Published in PC Hardware

TSMC and ARM announce 64-bit FinFET validation

by on30 September 2014

16nm FinFET big.LITTLE design

Foundry for hire TSMC and chip designer ARM have announced a key milestone in 16nm FinFET development – the first silicon validation of an ARM big.LITTLE core on TSMC’s cutting edge 16nm process.

The news comes just days after TSMC and HiSillicon announced the successful manufacture of 16nm FinFET processor based on ARM’s Cortex-A57 core.

Boba FinFETT Cortex-A57 at 2.3GHz

TSMC’s 16nm FinFET process (16FinFET or 16FF) is now in risk production. The companies collaborated to optimise ARM’s 64-bit ARMv8 processors designs for the new node.

The designs are now being optimised for TSMC’s improved 16FF+ process, which promises to deliver a 11% performance gain in the same power envelope for the A57 and a 35% reduction for the A53. The companies hope to finalise their work on 16FF+ optimisation by the end of the year.

The initial results for the first generation 16FF process are encouraging. The big Cortex A57 core manufactured in 16FF can achieve a sustained clock of 2.3GHz. It is backed by Cortex A53 cores running at the same clock and consuming an average of 75mW for most workloads.

Substantial gains for big.LITTLE designs

Pete Hutton, executive VP and president of ARM product groups, said the joint effort will eventually transform end-user experiences across the next generation of consumer devices and enterprise hardware.

"This silicon proof point with ARM Cortex-A57 and Cortex-A53 processors demonstrates the additional benefits in performance and power efficiency that 16nm FinFET technology delivers to big.LITTLE implementations," said Hutton.

Cliff Hou, VP of R&D at TSMC, said the company is proud to be the first foundry to validate an ARM big.LITTLE chip on 16nm technology. He said the collaboration with ARM will continue and the companies will strive to deliver market-leading SoCs for mobiles, servers and infrastructure applications using new nodes.

Rate this item
(0 votes)

Read more about: