Voltaire Open MPI Accelerator™ (OMA™)
Features & Benefits
- The best of both worlds – accelerated, open source Open MPI
- Works transparently over InfiniBand or 10GbE
- HPC applications run much faster with OMA
- Improves inter-node communication by freeing intra-node resources
- Constant improvements, fixes and upgrades from Voltaire developers, who are key members of the Open MPI open source project
- Improved performance with less hardware
- Investing in software eliminates the need for huge investments under tight budgets
- Voltaire’s reputation for world-class customer support
The Open MPI Project is an open source MPI-2 implementation that is increasingly being adopted by both commercial and non-commercial high-performance computing environments. Due to fact that Open MPI is continuously being developed and maintained by a consortium of academic, research, and industry partners, Open MPI combines the technological expertise and resources from all across the High Performance Computing community.
At the same time, new multi-core CPU technologies are introducing new capabilities and enabling superior I/O capacities, resulting in an increasingly high demand for faster and more efficient HPC application performance.
Voltaire’s application acceleration family of products provides a set of applications that enhance the performance and value of various open source projects for server/storage environments.
As part of this family of application acceleration products, Voltaire Open MPI Accelerator™ (OMA™) significantly increases the performance of Open MPI in server and storage environments by effecting different communication patterns. Most OMA users see at least a 10% performance improvement, with many users seeing improvements of 20% or even higher. Voltaire’s extensive experience and knowledge in MPI and in Open MPI development, as well as its leadership in the HPC market, makes OMA an ideal performance boosting add-in for the community-grown Open MPI framework.
Key to its performance improvement capabilities is OMA’s Intelligent Rank Placement (IRP) feature. With IRP, OMA automatically determines the best performing algorithm based on rank placement and a given job and uses that algorithm for all future runs of the same application and cluster for optimal performance.
In an environment of shrinking budgets, OMA provides tremendous value on top of existing Open MPI infrastructures by offering an almost immediate return on investment for intensive HPC environments.
Proven Results from an Experienced HPC Leader
OMA has demonstrated proven improved performance on:
- ANSYS FLUENT
- ESI Pam-Crash
- OpenFoam
- HPCC
- IMB (Pallas)
- OSU
As a longtime contributor to the Open MPI project, Voltaire brings years of expertise to the development of OMA, as well as a long, successful track record of achievements in HPC. Voltaire has consistently been the switching leader on the TOP500 list of the world’s most powerful supercomputers, with more than 50 percent of all InfiniBand deployments — more than double the share of any other InfiniBand systems vendor.
