Chip distributors together with AMD and Intel have for years outlined a chip by a nanometer measurement equivalent to 14nm or 10nm, which additionally described the manufacturing strategy of a chip. Such “nm” designations was practically as vital as clock pace, energy, or any of the opposite varied metrics of a chip. Intel, nonetheless, could also be getting ready to de-emphasize it fully.
What does “nanometer” imply in semiconductor manufacturing? “Nanometers” seek advice from the dimensions of the person transistors contained in the chip. The smaller the transistor, the extra dense the chip. Smaller transistors additionally suggest that chips could be run at quicker speeds, at decrease energy, or some mixture of the 2. Chipmakers use phrases like “10nm” and 7nm” to explain the manufacturing course of know-how used to make these transistors, and this terminology has develop into broadly adopted to point what’s modern, and what isn’t.
Over time, although, the phrases have develop into considerably muddy. Insiders already perceive that the “definition” of a chip manufacturing course of really relies on many variables, together with transistor density. Intel has said beforehand, for instance, that its unique 14nm course of had a transistor density of 37.5 megatransistors (MTr) per sq. millimeter, and that this elevated to 100.eight MTr per sq. millimeter with its 10nm course of. Many take into account Intel’s 10nm course of know-how to be on a par with TSMC’s personal 7nm course of, however the deeper particulars can already be complicated.
Intel, after all, has been plagued with manufacturing issues all through the previous few years, and caught on the 14nm manufacturing node even via its most up-to-date desktop processor, Rocket Lake S. So Intel’s continued use of the time period “14nm” actually doesn’t assist its chips look superior, from a advertising and marketing perspective.
That’s why, in line with sources near Intel in addition to The Oregonian, Intel is seemingly trying to change or eliminate manufacturing terminology from the conversation fully.
Each Intel and AMD have additionally begun incorporating design parts that complicate the method know-how even additional. Intel’s “Superfin” know-how, for instance, is technically a “10nm” course of, however the tweaks and enhancements that it gives helps put Intel at practically an equal footing to AMD’s 7nm Ryzen 5000 in our overview. Intel’s Foveros and AMD’s associated chiplet know-how place fully totally different silicon inside the identical bundle, which is reviewed as a single, monolithic product.
There are nonetheless measurable, concrete variations between AMD’s Ryzen and Intel’s Core and Samsung’s Exynos and Qualcomm’s Snapdragon on the engineering stage, and people variations are substantial sufficient to impress arguments amongst electrical engineers and hardcore lovers.
Longtime chip nerds may additionally recall a interval within the late 1990s, when AMD, Cyrix, SGS-Thomson and others developed the idea of “efficiency rankings”—arguing that whereas their chips really ran at slower clock speeds, they have been simply as quick because the 386 and 486 chips Intel marketed on the time. It’s considerably ironic to see Intel hinting that it could undertake the identical technique.
What Intel seems to be saying, nonetheless, is that manufacturing is now advertising and marketing. Maybe it’s, to a point. However what hasn’t modified are metrics like energy, worth, and the way nicely a given processor runs on varied functions. These will stay fixed, and must be what determines which processor to purchase.
Correction: A earlier model of this story incorrectly referred to Intel’s SuperFin enhancements as a part of its 14nm course of node; it is really a part of Intel’s 10nm course of node.