North American Systems
International
Where Innovation
Meets Solutions
Sometimes, high performance computing applications require a different approach at the system level, to achieve a required level of performance.
The first Cell BE-based systems, the BladeCenter QS servers are ideal for applications such as:
that accommodate single-precision, float-point, compute-intensive workloads.
Designed in collaboration between IBM, Sony, and Toshiba, the Cell BE processor contains a Power Processing Element (PPE) and eight Synergistic Processing Elements (SPEs) for a total of nine computing cores per processors. This revolution in processing technology also features an on-chip local store for each SPE, as well as on-chip L1/L2 cache, memory, and I/O controller.
The PPE includes a general-purpose 64-bit RISC processing element with 2-way multithreading and 32KB L1 instruction cache and 32KB data cache. Each of the eight SPEs is a special-purpose RISC processor with 128-bit SIMD and 256KB of local memory.
These high-performance capabilities and ultra high-speed communications can be optimized with a freely-available SDK (Software Development Kit) from AlphaWorks.
| IBM QS20 Blade Server Specifications | |
|---|---|
| Processors | 2 x 3.2GHz Cell BE (Broadband Engine) |
| L2 cache | 512KB per Cell BE Processor and 256KB local store memory for each SPE (synergistic processing elements) |
| Memory | 1GB XDRAM, 512MB per Cell BE |
| Hard drives | 40GB IDE |
| Optional Connectivity | 1 or 2 Infiniband 4X via PCI-Express |
| Networking | Dual Gigabit Ethernet |
| Compatible Operating Systems | Fedora Core5-based Linux |
The BladeCenter QS20 is a component of the IBM Cluster 1350 system offering.
North American Systems’ extensive IBM experience and experience with the Blade platform makes us an ideal partner to work with for those who are looking to pilot or expand their blade server projects and deployment. Our systems engineers can provide in-depth technical insight and expertise, to help you configure a system that meets business requirements while reducing infrastructure costs.