Agilent, NVIDIA accelerate signal integrity simulations

Nvidia's CUDA-based computation acceleration hardware is expected to accelerate Agilent's ADS Transient-Convolution Simulator, allowing designers to perform fast "what-if" design-space exploration using circuit-level models that can be verified against measured data and EM simulation of the artwork.

By Ann Steffora Mutschler, Senior Editor -- Electronic News, 8/27/2008

To accelerate signal integrity simulations faster than previously possible, Agilent Technologies Inc said Tuesday afternoon that it is working with Nvidia to develop a commercial software release of a GPU-enabled Advanced Design System (ADS) Transient Convolution Simulator, based on Nvidia’s Compute Unified Device Architecture (CUDA)-based GPU.

With designs running at high data rates, signal integrity engineers must take into account physical phenomena like impedance mismatch, reflections, electromagnetic coupling, crosstalk, and microwave frequency attenuation due to the skin effect and dielectric loss tangent, and Nvidia’s CUDA-based computation acceleration hardware is expected to accelerate Agilent's ADS Transient-Convolution Simulator, allowing designers to perform fast "what-if" design-space exploration using circuit-level models that can be verified against measured data and EM simulation of the artwork, the companies said.

Agilent believes applications that will benefit from the CUDA-based GPU acceleration include design and verification of chip-to-chip multigigabit/s serial links which are found in almost all consumer and enterprise digital products, from laptop computers to data center servers, telecommunication switching centers and Internet routers, with the accelerated simulation expected to help manufacturers of these products improve time-to-market by arriving at an optimum design through rapid and complete exploration of the design space and avoiding costly and time-consuming prototype iterations.

“By employing the CUDA development environment to harness the parallel architecture of the GPU, Agilent has significantly enhanced and accelerated its tools, which solve critical simulation problems for Nvidia. Using Agilent's new CUDA-enabled tools, our engineering team was able to simulate our data path in parallel. We achieved a 14x improvement in simulation time, sped up our NPI process and further increased our design velocity,” noted Tommy Lee, VP of system design and manufacturing at Nvidia, in a statement.

Agilent said the ADS Transient-Convolution Simulator running on NVIDIA’s CUDA-based GPUs is expected to be available in Q1 2009 with beta evaluation of the simulator expected in October.