Multi-processor code improves performance fourteen timesLEDAS Ltd., a developer of CAD computational components and products and a provider of software development services for the CAD/CAM/CAE/PDM/PLM market, today announced that it successfully installed its LEDAS Geometry Comparison engine on Amazon’s cloud service. The results demonstrated an excellent increase in speed for complex geometric software running on parallel computation engine.
LEDAS Geometry Comparison (LGC) is a solution that pinpoints differences between similar 3D models, just as revision control systems do with text. This comparison function is applicable to many areas of engineering software, including model interoperability in CAD, CAM, and CAE systems, comparing revisions in Product Data Management software, and validation and regression analysis of 3D models as a part of automated quality assurance systems.
The development of LGC technology began just last summer, with the goal of building from scratch a reliable, fast, and really parallel solution. Early studies showed impressive scalability for the LEDAS code, thanks to the programming team avoiding the Boolean operations that are typically used in other, less efficient implementations.
Amazon cloud was chosen as a mature Web service infrastructure on which to test LGC’s parallel capabilities. Through these tests, LGC showed a performance boost of 6x-7x when the cloud installation used 10 cores, then leaping to an amazing 14x on 40 cores.
LGS speed up factor as function of the number of cores