilitate that future, and more often than not, they are on point. They always bring something new to the table. 

This year they did the same as they unveiled their all-new M1 Chip, which commemorated their shift from Intel chipsets to their very own. With reviews all over the web swearing by its speed, the most debated question now is, “how is it so fast?”

SoC Design

It is the world’s first 5nm process technology, designed to be both powerful and efficient at the same time. This is because it’s not just a CPU. It is a SOC, Meaning it has the CPU, GPU, Memory, and much more. The SOC has all the components baked onto a single chip allowing for easy communication between the components. 

Our brain has billions of brain cells that help us think or remember things, around a hundred billion of them. Similarly, chipsets also have transistors that store binary data. The M1 has the largest number of transistors ever on a single chip. Sixteen billion, to be exact. 

Architecture

The M1 Chip is built based on an ARM architecture or Advanced RISQ Machine Architecture developed specifically for mobile computing. Therefore, it runs on an ARM instruction set and produces less heat as opposed to Intel’s or AMD’s architecture that uses an x86 instruction set. What is a bit lesser known is that Apple has a general-purpose CPU in the M1 Chip called Firestorm that is already fast by itself.

Processing

In order to make this even faster, there are two ways one can go: sequential Processing and Parallel Processing. Apple chose to go with the latter option. By increasing the Parallel processing speed, multiple tasks can be performed at once or parallel to each other. 

This is also known as Threading Code. Although efficient, it is very difficult to write a threaded code by most developers. So Apple devised a system with the M1’s Firestorm CPU to write this Threaded Code using a technique commonly known as Out-of-Order Execution. 

Decoders

This process further becomes faster and effective because the M1 performs the task of filling up the buffer of micro-operations faster using decoders’ help than what Intel is capable of. How? Intel’s most powerful chips have only four decoders, and with its x86 instruction set we mentioned before, the instruction set can range anywhere between one and 15 bytes long. This method has made it very difficult to split these bytes into instructions due to varying lengths. 

The M1, on the other hand, has eight decoders with an instruction buffer that is 3 times larger than usual, allowing it to fill the buffer faster while having a larger space. Besides that, the ARM instruction set uses a 4-byte long instruction set that is a fixed size. This reduces the amount of work for the decoders to split these bytes, making it a relevantly easy task, and therefore, resulting in higher efficiency and speed.

Unified Memory

Generally, when it comes to the average system, they have integrated CPUs and GPUs. The way each of these components assimilates data is different as they are separate entities. In order to open a file, the CPU needs to gather information regarding the data file and the application required to open or access that particular file.

It then takes all this data and loads it into the system’s memory but has to access all this data from various memory pools. It’s like having to take data from one storage room, carry it to another storage room before you access it, and take it out.

Besides this, as they are separate entities. CPUs naturally tend to assimilate data quickly but in smaller volumes as opposed to GPUs that prefer data at a slower rate but in larger volumes. This results in the GPU becoming drastically slower as the CPU starves it of data, therefore, resulting in slower performance due to the variances in data assimilation combined with the shared memory protocols of separate data memory pools .

The M1, though, has something called a unified memory architecture. This allows the Chip to have access to data from one single pool rather than having to copy the same between multiple pools of data, resulting in low latency. Simply put, the various parts of the processor can access all the information from one common Storage room. This approach cuts out the need to shuttle information from one data pool to another, making processes faster than ever before. 

“Core” Strategy

The Chip has an eight-core CPU that gets split into four High-Performance Cores and four High-Efficiency Cores. These High-efficiency cores take care of lighter day-to-day tasks while using just one-tenth of the power allowing the High-performance cores to handle the heavier workloads even while multi-threading. This results in twice the computing power while consuming twenty-five percent of the energy used by an average computer.  

The GPU also has eight cores that Apple claims can execute 25000 threads at any given point in time. This allows for outstanding graphic performance, with the GPU delivering twice the performance at just thirty-three percent of the power used by the latest laptops. 

The other guys

Aside from this, it has an Advanced Image Signal Processor. This ISP is a specialized component that increases the image processing capabilities of the M1, which is why the image processing capabilities of the M1 are much faster than the general industry standard processors. Another Specialised component is the Digital Video Decoder and Encoder. It helps in the conversion of highly demanding video files to various other formats. 

The M1 also has a Digital signal processor that assists in tasks that are more mathematically oriented. It also includes tasks such as decompressing music files. The 16 core Neural engine is another essential component that makes the Chip much faster. It helps with AI tasks, voice recognition, and camera processing. The machine learning accelerators further increase its performance in leaps and bounds. How fast? A whopping 11 Trillion calculations per second. 

End

With this much thought put into the whole design of Apple’s new Chip, it’s not surprising to see why the M1 Chip is so fast in performance. You can also get your hands on this kind of performance with the new Apple M1 Chip. Hop over to the Poorvika website to check out our range of M1 Macbooks and have your device delivered right to your doorstep with our 2 Hours Fast Delivery.