GPU Accelerated Computing

Maximizing performance for the 24/7 semiconductor manufacturing environment.


The computing applications used in semiconductor design and manufacturing have ever-increasing requirements for speed, accuracy and reliability. The continuation of Moore’s Law creates a perpetual demand for greater accuracy as, with each new process node, larger numbers of increasingly smaller features are crowded onto each mask and wafer. Computing farms, where thousands of central processing units (CPUs) are strung together to handle many jobs in parallel, have become ubiquitous to address the need for faster computation. However, CPU clock-speed gains began to decrease at about 3GHz, and CPUs turned to greater bit-widths and multi-core and multi-processor configurations so that their performance could continue to scale. Graphics processing unit (GPU)-acceleration offers the needed boost to CPU-only computing to address the on-going requirements of the semiconductor industry. Market-proven GPU-acceleration solutions combine the strengths of both CPU and GPU computing to achieve optimal acceleration and cost/performance ratios. The semiconductor-manufacturing environment is sensitive to any downtime, particularly in the 24/7 clean room. A computing platform with mean-time-between-failures and mean-time-to-repair good enough for the clean room is ready for any deployment in semiconductor manufacturing. Today, GPU-accelerated systems have been deployed successfully in semiconductor manufacturing by companies such as NuFlare and Advantest.

To read more, click here.

Leave a Reply

(Note: This name will be displayed publicly)