By Peter Clarke, Silicon Strategies, 04/26/2004
This article © CMP Media, LLC
REDWOOD CITY, Calif. - Arithmatica, Inc., a company that claimed it could improve silicon performance and reduce die area with its innovations in mathematical circuit blocks, has revealed that Xilinx Inc. and Nvidia Corp. are two companies that have used its CellMath libraries.
The company's chief executive has also told Silicon Strategies that Arithmatica is already applying its technology to a 65-nm manufacturing process with a secret partner.
Xilinx has implemented the math circuits in a forthcoming Virtex-DSP field programmable gate array while Nvidia has used the technology with a forthcoming graphics chip. This is thought to be code-named NV40 which is expected to launch soon. Arithmatica has also worked with Layer N Networks, Inc.
Silicon Strategies wrote about the CellMath libraries in March.
Applying the libraries to math-critical blocks in graphics chips can reduce overall chip area by up to 10 percent and can improve performance in processor designs to varying degrees, depending on the application, the company claimed.
"Our libraries sit above those of Artisan and Virage and target process through them," said Dave Burow, chief executive officer of Arithmatica. As a result, Burow said, Arithmatica already had experience of targeting its math circuitry at 130-nm manufacturing processes technologies from TSMC and UMC, 90-nm processes from TSMC, IBM and "a few other companies."
"I won't say who we are working with at 65-nm," said Burow.
"Using Arithmatica's CellMath Graphics Library, we were able to achieve a new threshold of performance in our next-generation of graphics processors while reducing the chip area dedicated to calculations in many of our major blocks by typically 20-30 percent," said Gopal Solanki, vice president of platform products at Nvidia, in a statement issued by Arithmatica.
"Arithmatica helped Xilinx achieve a very aggressive performance goal by increasing the speed of next-generation DSP operations by more than 50 percent - setting a new industry benchmark," said Jennifer Wong, director of IC Design at Xilinx, in the same statement.
"The explosive growth in consumer electronic products that require 3D graphics, high-quality audio and video, and high-bandwidth data transmission is driving the demand for powerful, easily implemented, cost-effective silicon math. We are pleased that technology leaders like NVIDIA, Xilinx, Layer N and others have successfully deployed our technology to improve the speed, area and power consumption of their math-intensive chips for their next-generation products," said Burow.
Arithmatica's licensing model includes a project-based, non-royalty fee pricing structure. The CellMath graphics library, processor library and configurable instances are available now for licensing, priced from US$175,000.
Arithmatica was included on the recently published Silicon Strategies' 60 Emerging Start-ups list.
