World's Fastest Multi-Core Geometry Engine

GPP Logo

Fixstars' Geometric Performance Primitives (GPP) library is perhaps the only computational geometry engine that's optimized for massive parallelization with accelerated NVIDIA GPU technology. This gives GPP incredible speed and the ability to process truly enormous datasets.

Computational geometry engines play a pivotal role
in advanced Graphical Information Systems (GIS),
Electronic Design Automation (EDA), computer vision, and motion planning solutions. Visualization and analysis of ever increasing amounts of data requires a computational foundation that can meet the challenge with flexible algorithms that are optimized for the latest generation of multi-core processors.

Key Features

  • High performance with accelerated GPU parallelization
  • Robust functions based on cutting-edge algorithms
  • Up to 25 times faster performance than reference CPUs with reduced memory footprint

Applications

The GPP engine provides a highly optimized computational foundation with the flexibility of deploying on accelerated GPU and CPU hardware targets. Designed from the ground up for robustness, GPP allows up to 53-bit coordinates, with no artificial limits on data size, and processing of all-angle geometry. Combined with overlay analysis and geometry-on-geometry checks, GPP is the only computational geometry library that combines state of the art analysis with massive multi-processor flexibility.

Electronic Design Automation (EDA)

CPU die

Many current EDA solutions rely on deployments that can only operate in narrow processor configurations. Faster turn around times, high-precision designs, tests and verifications require ever more computational power. Parallelism and flexible hardware targeting, along with support for boolean operations, line intersections, and polygon relations, makes GPP an ideal solution for design-rule checking.

Graphical Information Systems (GIS)

Smart phone

Common GIS software is characterized by sequential processing, size-limited datasets, and CPU-bound computation. GPP on the other hand, provides access to GPU acceleration, unrestricted input layer size, and spatial databases while providing support for user-requested features, such as map overlays and snap rounding.

Performance

GPP performance chart

In tests with common GIS dataset layers, GPP runs 18 times faster than conventional algorithms, leveraging the computational power of GPUs, and seven times faster through multi-core parallelization on standard CPUs. In addition, GPP uses roughly half the memory required by competing algorithms.

For inquiries regarding Geometric Performance Primitives

Please contact us here.