Using Platform LSF To Harness Non-Dedicated Computational Resources for LS-DYNA Crash Simulations at DaimlerChrysler

Computer aided engineering (CAE) tools are an important part of the product development lifecycle. These tools are used to perform complex simulation and analyses during the design phase. CAE requires significant computational resources in order to meet the response time requirements of the design engineers that use these applications. The speed and volume of CAE work provides a competitive advantage to the manufacturer by helping to bring products to market faster, reducing the need to build costly prototypes, and by increasing the quality of the end product. This paper documents a comprehensive approach to extending the HPC cluster grid to DaimlerChrysler’s engineering workstations. This approach will allow DCX to transparently make use of idle workstation CPU cycles and thereby significantly increase overall computing power at a fraction of the cost of a comparable dedicated HPC solution. This solution will also provide the foundation for further grid deployment to the DCX infrastructure so that additional benefits can be achieved in the future. By collectively harnessing the latent power of existing resources, Platform and DCX feel it maximizes the value of assets already owned while it gains compute power to accelerate and refine research and analysis, without any impact on the daily usage of these workstation users. The typical utilization of a desktop in the enterprise is about 5-7%, and large enterprises have tens of thousands of desktops. This will provide a cost effective path for increasing computing power while avoiding additional procurements of HPC cpu’s during usage spikes.