ARM & Cadence Tape Out 20nm Cortex A15 Test Chip
by Anand Lal Shimpi on October 18, 2011 10:30 AM EST- Posted in
- CPUs
- Arm
- Cortex A15
- 20nm
- SoCs
Although we won't see the first ARM Cortex A15 based designs until the second half of next year, and even then only on 28/32nm processes, ARM and design tools supplier Cadence have announced the first tape-out of a 20nm Cortex A15 based test chip. Tape out signals the end of an overall design phase and the release of the design to the foundry for manufacturing. The Cortex A15 is expected to be a significant step forward for ARM, bringing its designs further up the chain into the low-end x86 notebook market in addition to current smartphone/tablet targets. Cortex A15 based designs will also go head to head with Qualcomm's Krait based Snapdragon S4.
The test chip will be fabbed at TSMC on it's next-generation 20nm process, a full node reduction (~50% transistor scaling) over its 28nm process. With the first 28nm ARM based products due out from TSMC in 2012, this 20nm tape-out announcement is an important milestone but we're still around two years away from productization.
Source: Cadence
Facebook
Twitter
RSS
13 Comments
View All Comments
gradjoh - Tuesday, October 18, 2011 - link
Nice to see Cadence get a place in the spot light, usually nobody pays attention to the cutting edge software that makes the design of these chips even possible.bravomail - Tuesday, October 18, 2011 - link
As iPhone 4S showed compared to iphone 4 - there is not much difference in day-to-day tasks now. But battery life could be improved on some Android phones, LTE phones - you name it.Wolfpup - Tuesday, October 18, 2011 - link
Are you kidding?!? A8 is DIRT SLOW. Dual core A9 is at least kind of tolerable, and feels 900,000,000x faster than an A8.gevorg - Tuesday, October 18, 2011 - link
Will this quad-core 20nm A15 chip be fast enough to run OS X? I'm wondering if in two years or so, Apple will be able to merge their desktop and mobile OS into one.Gauner - Tuesday, October 18, 2011 - link
Not yet and not for 3 or 4 years at least.It could run a specially designed version of OSX(much in the same way windows 8 can run in a cortex a9), but it would not count as convergence yet since it would still be very limited.
I dont think apple is going to do anything like that until you can at least open photoshop in it while you have a webbrowser and itunes on the background.
B3an - Tuesday, October 18, 2011 - link
Theres a better and more capable alternative called "Windows 8". It will be out sooner too. It even runs on current ARM CPU's.DanNeely - Tuesday, October 18, 2011 - link
Not well. Quadcore A15 should be somewhere in the same ballpark as the atom; so while you probably could run lighter apps without trouble it'd still collapse under anything that needs real processing power.Hector2 - Tuesday, October 18, 2011 - link
Well, I guess every baby step going forward is still a forward step. But, seriously, when is the last time anyone "announced" tapeout ? These days, maybe it'll be announced when silicon is produced, but obviously that's not going to happen with this. Normally, the immediate next step right after a tapeout is starting wafers. In this guess, starting 20nm wafers.Does ANYONE think they're starting 20nm wafers within even the next year ? Absolutely not. Ship a 20nm ARM processor for revenue in 4 years ? Yeah, right.
I used to be a front end (wafer) product engineer after switching from design. It was my job to get masks built after tapeout and get wafers started. I've done this step many times.
This was just a Cadence tool exercise, nothing more. They took an existing ARM design for a current process technology, did some simple shrinking of the CAD design, and Voila !, they taped out 20nm ! All it takes is compute resources.
20nm ? They could have done the same thing for 5nm and it would mean just as much.
jeremyshaw - Tuesday, October 18, 2011 - link
So it's just a non-working die? Just a test pattern of sourts?DanNeely - Wednesday, October 19, 2011 - link
It was probably a working die; but there's a long gap between the first working things you get on a process in the lab and when the process is ready for mass production.