Go to Fixstars Solutions website open_in_new
Speed up your Grid Computing
M3 for Windows HPC cluster
M3 for Windows HPC cluster
Grid computing solution with high track records at financial institutions
M3 for Windows HPC cluster is a framework that provides functions necessary for financial institutions to use Windows HPC Server 2008 R2 as OS / middleware of cluster systems.
In the financial application such as derivative plicing tool and risk management system which require high computing power, it becomes possible to scale out the conventional server and shorten the calculation time.
M3 for Windows HPC cluster is adopted by Japanese bank's grid system and is in operation.
Even when the computational resources of thousands of cores are needed in a financial cluster system, which is rapidly increasing in computational complexity, it achieves high performance.
By utilizing the functions of Windows HPC Server 2008 R2, we can realize low price. The M³ for Windows HPC Cluster satisfies the missing functions of the financial cluster system more than enough to make a high-performance financial cluster system.
High fault tolerance is realized so that calculation can be continued even if a hardware failure occurs in a part of a large-scale cluster system or even a part of a large number of calculation applications fail.
M3 for Windows HPC cluster configuration
M3 for Windows HPC cluster is configured as a framework that users can easily use by hiding the complicated control of MS-MPI. We provide various functions necessary for financial institutions to use Windows HPC Server 2008 R2 as a cluster system.
Function detail
Main function
We provide various functions necessary for financial institutions to use Windows HPC Server 2008 R2 as a cluster system.
- Automatic rerun function on error
- Load balance function
- Easy to develop with SOA type API
Benefits of using Windows HPC Server 2008 R2
Recently, using Windows HPC Server 2008 R2, which attracts attention as a solution for financial cluster systems, has many advantages in using it in the financial grid.
- Enhanced job scheduler, management tool, reporting tool
- Low price with OS licensing pricing limited to HPC applications
- High-speed job execution function using MS-MPI
- Various functions such as priority execution of calculation job
Windows HPC Server 2008 R2
Windows HPC Server 2008 R2 is a product of Microsoft that bundled HPC middleware in server OS. We provide necessary functions for cluster system including Microsoft Message Passing Interface (MS-MPI) which maximizes performance of HPC.
For details, please see the manufacturer's product information open_in_new.
Benchmark
High speed communication of MS-MPI compared with SOA type scheduler
- Execute parallel processing similar to that of SOA type scheduler using MPI and investigate utilization status for each CPU core
- With MPI, parallel processing is performed very efficiently by shortening the data transfer time and reducing the time to start calculation (proven even in thousands of core environments)
- Perform about 1,000 tasks in 320 core grid environment
Benchmark environment
| Network | 1GBit Ether x 1 |
|---|---|
| Common data size | 30,000,000 (bytes) |
| Task data size | 160,000 (bytes) |
| Task execution time | 10,000 (ms) |
| Total number of task | 1,000 |
| Number of used cores | 320 |
SOA type scheduler
M³ for Windows HPC cluster
