D2S Computational Design Platform (CDP)
GPU-Acceleration for Semiconductor Manufacturing
The computing applications used in semiconductor design and manufacturing have ever-increasing requirements for speed, accuracy and reliability. CPU-only solutions have limited scaling options. GPU-acceleration offers the needed boost to semiconductor manufacturing applications. Just as importantly, GPU-acceleration opens up the opportunity to provide differentiation for your semiconductor manufacturing equipment through software solutions that add value for your customers.
Benefits
  • Offers software-based differentiation for semiconductor manufacturing equipment
  • Enables real-time, inline processing of image-based data
  • Comprises an off-the-shelf solution: custom ASICs and FPGAs are not required
  • Provides solutions that ease migration to new generations of equipment
  • Avoid reimplementation and re-certification
  • Spend engineering resources for enhancements and additional solutions
  • Costs less to create, has lower cost of ownership than custom-hardware solutions
D2S CDP offers lower cost of ownership

To realize the >10X acceleration potential of GPU-based computing, you must understand when to deploy GPUs and when to use CPUs. The D2S CDP combines the strengths of each to achieve optimal overall performance.

Economics also play a part. The D2S CDP offers a cost-effective, off-the-shelf solution that is less expensive to build, and to own (including power consumption and space requirements, and repairs).

The table below shows a comparison of the D2S CDP solution with a new-generation CPU-only configuration. The D2S balanced computing solution has better performance/$ and performance/watt.
Gaussian Convolution run on CPU-Only and on CPU+GPU
Data size: ~80 μm by 80μm, 10nm pixels
In 7U Rackmount Space CPU-Only CPU+GPU
Processor Cost Per 7U $164,000 $67,900
Power Consumed Per 7U 5800W 6090W
Speed of eBeam Simulation 2.27 Units of Time 1.67 Units of Time
Performance/$ 2.7 8.8
Performance/Watt 76 98


Hardware Configurations:
CPU-Only 40 E5-2699 v4, a 22-core 2.8GHz CPUs
CPU+GPU 14 Intel Xeon E5-2630 v3 2.6GHz CPUs 14 NVIDIA K-80 GPUs
D2S CDPs are deployed worldwide in a variety of manufacturing settings

Reliability is always a key concern for any computing platform. The clean-room proven D2S CDP keeps mission-critical production lines up and running, and minimizes costly downtime.

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.