 |
ClusterMax Apex | CPU-Intensive
This high performance supercomputer cluster is ideal for large-scale deployments that integrate the latest CPU-based computing technologies and require maximum availability and extreme scalability.
 |
|
 |
ClusterMax Eco | Power Efficient
To significantly reduce datacenter energy consumption, the ClusterMax Eco utilizes 94%+ efficient power supplies. It delivers the highest performance per watt, per dollar and per square foot compared to competing solutions.
|
|
 |
ClusterMax Mobile | Compact Personal Cluster
ClusterMax Mobile combines the immense power and performance capabilities of an HPC with the mobility and space-savings of a workstation to capitalize on greater levels of performance, flexibility and scalability.
|
|
 |
ClusterMax SuperG | GPU-Intensive
The SuperG offers an optimized deployment for applications that require high performance GPGPU integration. It includes support for NVIDIA® Tesla™ 20-series GPU.
|
|
 |
ClusterMax Density Blade | Ultra-Dense HPC Cluster
Delivers a modular, flexible and energy-efficient cluster architecture that addresses today’s data center needs for high performance, energy efficiency, and high-density computing applications
|
|
 |
ClusterMax Visual-X | GPU Visualization
This unique high computing solution supports collaborative visualization environments. It dramatically increases simulation performance and 3D design productivity by reducing project render times.
|
|
 |
ClusterMax Stor-X | Massive Storage
ClusterMax Stor-X is built for petabyte-scale HPC storage deployments requiring extreme performance and includes robust parallel file storage system management features.
|
|
 |
ClusterMax G-Cloud | Cloud 3D Modeling
ClusterMax G-Cloud is the first Web services platform that enables users to interact remotely with complex, 3D models and environments, from any perspective and under customizable lighting conditions.
|
|
|
AMAX HPC Management Suite |
| |
| |
Easiest HPC Management Product
AMAX's HPC Management Suite makes it easy for you to harness the power, flexibility, and scalability of your HPC cluster. Whether you have a small or large cluster, AMAX's HPC Management Suite quickly gets your cluster up and running. It simplifies the application integration process so that you can focus on your work, instead of managing your cluster.
Most Complete HPC Management Product
Other HPC cluster solutions combine multiple tools and interfaces, which are not integrated, certified, or tested together. AMAX's HPC Management Suite is the industry's easiest and most complete management product. It includes a mature, robust set of management capabilities that are accessible through a unified portal interface. Backed by the industry's best customer support, AMAX's HPC Management Suite incorporates nearly two decades of product and technology leadership.
Key Benefits
Unified Web Portal
- Easy to submit, monitor & manage jobs
- Remote support
- Customizable interface
- Job monitoring and control
- Job status notification
- Simplified application deployment
- Easily package software as kits
- Integrated commercial application support
- Increased security
- Restrict access to servers
- Restrict access to data
Effective Cluster Management -- A better way to deploy & manage HPC clusters
- Rapid provisioning
- Manage HPC clusters with unprecedented ease
- Ensure maximum uptime
- Designed to manage both small & large clusters
Powerful Workload Management, based on AMAX LSF
- The most powerful, comprehensive, policy driven workload manager
- Very easy to use and maintain
- Extremely flexible and customizable
GPU Scheduling via Unified Web Portal
Quickly deploy workload to GPU resources
- Unified web portal makes in simple to submit jobs to GPUs
- Remotely view the status of your jobs
Easily manage heterogeneous clusters
- Install CUDA across a cluster is a couple of clicks
- Deploy and manage both CPU & GPU resources in the same cluster
Powerful Application Support -- Harness the power of your cluster to optimize application performance
- Fast MPI
- Make it easy to get your parallel applications up and running
- Minimize job submission errors
- Application Templates
- Simplify job submission
- Reduce setup time
- Minimize operation errors
- Workload Smart, Resource Aware
- Improved application performance
- Optimal resource utilization
- High Performance
- Ensure that your cluster will deliver the best performance without manual tuning
Robust Commercial MPI
In order to make it easier for you to get your parallelapplications up and running, the industry ' s most robust and highest performing MPI implementation. AMAX MPI comes integrated with AMAX's HPC Management Suite. AMAX MPI provides consistent performance at application run-time and forapplication scaling, resulting in top performance resultsacross a range of third-party benchmarks.
Flexible OS Provisioning - Dynamically changes the OS based on workload
- Linux & Windows operating system multi-boot
- Based on resource policies to simplify management
- Dynamic; transparent to end users
- Higher cluster utilization for lower TCO
Workload Monitoring & Reporting -- Operational dashboard
- Comprehensive reports on workload & system load
- Proactively manage compute resources
- Drill-in/out capability for operational insight
- Troubleshoot job and cluster issues
|
Infiniband Technology
Because of its low latency and high bandwidth features, Infiniband has become the de facto interconnect standard for HPC clusters. As a result, Infiniband fabric is one of the key building blocks for the AMAX ClusterMax Series clusters and provides IB QDR/DDR products as either standard or optional interconnect solutions. AMAX works closely with leading Infiniband technology developers, such as Mellanox and QLogic, to bring our customers the latest Infiniband technologies.
Key Features of IB QDR HCA
- Virtual Protocol Interconnect
- 1us MPI ping latency
- Selectable 10, 20, or 40Gb/s InfiniBand or 10GigE per port
- Single- and Dual-Port options available
- IBTA version 1.2.1 compatible
- PCI Express 2.0 (up to 5GT/s)
- CPU offload of transport operations
- CORE-Direct application offload
- GPU communication acceleration
- End-to-end QoS and congestion control
- Hardware-based I/O virtualization
- TCP/UDP/IP stateless offload
- Fibre Channel encapsulation (FCoIB or FCoE)
- RoHS-R6
Key Features of IB QDR Switches
- 2.88Tb/s switching capacity
- Signal optimization for longer cable length
- Quality of Service enforcement
- Temperature sensors and voltage monitors
- Reversible air flow
- Redundant power supplies
- Replaceable fan drawers
- Fan speed controlled by management software
- IBTA Specification 1.2.1 compliant
- Integrated subnet manager agent
- Adaptive routing- Congestion control
- 256 to 4Kbyte MTU
- 9 virtual lanes: 8 data + 1 management
- 48K entry linear forwarding data base
- Port Mirroring
Switch Management
- Fast and efficient fabric bring-up
- Fabric-wide bandwidth verification
- Comprehensive chassis management
- Mellanox API for 3rd party integration
- Intuitive CLI and GUI for easy access
Key Features of Infiniband Fabric Management
Efficient Performance
- Industry's lowest end-to-end latency
- World-record message rate performance
- The only solution with true liner scale of core count
Efficient Management
- Automation to accelerate installations and upgrades
- Detection and diagnosis of fabric issues in seconds
- Initialization of 2,000 node clusters in seconds
Efficient Capacity
- 50-percent higher node count in same footprint
- Up to 67 percent improvement in bandwidth utilization
- Better performance with less hardware
Fabric Manager - Offers administrative functions for subnet, InfiniBand fabric, and individual component management through HTML and a Java™-based console.
Advanced Fabric Services - Delivers efficient fabric performance with industry-leading technologies to automatically tune and optimize an organization's network.
FastFabric Toolset - Includes easy-to-use tools that ensure effortless installation, configuration, and verification of the cluster network and virtual gateway I/O resources.
Fabric Viewer - Displays the Fabric Manager subnet management facilities using a Java-based, stand-alone GUI.
Host Services - Provides a wide range of drivers and installers for products from all adapter vendors, as well as optimized MPI user tools.
1. Application Analysis
The AMAX HPC team is a group of well-trained IT consultants with strong knowledge in high performance computing. By working with customers from every industry and market segment for over 30 years, our sales engineers have gained extensive experience with many different HPC applications. They will provide a thorough analysis of each customer application and walk them through our comprehensive HPC solutions.
2. Architecture Design
AMAX houses a dedicated HPC engineering team that consists of HPC solution architects, OS and application experts, networking experts, as well as manufacturing and industrial experts. Based on the analysis of the application, our engineers will design a custom HPC cluster architecture based on your unique requirements, from application, middleware, operating system, and node configuration, to interconnect and storage. Our engineers leverage the latest CPU, GPU and networking technologies to design your solution with maximum throughput, optimized performance and the best power and cooling efficiency.
3. Evaluation and Test Drive
Customers working with AMAX can enjoy guaranteed satisfaction by participating in our System Evaluation Program and Cluster Test Drive Program. We utilize an advance, in-house HPC cluster as a test bed for the HPC community. HPC users can create an account and remotely log-in to run their codes for proof-of-concept before purchase. Our Technology Partners can also utilize on our HPC testing environment for their research projects.
4. Production and Performance Tuning
Combining over 30 years of engineering expertise and an ISO 9001 certified facility, AMAX builds turnkey HPC clusters with today's highest quality standards. From assembly and cabling to software image loading, our production engineers strictly follow a well-designed manufacturing process instruction (MPI). We have comprehensive cluster testing and performance tuning methodologies, which pay close attention to every single detail. From patching kernels, installing libraries, provisioning nodes, and writing shell scripts to setting up user environment modules, we do the "Whole Nine Yards".
5. Service and Support
AMAX ships turnkey HPC clusters with complete rack cabinets, comprehensive documentation and user manuals. Customers seeking help can also rely on our professional engineering support team with strong troubleshooting and problem solving skills. From repairing a single node to performing diagnostics of an entire cluster, AMAX will take care of everything with quick responsiveness. We provide both domestic and global on-site support.
AMAX works closely with industry-leading ISVs to incorporate the latest software stacks, including operating systems, middleware, and cluster-provisioning software to AMAX’s cluster solutions, to deliver the highest levels of performance and scalability with proven application compatibility.
Below is a list of software stacks we incorporate with our cluster solutions.
|
| Operating System |
- Fedora
- CentOS
- Scientific Linux
- Red Hat Enterprise Linux
- SuSE Enterprise Linux
- Microsoft Windows Server 2008
|
|
| MPI Library |
- Intel® MPI Library
- Platform MPI
- OpenMPI
- MPICH
- MPICH2
- MVAPICH
- MVAPICH2
- MPICH-MX
- MPICH2-MX
|
|
| Cluster Management |
- Bright Computing Cluster Manager
- Clustercorp Rocks+
- Microsoft Windows HPC Server 2008
- Platform Cluster Manager
|
|
| Scheduler/Resource Manager |
- Platform LSF
- Moab
- Torque
- Sun Grid Engine (SGE)
- Maui
- ScaleMP
|
|
| Complier |
- Intel® C/C++ Compiler
- Intel® C/C++ Composer
- Intel® Fortran Compiler
- PGI
- GNU
- Open64
|
|
| Tools |
- CUDA Toolkit SDK
- Intel Cluster Ready Checker
- Intel® Cluster Studio
- Intel® OpenCL SDK
- Intel® Parallel Studio
- Intel® Parallel Studio XE
|
|
|
|
