SANTA CLARA, Calif. Optical proximity correction (OPC) can take days or weeks for 65 nm ICs, and the problem will only get worse at 45 nm. Hoping to head off a computational nightmare, Mentor Graphics Corp. and Mercury Computer Systems are bringing out what they say is the first Cell Broadband Engine (BE) based acceleration platform for the EDA market.
Mentor Graphics announced Wednesday (Nov. 29) the availability of Calibre nmOPC, a next-generation OPC tool aimed at foundries and mask shops. While it has a number of software improvements aimed at 45 nm ICs, the star attraction is the "coprocessor acceleration" available from Mercury Computer's Dual Cell-Based Blade, which is just now going into production use.
Mercury Computer is a provider of computing systems and software for data-intensive applications, including image processing, signal processing, and visualization. In 2005, Mercury and IBM announced an alliance focused on the joint development, engineering and commercialization of Cell BE processor-based products. Mercury followed up with the release of the
Dual Cell-based Blade last year.
The Blade serves a number of compute-intensive applications, and was not built specifically for OPC. But Joe Sawicki, vice president and general manager of Mentor's design to silicon division, noted that Mentor and Mercury worked closely together to port Mentor's OPC modeling infrastructure onto the Cell, take advantage of parallelization, and optimize Calibre nmOPC for the Cell's Fast Fourier Transform (FFT) library.
"Absent innovation, we are facing a crisis in turnaround time with OPC," Sawicki said. As process sizes shrink, and model complexity grows, more and more CPUs are needed to get a reasonable turnaround time, he said. Even at 65 nm, according to Sawicki, some customers are using 1,000 processor nodes to run OPC.
With a coprocessor acceleration (CPA) cluster of 25-50 Cell processors, Sawicki said, 100 to 200 general-purpose CPUs can achieve the same computing capacity as 750 to 1,000 CPUs without a CPA. Cost, space, and power consumption will be dramatically reduced, he said.
IBM's multicore Cell architecture is "uniquely suited" to tackling OPC, Sawicki said. Originally aimed at gaming applications, the Cell contains one PowerPC processor and 8 "synergistic processing elements." The Cell's strength is rapid image processing. Compared to an Opteron processor, a Cell processor can speed OPC simulation using Fast Fourier Transforms (FFTs) by 50 to 100 times, Sawicki said.
Mercury's Dual-Cell-Based Blade offers peak performance of 400 Gigaflops, features two Cell BE processors, and includes XDR memory from Rambus Inc. Mercury also offers the MultiCore Plus Software Development Kit (SDK) to program the Cell BE.