This site may earn chapter commissions from the links on this page. Terms of utilize.

Nvidia-Feature

Nvidia's Pascal GPU family has been tremendously successful, only the company is already looking alee to a new architecture, codenamed Volta. Details on the scrap are scarce, though Nvidia has previously given guidance that Volta would be its first consumer HBM2 product and would feature a stacked retention architecture. Now, rumors are flight that the chip will actually be congenital on TSMC's "new" 12nm node.

The reason nosotros've put the word in quotes is because 12nm isn't really a new node at all. Here's how TSMC's CC Wei described it in TSMC's terminal quarterly conference phone call.

[O]ur strategy is continuously to improve every node in the performance, such as 28 nanometer. And are standing to improve the xvi nanometers technology. And we have some very proficient progress, and you lot might telephone call information technology the 12 nanometer considering we're improve in the density, classical density, operation and power consumption. Yeah, we have that.

TSMC appears to exist performing the same kind of optimization and improvement to existing nodes that every foundry performs over time. It'due south only that this time, they're calling information technology something a little unlike.

TSMC-28nm

This photograph shows TSMC'south diverse types of 28nm technology and their respective characteristics at a very high level. The deeper you drill into the characteristics of each sub-blazon, the larger the differences become. A device congenital on TSMC's 28LP process would have very dissimilar characteristics than a device built on 28HP, even though both are based on 28nm.

Over the last 20 years, it'southward become increasingly difficult to cleanly define a node. In the one-time days, node size and gate length were synonymous. As time has passed, the metrics for a new node accept grown more complex. Samsung, TSMC, and Intel all tend to brand their products as belonging to the same node (save for TSMC going with 16nm as opposed to 14nm), but that doesn't mean they are identical, as the chart beneath demonstrates:

Intel-Features

Intel, TSMC, and Samsung feature sizes at 14nm.

Over the last few years, we've started to see more and more manufacturers taking this route. Samsung'south 14nm technology is based on a respin of its original 20nm process, not the 20nm procedure it publicly reported when information technology beginning discussed the engineering science. Information technology's debuted multiple variants of 14nm, while TSMC has both 16nm and 16FF+. Intel has respun 14nm into 14nm+, and some rumors advise the firm volition do a third respin (14nm++). TSMC likely isolated some additional improvements it could brand to its 16nm node and decided to adopt a new naming convention.

Several years ago, I predicted nosotros would begin to see more firms debuting their own methods of identifying procedure nodes as advances became more difficult and the speed of improvement fell. Samsung is positioning its 10nm node as a long-lived node, while TSMC believes its 10nm will exist a brusk-lived procedure. Calling its new 16nm refresh "12nm" allows the visitor to appear to stay in near-lockstep with the residue of the manufacture, as opposed to falling backside it (fifty-fifty if that 'falling behind' is illusory and based in marketing classification, non fact). Intel's 10nm node will be smaller than Samsung'due south, simply Intel generally marches to the beat out of its own drum. But we're seeing a fleck of jockeying for position and fifty-fifty different plans around specific nodes today that didn't used to be the example.

TSMC's 12nm procedure will undoubtedly deliver some improvements over its 16FF+ node, merely information technology'southward not going to be the same as a full-node shrink, and the improvements may range from small-scale to minor. I'm not too worried virtually how this could impact Volta, though. Nvidia has already demonstrated that information technology tin deliver a new GPU on the same process node with significant functioning improvements, and Volta could turn out to exist a major accelerate over Pascal, even if its 12nm node is more than of a 16nm tweak.