Industry News

What Makes the M1 Chip Faster? Apple UMA Memory Explained!


Apple (AAPL.US) is rethinking how the internal components of laptops should exist and work. By using the M1 chip in the new Macs, Apple computers have a new "Unified Memory Architecture" (UMA) that can dramatically improve memory performance, which is also the way things work on the internal Apple Silicon.


How Apple Silicon Handles RAM

In case you haven't heard the news, let's rephrase it. Apple released new Macs in November 2020, including the new MacBook Air, MacBook Pro and Mac Mini, all of which use the M1, an ARM-based processor custom-designed by Apple. For a long time, people have been looking forward to this change. This is also the climax of Apple's ten years of designing ARM-based processors for iPhone and iPad.


M1 is a system-on-chip (SoC), which means that the processor contains not only the CPU, but also other key components, including the GPU, I/O controller, Apple neural engine for AI tasks, and most importantly, the physical RAM is also part of the package. To be clear, the RAM is not on the same chip as the basic part of the SoC.


Adding RAM to SoCs is nothing new. Smartphone SoCs may contain RAM, and Apple has decided to set aside the RAM module at least from 2018 onwards. If you look at the iFixit teardown of the iPad Pro 11, you'll see that RAM is also placed side by side with the A12X processor.


The difference now is that this approach also applies to the Mac, a sophisticated computer designed for heavy workloads.


Basics: What is RAM and Memory?


RAM stands for Random Access Memory. It is the main component of system memory, which is the temporary storage space for the data that your computer is currently using. It can be anything from the files you need to run the operating system, to the spreadsheet you are currently editing, to the contents of an open browser tab.


When you decide to open a text file, your CPU receives these instructions and the program to use. The CPU then takes all the data needed for these operations and loads the necessary information into memory. The CPU then manages the changes it makes to the file by accessing and manipulating the contents of the memory.


Typically, RAM can be inserted into a dedicated slot on a laptop or desktop motherboard, or it can be a simple square or rectangular module soldered to the motherboard. Either way, RAM for PCs and Macs has traditionally been a separate component with its own space on the motherboard.


M1 RAM: Separate Roommates


Thus, the physical RAM modules remain separate entities, but they are located on the same green substrate as the processor. "Shut up," I hear you say. "What's the big deal?" Well, for one thing, it means faster access to memory, which inevitably improves performance. In addition, Apple is adjusting the way memory is used in the system.


Apple calls its approach "Unified Memory Architecture" (UMA). The basic idea is that the RAM in the M1 is a single pool of memory that all parts of the processor can access. First of all, this means that if the GPU needs more system memory, utilization can be increased, while the rest of the SoC can be reduced. Better yet, there is no need to allocate a portion of memory for each part of the SoC and then shuttle data between two spaces in different parts of the processor. Instead, the GPU, CPU, and the rest of the processor can access the same data at the same memory address.


To understand why this is important, imagine how a video game works. The CPU first receives all the instructions for the game, and then offloads the data required by the GPU to the graphics card. The graphics card will then take all the data and process it in its own processor (GPU) and built-in RAM.


Even if you have a processor with an integrated graphics card, the GPU maintains its own block of memory just like the processor. They both process the same data independently and then move it back and forth between the memories. If you give up the requirement to move data back and forth, it's easy to see how storing everything in the same virtual filing cabinet can improve performance.


Here's how Apple describes its unified memory architecture on the official M1 website.


"M1 also features our Unified Memory Architecture or UMA. M1 unifies its high-bandwidth, low-latency memory into a single pool in a custom package. This allows all technologies in the SoC to access the same data without having to replicate across multiple pools of memory. This dramatically improves performance and power efficiency. Movie applications are easier. Games are richer and more detailed. Image processing is lightning fast. And the entire system is more responsive."


It's not just that every component has access to the same memory in the same location. as Chris Mellor points out in The Register, Apple is using high-bandwidth memory here. The memory is closer to the CPU (and other components), and access is faster than accessing traditional RAM chips connected to the motherboard via a socket interface.


Apple is Not the First Company to Experiment with Unified Memory


Apple is not the first company to come up with this solution. For example, about six years ago, NVIDIA began offering a hardware and software solution for developers called Unified Memory.


For NVIDIA, unified memory provides a single memory location that "can be accessed from any processor in the system". In NVIDIA's world, as far as the CPU and GPU are concerned, they will store the same data in the same location. However, in the background, the system is paging the required data between the separate CPU and GPU memories.


As far as we know, Apple is not taking any behind-the-scenes technology. Instead, each part of the SoC has access to the exact same location of data in memory.


The bottom line of Apple's UMA is faster access to RAM and shared memory pools, thus improving performance and eliminating the performance loss of moving data to different addresses.


Related Posts