Appro - High-performance Workstations, Servers, Clusters and Storage Solution

High-Performance and Enterprise Computing

Contact  |  Buy Online  |  Shopping Cart






Overview

  Supercomputers

  Blade Servers

  Servers

  Workstations

  Storage

  Compute On Demand


Appro Xtreme-X1 Supercomputer

 

 

 

 

 

Q&A | link

 

  System Hardware Configuration

Nodes

64 blade nodes per rack cabinet

CPU

64-bit Quad-Core Intel Xeon Processors: 128 sockets/512 cores per rack cabinet
Cache
12MB L2 cache per processor
FLOPs
6.1 TF Peak Performance per Cabinet
Main Memory
Up to 4TB Fully-buffered ECC DDR2 SDRAM per rack
Interconnect
Dual-Rail redundant Fat-Tree
External I/O Interface
Gigabit Ethernet
10 Gigabit Ethernet
DDR Infiniband with ConnectX Host Channel Adapters
Local Disk Storage
Two 2.5" SATA II disk drives per blade
File System
Linux EXT3 for /Root and /Temp
Panasas Active Storage available ad Global File System
Reliability Features (Hardware)
Full ECC Memory Protection, Redundant Power Supplies, Redundant cooling fans, Redundant Management Servers, Redundant Management Network, Redundant Infiniband Network, Internal RAID1 Disk Capability.
Reliability Features (Software)
Redundant Management Servers with software support for failover, Redundant I/O Servers with software support for failover, Redundant Management Networks with software support for message retry and failover, Redundant Infiniband Networks with software support for load balancing, message retry, and failover, Stateless Management allows immediate Recovery. (more detail specs are below)
Operating System
RedHat AS Version 5
Message Passing Libraries
MVAPICH
Compliers
Intel Cluster Toolkit Compiler Edition (Try and Buy evaluation copy)
Power
28.8KW per Cabinet.
Input: Dual 208VAC 40A 3Phase
Connectors: Hubbell CS8365C
Cooling Requirements
Air Cooled
Required Air Flow: 3000 cfm (1.41 m3/s)
Intake: Floor
Exhaust: Front of Cabinet
Cabinet Dimensions
(H x W x D)
81.08 in. (2059.4 mm) x 24.0 in. (609.6 mm) x 46.47 in. (1180.4 mm)
Cabinet Weight
1,650 Pounds (748.4kg) per cabinet
Acoustic Noise
TBD
Regulatory Compliance
FCC Class A, CISPR 22, CE (EMC) : EN55022 (EMI), EN55024 (EMS), EN61000-3-2 (Harmonic), EN61000-3-3 (Flicker), MIC : KN22 (EMI), KN24 (EMS)
Safety
UL 60950-1, CAN/CSA – C 22.2 No 60950–1, CE (LVD) : EN60950, CB : IEC 60950, RoHS, WEEE

 

  System Software Management Capability

ACE
Appro Cluster Engine™ (ACE) Management System provides complete lights-out remote management capability with web based interface
System Management
Management of overall system configuration
Supports redundant management servers with automatic failover
Designed to anticipate and tolerate failures
Supports enterprise level availability requirements with 24 hr MTTR
Network Management
Automatic discovery of interconnect hardware
Supports multiple interconnect fabric topologies
Supports redundant paths and networks
Supports load balancing and failover
Provides network status to the management system
Server Management
Automatic discovery of server hardware
Remote server control (Power On/Off, Cycle)
Remote server Initialization (Reset, Reboot, Shut Down)
Support for scalable fast diskless or dataless booting for large node count systems
Supports server redundancy and failover
Provides server status to the management system
Cluster Management
Supports partitioning a cluster into multiple logical computers
Maps logical computers (clusters) onto servers (nodes)
Supports multiple independent OS configurations
Manages and monitors logical computer (clusters) status
Provides cluster status to management system
Integrated job scheduling and management
Manages and monitors operating system instances (nodes)
Provides node status to management system
Storage Management
Supports scalable root file systems for diskless or dataless nodes
Supports multiple global storage configurations
Supports high BW to secondary storage for data and check pointing
Provides server status to the management system
GUI
Web based
Reliable, Available, Serviceable (RAS)
Redundant management network (GbE and 10GbE) with failover
Redundant high bandwidth network (InfiniBand™) with failover
Redundant management servers (two levels) with failover
Redundant root file system with failover
Built-In multi-generation configuration management for software
Redundant storage via RAID
 
Appro Reference Architecture Designs
Reference Architecture Designs
The Appro Reference Architecture Designs are based on a 44U Rack Cabinet.
1 Rack scale to 6TF with 512 cores
2 Racks scale 12TF with 1,024 cores
4 Racks scale to 25TF with 2,048 cores
8 Racks scale to 49TF with 4,096 cores
18 racks scale to 111TF with 9,096 cores
36 racks scale to 221TF with 18,432 cores
72 Racks scale to 442TF with 36,864 cores
Closed-Loop Services and Solution Management
Support and Services
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.

Copyright © 2007 Appro International, Inc. All Rights Reserved.