AMD is promising a 5nm node with its upcoming chips that are “highly optimized for high-performance computing.” The process technology of a GPU or CPU is one of the key aspects of its design, but smaller does not automatically mean better. This is why AMD isn’t going straight to the full bleeding edge of the 3nm or 4nm process; Because one nanometer is not necessarily the same as another.
When we review a new CPU or GPU generation, we always talk a little bit about process technology. Companies have a never-ending desire to move to even smaller nodes. There are some major reasons for this. An otherwise identical chip on a smaller node usually provides better power efficiency, and therefore produces less heat. The second major reason is that smaller transistors and smaller dies mean that more chips can be squeezed out of a traditional 300mm wafer. More die per wafer means more final product to sell, and ever more profit.
Of course, it’s not always that easy. Power efficiency gains are often used in search of higher performance and sometimes mean that a chip on a smaller node may require more power and heat up more than its predecessor. Also, the node of one company is not always directly comparable with another. Sometimes, a bleeding edge node simply cannot compete with a mature node. Intel’s long gesturing 10nm process is a good example of one that struggled to carve a niche for itself.
At present, the 5nm node is the most advanced process in widespread use. Many mobile processors are made with the 5nm process. Apple is the obvious company to reference. In late 2020 Apple released its first 14nm products led by the A5 SoC powered iPhone 12 range.
The M1 CPU followed, which is found inside the iPad Pro and various Macintosh laptop models. These chips are perfectly optimized for low power use cases and battery life at the expense of performance, although they are far from being slow. In those situations, the drawbacks of adopting bleeding edge process technology are not obvious. A processor that consumes 15W is not the same as one that uses 125W, or a lot more.
moving forward and TSMC is already Prototype 3nm Product, with mass production expected later in 2022. There are AMD’s Zen 4 processors coming in late 2022, but they will be built using a 5nm node.
Why not 3nm? Spoke with AMD CEO Lisa Su Ian Cutress from AnandTech of the same point, and added: “Technology roadmaps are all about making the right choices and the right turns” and “Our 5nm technology is highly optimized for high-performance computing—not necessarily the same as some other 5nm technologies . “
So, while AMD may seem to be a node ‘behind’ the bleeding edge, it is more important that its 5nm process options are tailored to best suit its own products. The Zen 4 needs to be tuned for desktop, enterprise, and mobile applications that cover TDP everywhere from sub 10W to 400W or more for its higher core count Epic models.
best cpu for gamingTop Chips from Intel and AMD
best graphics card: Your perfect pixel-pusher awaits
best ssd for gaming: Get in the game ahead of the rest
The same can be said of Intel and its 12th generation CPUs. Intel calls its 10nm node ‘Intel 7’ – which is clearly a marketing tactic – although its transistor density is reportedly about as good as that of smaller 7nm nodes.
But other than that, see what Intel has been able to accomplish, despite its obvious process disadvantages: Its Core i9 12900KS can reach 5.5GHz out of the box. This is a rather surprising achievement. OK, it takes a lot of power to get there but still, 5.5 GHz!
What does all this mean for gamers? The main thing is not to focus too much on which product is using which process node. The core architecture matters as does the rest of the product or system. When considering your next component purchase, look at performance outright, but also look at power consumption, cooling requirements, and performance per watt. Oh, and there’s the small factor of price. Those raw process node numbers are only one part of the bigger picture, and are generally just pure marketing as well.