Microway - Technology you can count on, since 1982

sales@microway.com
Phone: 508-746-7341
Fax: 508-746-4678

Microway (R) Technology you can count on.

Faster and Greener - Microway NVIDIA^® Tesla^™ Powered High Density Clusters

Supercharge your next cluster with Tesla GPUs

Microway's Tesla Preconfigured Clusters deliver supercomputing performance at a lower power, lower cost, and using many fewer systems than standard CPU-only clusters. These clusters are powered by NVIDIA Tesla S2075/S2090, 1U systems that have the Tesla T20 GPUs based on the CUDA massively parallel architecture. With 6 GB of memory per GPU, up to 512 cores per GPU, support for IEEE 754 single & double floating point precision, and a fast GDDR5 interface to the memory, the S2075/S2090 speeds the transition to energy-efficient parallel computing and scales to solve the world's most important computing challenges more quickly and accurately. Successfully deployed in demanding applications at research institutes, universities, and enterprises - it powers the second most powerful supercomputer worldwide.

Installed Software
The Microway Tesla preconfigured cluster comes with preinstalled with:

Linux distribution of your choice, including Red Hat, SUSE, Debian, Fedora and Ubuntu
CUDA Toolkit and SDK
Microway Cluster Management Software (MCMS) integrates with optional MPI Link-Checker^™ Fabric Validation Suite and InfiniScope^™ Real-Time Network Diagnostics.
Optional Bright Cluster Manager software

CUDA Applications
The CUDA-based Tesla GPUs give speedups of up to 250x on applications ranging from MATLAB to computational fluid dynamics, molecular dynamics, quantum chemistry, imaging, signal processing, bio-informatics. Click here to learn more about these speedups with links to application downloads.

GPU Software Development Tools
Click here to find out more about the C-based software development tools and various libraries for GPUs.

NVIDIA S2075/S2090 GPU Compute System

Add GPU compute capability to your existing HPC infrastructure

Each Tesla 1U GPU Compute System Provides

Four Tesla Fermi T20 GPUs in a 1U chassis
2.06 double precision and 4.12 TFLOPS single precision GPU performance
S2090 with additional cores, 2.66 TFLOPS double precision, and 5.32 TFLOPS single precision performance
Total of 24GB GDDR5 GPU memory with ECC function
Connection to existing host server(s) with 2x PCI-E x16 2.0 host adapters (optional x8 with reduced performance)

Microway OctoPuter™ with 8 Tesla GPUs

Achieve high GPU/CPU ratio with OctoPuter-Tesla 4U System

Each OctoPuter has:

8 NVIDIA Tesla GPUs—Over 8 TFLOPS
2 Quad- or Six-Core Intel^® Xeon Nehalem/Westmere CPUs
Up to 288GB DDR3 memory

Applications: complex modeling, simulations and analysis.

Microway BioStack™ -LS — Scalable Personal Cluster

Implement Tesla Fermi GPUs for supercomputer performance: 84 CPU Cores and 6272 GPU cores in 9U

For more information see BioStack dedicated product page

Each BioStack has:

Total 7.21 TFLOPS double precision GPU performance (14.42TFLOPS single)
Seven GPU Compute Nodes: each with 2 Tesla Fermi M2075 448 core GPUs and 12 CPU cores
Intel Xeon CPU Cores at up to 3.06Ghz
Microway 36 port FasTree™ InfiniBand switch
Compact 9U Cabinet for excellent scalability

Available NVIDIA Bio Workbench Applications including: ACE MD, AMBER, BigDFT, GROMACS, GROMOS, HOOMD, LAMMPS, NAMD, TeraChem (QC), VMD, CUDA-BLASTP, CUDA-EC, CUDA-MEME, CUDASW++, GPU-HMMER, and MUMmerGPU

Sample Microway Tesla Cluster Specifications

Configurations	8 Tesla S2075/S2090 NumberSmasher Cluster with 2 GPUs per Server	8 Tesla S2075/S2090 NumberSmasher Cluster with 4 GPUs per Server	9U Compact OctoPuter GPU Cluster with 8 GPUs per Server	Microway BioStack 9U Scalable Personal Cluster
GPUs	32 Tesla T20 GPUs		16 Tesla T20 GPUs (8 Teslas per 4U OctoPuter)	14 Tesla T20 GPUs (2 Teslas per 1U GPU Node)
CPU Compute Nodes	16 Dual Xeon Compute Nodes with up to 192GB Memory	8 Dual Xeon Compute Nodes with up to 288GB Memory	2 Dual Xeon Compute Nodes with up to 288GB Memory	6 Dual Opteron Compute Nodes with up to 256GB Memory
Master Node	Dual Xeon Server with up to 144GB Memory, Optional GeForce/Quadro Graphics and RAID		N/A	1 Quad Opteron Compute Node with up to 512GB Memory
Memory	192 GB GPU memory Up to 3072 GB CPU memory	192 GB GPU memory Up to 2304 GB CPU memory	96 GB GPU memory Up to 576 GB CPU memory	84 GB GPU memory Up to 1280 GB CPU memory
Storage	Head Node: up to 64TB Compute Nodes: up to 48 TB	Head Node: up to 64TB Compute Nodes: up to 32 TB	Compute Nodes: up to 2TB each	Head Node: up to 6TB Compute Nodes: up to 6TB each
Network	Optional ConnectX-2 QDR InfiniBand Interconnect Gigabit or 10G Ethernet		Gigabit Ethernet	ConnectX-2 QDR InfiniBand Interconnect included Gigabit Ethernet
Cabinet and Cooling	Microway CoolRack Cabinet includes: 535-850CFM fans for superior cooling of high-density clusters		Microway 9U Compact Cabinet	Microway 9U Compact Cabinet
Green Features	High-efficiency (Gold-Level) Power Supplies Cooling Paradigms Designed to Reduce Power Consumption Optional Low-Power Intel or AMD Processors Integrated Software Reduces Power Consumption on Idle Cores		High-efficiency Power Supplies Cooling Designed to Reduce Consumption Optional Low-Power Intel Processors Software Reduces Idle Core Consumption	High-efficiency (Gold-Level) Power Supplies Cooling Paradigms Designed to Reduce Power Consumption Optional AMD Processors

Microway CUDA Cluster
Up to 48 TFLOPs performance in 24U

More Options
Please contact our Technical Sales Department to learn more about customizing Microway's GPU Compute Clusters to your needs. Also, don't forget to ask about our special cluster configuration to enter the Top 500 Supercomputing List.

sales@microway.com
Phone: 508-746-7341
Request a Quote

Privacy Legal Our mission is to provide customers with leading edge technologies for high performance computing solutions (HPC). We establish and maintain industry recognized products and expertise for beowolf cluster interconnect (InfiniBand), clustre management and HPC storage solutions.