By Margaret S. Gurney
Appro Xtreme-X1 Row of Clusters
After Appro (Milpitas, CA) announced last June that the Appro Xtreme-X1 Supercomputer adopted the certification methodology from the Intel Cluster Ready Program, the company demonstrated 40 gigabit per second (Gbps) InfiniBand Supercomputing Clusters at ISC/08, the International Supercomputing Conference held in Dresden, Germany, later that month.
The significance of Intel Cluster Ready certification is that the technology helps simplify the deployment, usage, and management of clustered computer systems through a standardized and replicable methodology to build clusters and run off-the-shelf high-performance computing (HPC) applications. Its automated and systematic diagnostic tool methodology enables independent software vendor (ISV) applications to run across certified cluster platforms.
|“Customers can now build a powerful, easy to manage cluster architecture that can dramatically reduce latency while accelerating I/O performance for a variety of enterprise and high performance computing applications.”
— Jim Ballew, Appro, CTO
When Appro demonstrated the 40Gbps InfiniBand Xtreme-X1 Supercomputer powered by Quad-Core Intel Xeon processors, it was using Mellanox ConnectX 40Gbps InfiniBand adapters and InfiniScale IV-based 40Gbps InfiniBand switch, and running an ANSYS-Fluent airplane structure application. The demonstration featured a powerful and balanced open supercomputer architecture, says the company, managed by Appro Cluster Engine software that combined high-performance capacity computing with fault-tolerant capability.
“Appro Xtreme-X1 Intel Cluster Ready-certification helps simplify deployment of high-performance computing clusters achieving faster and more accurate time-to-results,” said Daniel Kim, CEO of Appro. “IT departments in all high-performance computing organizations need to accommodate the pressure for shorter life cycles with a limited budget for software and hardware expenses.”
What the X1 Supercomputer Is
The Xtreme-X1 Supercomputer is designed as a building block to deliver a balanced, highly available supercomputer cluster with high reliability and easy-to use system management. It is suitable for such HPC applications as CFD (computational fluid dynamics), aerospace and automotive engineering simulations, petroleum exploration and production, scientific visualization for oil discovery and recovery, research in seismic, weather and environmental sciences, defense, and classified research production.
A fully integrated and prepackaged solution, the Xtreme-X1 Supercomputer supports Quad-Core Intel Xeon processors with 128 sockets with 512 cores and 4TB (terabytes) of memory per rack, scaleable to 6TF (teraflops) in a single 44U rack. It has redundant (Dual-Rail) InfiniBand Interconnect Fabric and (Dual-Rail) Ethernet Management Fabric, and additional racks can be added for greater capacity; e.g., 72 racks scale to 442TF with 36,864 cores. Appro can deploy its building-block reference architecture based on the 44U rack cabinet to enable a supercomputing environment with up to 1000TF. The X1 Supercomputer comes with a complete stack of management software for networking, servers, host, scheduling, and reliability, availability, and serviceability (RAS).
Appro Xtreme-X1 Single Cluster
The Xtreme-X1 Supercomputer can also reduce your data center floor space by as much as 30 percent. The key to this is Appro’s Directed Airflow cooling configuration. Designed to maximize cooling efficiency and minimize power consumption, the cooling system’s front access for maintenance allows racks to be mounted back-to-back to take advantage of the under-floor cooling available in most data centers. This delivers the cold air directly to the equipment to be cooled, thereby eliminating the need for aisles between the racks. Additionally, board components are positioned so that processor-heated air flows over the fewest components and exits through the perforated front-access panel. The X1 Supercomputer has three independently controlled cooling fans per node, redundant N+1 power supplies, and is reportedly 90 percent efficient.
The Xtreme-X1 Supercomputer uses the Appro Cluster Engine (ACE) management software, which features a complete remote-access “lights-out” management solution that comes pre-installed with every system. This software suite offers a web-based management interface that is said to be easy-to-use, making it possible to control the Xtreme-X1 from any location. The management modules include Network Management, Server Management, Cluster Management and Storage Management. In addition, ACE supports multiple networking topologies to achieve maximum reliability, performance, and high availability.
With Clusters Added
“With Appro Intel Cluster Ready deployment, Appro customers can now purchase their clusters with the confidence of knowing that the equipment and applications are certified to work together,” said Richard Dracott, Intel’s HPC Division General Manager. “Intel Cluster Ready allows software developers to validate just one time across multiple hardware platforms, speeding their development and time-to-market. This simplifies and accelerates the deployment of high-performance computing solutions — achieving quicker time to results and time to computing.”
Intel Cluster Ready includes a software registration process for compatible HPC applications and a hardware certification process that uses the Intel Cluster Checker. This tool checks the cluster hardware and software components to ensure they interoperate correctly. Intel Cluster Checker also includes fault isolation, helping to improve early detection of cluster problems that can decrease productivity and increase support costs.
“Appro offers best-in-class open supercomputers that improve performance bandwidth, scalability, reliability, manageability and availability packaged with outstanding cluster solution integration,” said Jim Ballew, CTO of Appro. “By integrating Intel multi-core processor technologies and the latest Mellanox 40Gbps InfiniBand ConnectX adapters into Appro Xtreme-X1 Supercomputers, Appro, Intel, and Mellanox provide the most advanced, high-performance computing solutions to help HPC users align their technology more closely with their business goals.”
Results for IT in HPC
“Customers can now build a powerful, easy-to-manage cluster architecture that can dramatically reduce latency while accelerating I/O performance for a variety of enterprise and high-performance computing applications,” Ballew adds.
Now Intel and Mellanox weigh in.
Appro Xtreme-X1 Datacenter
“Appro Xtreme-X1 offers an optimal combination of Intel multi-core technology to enable unprecedented cluster performance and energy-efficiency packaged with Intel advanced cluster tools and Intel Cluster Ready certification,” said Richard Dracott, General Manager, HPC Organization, Digital Enterprise Group, Intel Corporation. “Intel multi-core processors help Appro deploy high-performance computer clusters that provide faster time to results and time to computing.”
Thad Omura, vice president of product marketing at Mellanox Technologies, concurs. “Mellanox’s 40Gbps InfiniBand adapters and switch silicon deliver unparalleled I/O connectivity for latency-sensitive clusters, grids, and virtualized environments. The Appro Xtreme-X1 Supercomputer unified-scale cluster platform with Mellanox 40Gbps InfiniBand connectivity enables applications to achieve unsurpassed efficiency and performance from multi-core processors.”
In an era where budgets for IT are growing ever tighter, Daniel Kim, CEO of Appro, explains that “Appro Xtreme-X1 Intel Cluster Ready-certification helps simplify deployment of high-performance computing clusters achieving faster and more accurate time-to-results. IT departments in all high-performance computing organizations need to accommodate the pressure for shorter lifecycles with a limited budget for software and hardware expenses.”
For more information from the company, go directly to Appro.
Read why DE’s editors chose this as a Pick of the Week.
Access Appro’s Xtreme-X1 main page.
Read the details about Xtreme-X1 + Intel Cluster Ready.
Here are the specifics about the Xtreme-X1 Configuration.
Users may request a quote from Appro.
View ISC videos on First Experiences Building a 40 Gb/s InfiniBand Supercomputer.
Read the White Paper titled “Appro Xtreme-X Supercomputer: Designed for High Availability and Price/Performance.”
Read earlier DE coverage: “New Xtreme-X2 to Advance CFD Research Facility in UK” (April 2008).
Margaret S. Gurney is the editor for new products at Desktop Engineering magazine.