Quick Product Finder Intel Powered Products Clusters   » EcpliseCluster Servers   » Ecplise7Quad   » Ecplise7   » Ecplise7 TWIN   » Ecplise7M   » EcpliseSA Lustre Storage   » EcpliseSL Storage   » EcpliseSA Workstations   » Ecplise7 AMD Powered Products Clusters   » EcpliseCluster Servers   » EcpliseA64   » EcpliseA4   » EcpliseA8   » EcpliseSA Storages   » EcpliseSA   » EcpliseA64 Workstations   » EcpliseA64 More Products Co-Processing Accelerators   » EcpliseGP   » GPGPU Add-On Systems   » EcpliseFP Storage Only   » EcpliseNAS   » HorizonRAID

Intel Powered Products
AMD Powered Products

HPC Solutions
Servers
Storage
Workstations

Datasheets
Request a Quote

Aeon Computing EclipseCluster™ HPC solutions are built with the latest compute, storage, software tools, and fabric technologies to accelerate any application. Each solution is tailored to your specific application, physical requirements, and budgetary needs. In addition to these dynamics, EclipseCluster systems employ the latest methodologies in deploying,managing, and optimizing a HPC resource.

Nodes

The key ingredients to any HPC system are the computational platforms...ie, the nodes. EclipseClusters™ are built with our extensive line of Eclipse™ servers powered by the latest multi-core technology from AMD and Intel. From 4 to 24-cores and 2 to 256GB of RAM per node, there is a optimal node configuration for your application(s).

In addition to multi-core x86-64 CPUs, EclipseClusters can leverage the latest in alternative computing:

• GPGPU - Nvidia Quadro/Tesla & AMD/ATI FireStream
• FPGA - Altera StratixII

Servers

The Eclipse7Quad™ was designed with HPC and high density application computing in mind. Powered by the latest Quad and Six-Core Intel® Xeon® processors, the Eclipse7Quad™ has the computational power of 4 servers in the space of one 2U chassis.

• Powered by Intel® Xeon® 5600 series processors

The Eclipse7™ line of systems provideS breakthrough application performance and new levels of power efficiency to the data center. The Eclipse7™ offers higher performance and less energy consumption compared to previous servers.

• Powered by Intel® Xeon® 5600 Quad or Six-Core processors

The Eclipse7twin™ is two full-featured, independent, high-performance servers in the space of one. Featuring the latest Intel Xeon processors, each server in the Eclipse7Twin™ supports up to 12-cores and 192GB of memory.

• Powered by Intel® Xeon® 5600 Quad and Six-Core CPUs

The Eclipse7M™ is a true enterprise level high-performance 4 to 24-core 4-socket SMP server with support for up to 256GB of 667MHz FB-DIMM memory.

• Powered by Intel® Xeon® MP 7400 Quad and Six-Core CPUs

The EclipseA64™ is ideal for a wide range of intense computing applications such as HPC, databases, rendering, and virtualization to name just a few.

• Powered by Quad and Six-Core AMD Opteron™ 4000 Series Processors

The EclipseA4™ is a 4way multi-core SMP server platform that can accelerate multiple applications for a wide range of budgets. With up to 24 64bit AMD Opteron™ cores.

• Powered by Quad and Six-Core AMD Opteron™ 8000 Series Processors

Server/Storage

The EclipseSA™ is two powerful products in 1 robust, dense solution. It combines a multi-core compute server with a high-performance SATA-II or SAS RAID.

• Powered by Intel® Xeon® 5600 series Quad or Six-Core CPUs or Quad or Six-Core AMD Opteron™ 4000 or 6000 Series CPUs

Alternate Computing

The EclipseGP™ is a fully-integrated GPGPU system with up to 4 nVIDIA Tesla, Quadro FX6000, or GTX580 GPUs. Utilizing the latest DDR5 memory and 128-48 "stream" processors, the EclipseGP has a peak throughput of 1.55Tflops.

With the EclipseFP™ FPGA accelerated solution, an increase in speed, density, and bandwidth can be achieved while lowering power consumption.

• Systems come installed with a XtremeData FPGA-based In-Socket Accelerators™ plus a dual-core or quad-core x86 CPU

Interconnects

Depending on your application, I/O requirements, and/or budget, EclipseClusters™ are available with the latest fabric technology from low latency Infiniband, 10GbE, to Gigabit Ethernet.

InfiniBand

A high-performance low-latency fabric technology designed for parallel computing and/or demanding I/O applications. MPI ping latencies as low as 1.2 microseconds and up to 3.4GBytes/second of uni-directional bandwidth. InfiniBand can scale from a small 8-node cluster to a large 1000+ node multi-site grid supporting different applications.

QLogic InfiniBand HCAs feature the industry's highest message rate. Combined with extremely low MPI latency and very high effective bandwidth, this enables MPI and TCP applications to scale to thousands of nodes with unprecedented price-performance.

The 12000-series 40Gbs (QDR) and 9000-series 20Gbs (DDR) high-performance Infiniband switches can be configured from entry-level 8-node systems up to 1000+ nodes. The 12000 and 9000-series switches are available in 24, 36, 48, 60, 72, 84, 96, 144, and 288-port configurations and optional 10GbE or Fiberchannel bridging.

Mellanox ConnectX™ HCAs deliver low-latency and high-bandwidth for performance-driven server and storage clustering applications in Enterprise Data Centers, High-Performance Computing, and Embedded environments. ConnectX IB simplifies network deployment by consolidating clustering, communications, storage, and management I/O and providing enhanced performance in virtualized server environments.

10Gb and 1Gb Ethernet

Standards-based protocols utilizing cost-efficient on-board gigabit ports to high-performance 10Gb with TCP-offload technology "TOE". Some application are perfectly suited for Gig E, even some MPI apps that don't require extremely low-latency. For a extremely cost-effective MPI solution, an optimized MPI such as Intel™ MPI can be utilized to dramatically improve latency performance using standard gigabit ethernet or 10GbE.

OS/Software

Each and every EclipseCluster ships fully configured and tested with OS, HPC software tools, and EclipseMon management for a truly turn-key solution.

OS Support

EclipseCluster™ HPC systems support the latest Linux open-source and PAYware revisions such as CentOS and RHEL. Contact our engineering staff for recommendations and qualification of specific variants of Linux that may be required for legacy reasons.

Microsoft's Windows Compute Cluster Server 2008 is available as a unique alternative to traditional Linux driven cluster computing. Windows CCS is simple to deploy, operate, and manage. It includes MS MPI, an optimized MPICH2, job scheduler, and compiler tools. Windows CCS supports both gigabit Ethernet and low-latency Infiniband.

HPC Software

EclipseStack™ is a complete Linux based HPC suite of tools imaged and optimized for your specific EclipseCluster™ HPC system. It is the "glue" of the solution. With EclipseStack™, you can compile, schedule jobs, manage user accounts, and manage nodes in parallel.

EclipseStack Features:

• MPICH 2 (use of MPICH 1 requires engineering review)
• MVAPICH 2
• AdaptiveComputing™ Moab™, TORQUE/Maui or SGE job scheduling
• System Imager with full PXE bootimg, imaging and management
• PDSH / C3 parallel shell tool
• Sync-Nodes(tm) user management sync tool
• GNU compilers (optional Intel™ or Portland Group™ compilers)
• AMD or Intel math kernel libraries

Optional supported Software Tools for Maximum Performance:

• AdaptiveComputing™ MOAB Cluster Suite
• Intel Cluster Toolkit: MPI, Trace Analyzer, math kernel libraries
• Intel Compiler Professional Edition
• PGI Compilers

Cluster Monitoring

Observer is a web-based cluster monitoring tool that provides the administrator with an aerial view plus individual node level status of the EclipseCluster™. Utilizing an industry standard IPMIv2 interface, Observer monitors system and I/O load, resource utilization, CPU temp, system temp, and fan speeds of each individual system in an easy-to-read graphical interface. EclipseMon™ also allows the administrator to remotely power cycle a distressed node with the click of an on-screen button.

Additional Local and Remote Monitoring Options:

• Out-of-Band Serial Console access via IP
• KVM and KVM-over-IP
• 1U rackmount slide out LCD display with keyboard and mouse

Storage

Storage is a key component to a efficient HPC solution. It is often overlooked and can be the one single item holding back a system. EclipseCluster HPC systems support a wide range of storage solutions using standard NFS to high-performance Global Parallel File systems. Regardless of the level of complexity or price, there is a solution we can optimize for your applications(s).

EclipseSA™
The EclipseSA™ line of storage servers are very effective storage solutions that also double as the cluster head-node in small-medium sized EclipseClusters. The EclipseSA is a high-performance RAID system combined with a multi-core server

Aeon Computing offers the latest in high-performance global parallel filesystem storage from Lustre and GlusterFS. Depending on your application requirements, these solutions can greatly accelerate application performance, plus offer linear capacity growth. Please contact our Sales Engineering group for details and how a solution may benefit your application.

HorizonRAID™
The HorizonRAID™ line of external direct attached RAID arrays offers the latest in disk technology, such as SATA-II and Serial Attached SCSI (SAS). HorizonRAIDs support multiple host connections with SAS and Fibre-channel as well as redundancy in the way of redundant controllers. Large storage fabrics can be built using Qlogic fibre-channel switching for growth, increased I/O, and multi-path redundancy.