Arm Opens Up To Using Intel's 18A Process For Leading-Edge SoCs

[phoronix]

Phoronix: Arm Opens Up To Using Intel's 18A Process For Leading-Edge SoCs

Intel Foundry Services (IFS) has racked up a big win today with Arm over enabling chip designers to make use of Intel's upcoming 18A process for low-power Arm SoCs...

https://www.phoronix.com/news/Intel-IFS-With-Arm-18A

[cl333r]

This from the article's end:

Intel's 20A and 18A processes are similar to TSMC's N2 (2nm) and should be entering mass production during the back half of 2024.

should be at its start.
I googled (what's Intel 18A) after reading the first few sentences only to find it explained at the very end.

[ryao]

This from the article's end:

should be at its start.
I googled (what's Intel 18A) after reading the first few sentences only to find it explained at the very end.

Here is a decoder:

Intel 7nm = Intel 5
Intel 7nm+ = Intel 4
Intel 5nm = Intel 20A
Intel 5nm+ = Intel 18A

I suspect that Intel went from understating their process technology to overstating it, since Intel 20A should be similar to TSMC N3, while Intel 18A should be similar to TSMC N2, if past trends hold.

[Ironmask]

I want to make fun of intel for making the angstrom standard because they got tired of losing the nm battle and just decided to switch to another standard entirely, but honestly the nm scale is getting a bit ridiculous and I don't blame them, angstrom is a neater measurement.

... But you know they totally switched to angstrom because they got sick of losing the nm battle.

[Yalok]

Looking at the recent history, I bet 18A will be delayed for at least 2 more years.

[Anux]

I want to make fun of intel for making the angstrom standard because they got tired of losing the nm battle and just decided to switch to another standard entirely, but honestly the nm scale is getting a bit ridiculous and I don't blame them, angstrom is a neater measurement.

I'm not sure what angstrom does better than nm before? It saves them from using a decimal point but it's not more accurate or better in any way. The core problem still is, that those numbers tell you nothing about transistor density of a real product and are not at all compatible between manufacturers.

[coder]

Here is a decoder:

Intel 7nm = Intel 4
Intel 7nm+ = Intel 3
Intel 5nm = Intel 20A
Intel 5nm+ = Intel 18A

Fixed that, for you.

The names should be treated as somewhat opaque. They publish descriptions of what changes the nodes actually entail (EUV, GAA, backside power-delivery, etc.), as well as the relative density, performance (ISO-power), and power (ISO-performance) improvements.

https://www.anandtech.com/show/16823/intel-accelerated-offensive-process-roadmap-updates-to-10nm-7nm-4nm-3nm-20a-18a-packaging-foundry-emib-foveros

[coder]

I want to make fun of intel for making the angstrom standard because they got tired of losing the nm battle and just decided to switch to another standard entirely, but honestly the nm scale is getting a bit ridiculous and I don't blame them, angstrom is a neater measurement.

I though they were moving in the right direction by going unit-less, with names like "Intel 7" and "Intel 4". To then speciously re-introduce a new unit is pretty sad. I'd have rather seen them go on to a letter-based naming scheme.