ARM’s close collaboration with the foundries on next generation processes is paying off.
Globalfoundries says that an ARM chip made on its 14nm finfet policy delivers twice the power efficiency of its ‘super low power’ (SLP) 28nm bulk process, says the foundry.
The comparison was done using an IC containing dual Cortex ARM A9 cores.
“Early collaboration on manufacturing process technologies allows us to identify and address SoC design challenges and reduce risks to adoption by our mutual customers,” says ARM’s Dipesh Patel.
At the moment the best processes being used by anyone to make ARM cores are 28nm, and ARM CEO Warren East says: “It’s too early to say when customers will move to 20nm.”
ARM-GoFo’s 14nm finfet IC represents the process generation after 20nm.
“Our deep collaboration with ARM is continuing to pay off as we work together to optimise ARM IP for our process technology,” says GloFo’s Mike Noonen.
Other findings from the 14nm implementation included: ARM processors on 28nm-SLP typically use 12-track libraries, however on the14nm process higher performance and more energy-efficient ARM processors can be implemented using 9-track libraries.
At constant power, the frequency achieved with 14nm technology based implementation (using 9-track libraries) is expected to be 61% faster than the frequency achieved with 28nm-SLP technology based implementation (using 12-track libraries).
At constant frequency, the power consumed by 14nm technology based implementation is expected to be 62% lower than the power consumed by 28nm-SLP technology based implementation.
The performance-power efficiency of IC made on the14nm process (expressed as DMIPs/milliwatts) is expected to be more than twice that of the 28nm-SLP technology based implementation, while using half the silicon area.