Features/ Benefits

Designed for HPC and Built to Scale
• Delivering 13.5TF to over 1,000TF of high performance and availability computing
• Scalable performance to over 1000s of nodes
• Support for Eight/Twelve-Core AMD Opteron™ processors
• Dense architecture: 64 nodes/128 CPUs/1,536 cores in a standard rack cabinet.
• Appro Cluster Engine (ACE) Management Software including network, server, cluster, and storage management.
• Support options for NVIDIA Tesla 20 Series GPU computing

Balanced Architecture (Processors, Memory, Network)
• Scalability (SW & Network)
• Reliability (Real RAS: Network, Node, SW)
• Single or Dual Rail Fat Tree IB or 10GigE Network
• 3D Torus Option to be available in Q3 2010
• Standard rack air cooled or water cooled option
• Efficient Cooling Architecture: 22~28 KW/Rack
• Redundant PSU 90% Efficiency, 5+1, PFC
• MVAPCH2 or Intel MPI Library
• Intel Cluster Toolkit Compiler Edition
• Appro Cluster Engine™ Management Software
• Professional Services and Support

Energy-Efficient (Green) Architecture
• Reduced TCO with improved system density, reliability and computing capability while keeping the datacenter cool.
• Appro Water Cooled Rear Door cabinet options improves cooling efficiency of the entire system by up to 8X.
• “Power-Factor-Corrected” redundant power supplies that are 94% efficient compared to the 70% to 80% of regular power supplies.

Simplifies IT Management and Administration
• Appro Cluster Engine (ACE) Management Software - Robust management system

• ACE supports either local or remote disks. This capability with network boot provides users with the best of both worlds. The reliability and speed of network boot ensures that the system is ready quickly, allowing them to boot 100 to 1,000 blades about the same time while the local disks provide options for faster distributed disk access. This combination provides the added performance and reliability of ACE and the application flexibility of local disks.

Appro Cluster Engine™ Management Software
• Network Management
• Server Management
• Cluster Management
• Storage Management

Appro Cluster Management Architecture
• Scalable to very large configurations
• Hierarchical architecture
• Supports multiple interconnect topologies
• Network booting operation without lightweight kernel
• Reliable, available, serviceable (RAS)
     - Run-through network fault tolerance

Support and Services
• Closed-loop solution management, up-front consulting, pre-integration and testing with pre-sales and post-sales support.
• Support for Linux operating environments, with configurations that provide exceptional performance across a wide range of HPC applications.

AMD Opteron™ 6100 Series Processor

The new AMD Opteron™ 6100 Series processor performs when the workloads get tough. Based on the next-generation AMD Direct Connect Architecture 2.0 and up to 48 total cores in a configuration, this platform more than doubles the memory bandwidth, and provides nearly 2x the I/O bandwidth of previous generation 2P and 4P servers, helping your business tackle complex jobs with greater throughput, exceptional value and readiness to scale

Performance for Demanding Workloads
- Increased performance with higher core count 8 and 12 physical cores options with improved memory and I/O bandwith providing the power to tackle larger, increasingly complex and multi-threaded applications.

- Achieve STREAM bandwidth up to and exceeding 100GB/second with the high memory scaling of Direct Connect Architecture 2.0

- Total memory bandwidth without comparison with support for four memory channels running at up to 5.2 GT/s each

- Delivering the capacity and throughput to help avoid bottlenecks and provide users a highly responsive experience with high memory and I/O bandwidth to support online workloads

Platform Consistency that Scales
- Large-scale memory capacity on an x86 platform with DDR3 memory support for up to 50% more DIMMS, and up to 66% faster memory performance, and 12MB of L3 cache and options for 8 to 12 processor cores

- Massive memory footprints with up to half a terabyte of more total system memory giving you more bandwidth than ever before to handle large-scale workloads

- Get double the memory bandwidth and nearly 2x the I/O bandwidth with four, faster HyperTransport™ 3 (HT3) links and integrated CPU/chipset platform

Business Value without Compromise
- Take budgets further with affordable server choices and gain the advantages normally reserved for high-end systems, with exceptional value and low total cost of ownership

- RAS features to help keep applications online including the ability to turn off cores and bad memory cells, and enable automatic switching to online spares in the event of problems, helping to reduce downtime risks

- Stable platform and compatibility with the next generation AMD Opteron processor, with more cores, larger memory and faster I/O