ARM and TSMC tape out 20nm FinFET Coretex-A57 CPU

History is made at night, character is what you are in the dark. – Lord John Whorfin

Apr 2, 2013 by

ARM logoIt looks like TSMC and ARM are getting serious about 64-bit CPUs, they just taped out a Coretex-A57 CPU on 20nm. It looks like you need v8 power to swim nowadays, FinFETs aren’t enough on their own.

Bad puns aside there are three things to care about here, the tapeout on 20nm, a new process nomenclature, and where that puts TSMC in the nebulous world of PR roadmaps. Most important is the basis of the release, there is now a tapeout of a Coretex-A57 CPU, not SoC so it is likely a test part, on TSMC’s 20nm process. This process is a FinFET design, but that is all that was said. Basically, don’t look to buy a TSMC 20nm A57 chip around Computex 2013 time, that’s not the point.

Next up is the rather nebulous claim that this allows a company to leverage 16nm efficiencies through magic hand waving and PR distractions or something. To use non-technical terminology, this is basically BS, but the press will eat it up and parrot it back until people think it is ‘truth’. That said, this is nothing that will actually be real to the end user, so they could claim it will bring world peace next week too. The point of all this is that TSMC is calling their next node 16nm not 14nm like the Common Platform guys and Intel.

Last up is probably the most important part of the release, essentially where it puts TSMC relative to the Common Platform crew, Global Foundries, Samsung, and IBM. GloFo announced a 20nm FinFET tapeout of an A9 in December of 2011, and like TSMC’s 20nm A57, it is a test part. You still can’t buy a 20nm GloFo anything unless they started selling competitive intelligence laden key chains in the lobby gift shop.

If press releases about ARM CPU tapeouts map to the reality you and I occupy, then that puts Global Foundries 15+ months ahead of TSMC on 20nm. The chip itself is irrelevant as a measure of progress, the pretty designs themselves mean nothing to the end result. More importantly, these tapeout announcements have nothing to do with the progress of the process itself, that stuff was, is, and will be done behind closed doors. The tapeouts that mattered were likely done a long long time ago and will never be made public.

If they were it would be the aforementioned competitive intelligence goldmine as well as something the public really doesn’t care about. How does the headline, “TSMC announces 20nm tapeout of rev3 of their 6th test structure die, the 19th and final test chip before 20nm is finalized!” sound? I know, heady stuff but even those in competitive intelligence would probably not read the full release before they died of boredom. The same goes for today’s 20nm Coretex-A57 tapeout announcement, it is a nice headline grabber but totally irrelevant to real news or progress.S|A

The following two tabs change content below.
My Twitter profileMy Google+ profileMy LinkedIn profile

Charlie Demerjian

Roving engine of chaos and snide remarks at SemiAccurate
Charlie Demerjian is the founder of Stone Arch Networking Services and SemiAccurate.com. SemiAccurate.com is a technology news site; addressing hardware design, software selection, customization, securing and maintenance, with over one million views per month. He is a technologist and analyst specializing in semiconductors, system and network architecture. As head writer of SemiAccurate.com, he regularly advises writers, analysts, and industry executives on technical matters and long lead industry trends. Charlie is also available through Guidepoint and Mosaic. FullyAccurate