Azul Systems Reduces Leakage Power with Sequence CoolTime

Azul Systems, pioneer of the industry’s first network attached processing solution, has adopted low-power EDA leader Sequence Design’s CoolTime to reduce leakage power in its massive SoCs without negatively impacting performance.

“What do you leave on the table, power or performance?” asked Paul Koike, Senior Director, Silicon Engineering at Azul. “We needed a tool to reduce leakage and fix all clock-slowing crosstalk at the same time –the only one that can handle all that and more is Sequence’s CoolTime.”

Eliminating the need for multiple point tools and iterations, CoolTime is the only nanometer SoC design tool that takes dynamic voltage drop into account while computing timing and signal integrity in a single tool.

CoolTime reduces leakage power using multi-threshold standard cell libraries while fixing crosstalk noise simultaneously. Its SI repair technology and multi-VT optimization are part of the same database which can run at the same time. Competing tools require these be run independently, leading to chip-closure convergence problems that force users to choose between fixing the leakage or fixing noise. For more information, go online.

“CoolTime is emblematic of our holistic approach to power,” said Sequence president and CEO, Vic Kulkarni. “Tackling it in isolation may solve one problem, but it creates others, and if your tool isn’t speeding design closure, what good is it?”

About Azul Systems
Azul Systems Inc, has pioneered the industry’s first network attached processing solution, designed to deliver massive amounts of compute capacity as a shared network service to transaction-intensive applications and services. Without application level modifications, binary compatibility requirements or operating system dependencies, this transparent new compute model significantly increases utilization across existing infrastructure, enables unprecedented application scalability and predictably ensures high levels of service with up to 50% less cost than traditional computing models.