TSMC and ARM Unveil Roadmap for 64-bit ARM-based Processors on 10FinFET Process

[btarunr]

ARM and TSMC today announced a new multi-year agreement that will deliver ARMv8-A processor IP optimized for TSMC 10FinFET process technology. Because of the success in scaling from 20SoC to 16FinFET, ARM and TSMC have decided to collaborate again for 10FinFET. This early pathfinding work will provide valuable learning to enable physical design IP and methodologies in support of customers to tape-out 10FinFET designs as early as Q4 2015.

"ARM and TSMC are industry leaders in our respective fields and collectively ensure the availability of leading-edge solutions for ARM-based SoCs through our deep and long-term collaboration," said Pete Hutton, executive vice president and president, product groups, ARM. "Our mutual commitment to providing industry leading solutions drives us to work together early in the development cycle to optimize both the processor and the process node. This joint optimization enables ARM silicon partners to design, tape-out and bring their products to market faster."

TSMC will be applying the learnings from prior generations of 20SoC and 16FinFET in the ARM ecosystem to offer performance and power improvements at 10FinFET that will be better than previous nodes. The ARM ecosystem can also take advantage of TSMC's Open Innovation Platform (OIP) which includes a set of ecosystem interfaces and collaborative components initiated and supported by TSMC.

"TSMC has continuously been the lead foundry to introduce advanced process technology for ARM-based SoCs," said Dr. Cliff Hou, TSMC vice president of R&D. "Together with ARM, we proved out in silicon the high performance and low power of the big.LITTLE architecture as implemented in 16FinFET. Given the successful adoption of our previous collaborative efforts, it makes sense that we continue this fruitful partnership with ARM in future 64-bit cores and 10FinFET."

The joint innovations from previous TSMC and ARM collaborations have enabled customers to accelerate their product development cycles and take advantage of leading-edge processes and IP. Recent benefits have included early access to ARM Artisan Physical IP and tape-outs of ARM Cortex-A53 and Cortex-A57 processors on 16FinFET.

View at TechPowerUp Main Site

 

[deviant88]

"as early as Q4 2015" wow wasnt 16FinFET supposed to be released in 2015 how can they do 10nm FinFET by the end of 2015 seems unrealistic to me,not even intel is that close they just announced they will build a 6B$ fab in Israeli,let alone be ready for production next year or 2016,not even close.

 

[iO]

If it takes as long as with 16FinFETs, which taped out in April 2013, there might be actual products maybe around 2017...

 

[Sony Xperia S]

Liars!

16nm FinFET and 16nm FinFET+ are expected to enter TRIAL production at some time now, or the end of 2014.
16nm FinFET Turbo is not even expected to begin this year but maybe ~ a year later.

So, to speak about 10nm now is something very very ugly.

 

[HumanSmoke]

"as early as Q4 2015" wow wasnt 16FinFET supposed to be released in 2015 how can they do 10nm FinFET by the end of 2015 seems unrealistic to me

The PR release specifies tape out, not production. An idea of the manufacturing process time and process ramp should be readily apparent from 16nmFF. First tape out February 2014...........First silicon September 2014.....Commercial silicon is slated for early 2015, so you're looking at a year (or slightly longer) between tape out, through risk production to volume production.

 

[Sony Xperia S]

The PR release specifies tape out, not production. An idea of the manufacturing process time and process ramp should be readily apparent from 16nmFF. First tape out February 2014...........First silicon September 2014.....Commercial silicon is slated for early 2015, so you're looking at a year (or slightly longer) between tape out, through risk production to volume production.

Actually, to be honest, I wouldn't be surprised at all if they have had 10nm chips for while now which will be put in the wild much late.

I doubt that here someone cares about apple and what they will do in the beginning of 2015, I would ask you about AMD and nvidia. GPUs is the thing that matters.

 

[HumanSmoke]

Actually, to be honest, I wouldn't be surprised at all if they have had 10nm chips for while now which will be put in the wild much late.

Probably depends more on ASML's lithography tool validation and refinement as well as fulfilling tool orders. Running a process once you have the tooling installed and validated is only the first step. Bringing the production speed (wafers per hour) up and lowering overall power demand - commercial viability- would be the next stages.

I doubt that here someone cares about apple and what they will do in the beginning of 2015, I would ask you about AMD and nvidia. GPUs is the thing that matters.

Large IC's will always trail far behind the small chips. If any wafer has a quantifiable range of defects, then spreading those defects out over a high number of chips per wafer allows the IHV and foundry to better validate the working silicon as well as fix the defective chips. A large IC might have numerous defects and that would add considerably in tracking down causation, as well as limiting working chips for the validation/testing phase.