AMD is promising a 5nm node for its upcoming chips “highly optimized for high-performance computing.” The processing technology of a GPU or CPU is one of the key aspects of its design, but smaller is not automatically better. That’s why AMD didn’t just jump straight to the absolutely cutting-edge 3nm or 4nm process; because one nanometer isn’t necessarily the same as the other.
When we review a new generation of CPU or GPU, we always talk about some process technology. Companies have been eager to move to smaller nodes. There are several key reasons for this. Other comparable chips on smaller nodes generally offer better power efficiency and therefore output less heat. Another key reason is that smaller transistors and smaller dies mean more chips can be squeezed out of a traditional 300mm wafer. More dies per wafer means more end product to sell and higher profits.
Of course, things are not always that simple. Power efficiency gains are often used in pursuit of higher performance, which sometimes means that chips on smaller nodes may end up requiring more power and running hotter than their predecessors. Also, one company’s nodes are not always directly comparable to another company’s nodes. Sometimes, frontier nodes simply cannot compete with mature nodes. Intel’s long-awaited 10nm process is a good example of trying to make room for itself.
Currently, the 5nm node is the most advanced process in widespread use. Many mobile processors are manufactured using the 5nm process. Apple is clearly a company worth considering. In late 2020, Apple released its first 5nm product, led by the iPhone 12 series powered by the A14 SoC.
Then came the M1 CPU, which is found in the iPad Pro and various Macintosh laptop models. These chips are optimized for low-power use cases and battery life at the expense of full performance, though they are far from slow. In these cases, the disadvantages of using cutting-edge process technology are not obvious. A processor that consumes 15W is not the same as a processor that uses 125W or more.
Moving forward, TSMC has Prototype 3nm product, mass production is expected later in 2022. AMD’s Zen 4 processors are also due in late 2022, but they’ll be made using the 5nm node.
Why not 3nm? AMD CEO Lisa Su and Ian Cutress from Anandtech On the matter, saying: “The technology roadmap is all about making the right choices and the right junctions” and “Our 5nm technology is highly optimized for high performance computing – it doesn’t necessarily have to be the same as some other 5nm technology. same. “
So while AMD may appear to be a node that’s “behind” the cutting edge, it’s more important that its 5nm process choices work best for its products. Zen 4 needs to be tuned for desktop, enterprise, and mobile applications, and its high-core-count Epyc models cover TDPs from less than 10W to 400W or more.
The same goes for Intel and its 12th Gen CPUs. Intel refers to its 10nm node as “Intel 7” – apparently a marketing ploy – even though its transistor density is reportedly as good as the competing smaller 7nm node.
But beyond that, see what Intel is able to achieve with a clear process disadvantage: its Core i9 12900KS hits 5.5 GHz out of the box. This is quite an astonishing achievement. OK, it takes a lot of power to get there, but it’s still 5.5 GHz!
What does all this mean for gamers? The key is not to pay too much attention to which product uses which process node. The core architecture is just as important as the rest of the related product or system. When considering purchasing your next component, consider not only overall performance, but also power consumption, cooling requirements, and performance per watt. Oh, and the little factor of price. These raw process node counts are only part of the bigger picture, and often just pure marketing.