Sprinx HPC - Hybrid GPU clusters

Hybrid GPU Clusters

Products and services > Hybrid GPU clusters

Supercharge your cluster with Tesla CUDA-based GPUs

The Sprinx Systems Tesla Preconfigured Clusters deliver supercomputing performance at a lower power, lower cost, and using much fewer systems than standard CPU-only clusters. These clusters are powered by the NVIDIA® Tesla™ S1070 Computing System, a four-teraflop 1U system that has the Tesla T10 GPUs based on the CUDA massively parallel architecture. With 4 GB of memory per GPU, support for IEEE 754 single & double floating point precision, and a fast 102 GB/sec GDDR3 interface to the memory, the S1070 1U 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 at research institutes, universities, and enterprises – it powers the 29th most powerful supercomputer in the world.

Sprinx Systems offers several configurations of the Tesla preconfigured clusters and we can customize them to your needs:

Configurations	32 TeraFlop Tesla Cluster	16 TeraFlop Tesla Cluster
GPUs	32 Tesla T10 GPUs	16 Tesla T10 GPUs
CPU nodes	32 Intel Xeon E5550 Quad-core 2.66 GHz	16 Intel Xeon E5550 Quad-core 2.66 GHz
Memory	128 GB GPU memory Up to 1 TB CPU memory	64 GB GPU memory Up to 512 GB CPU memory
Storage	Up to 96 TB SATA drives	Up to 48 TB SATA drives
Network	Gigabit Ethernet switch Infiniband switch	Gigabit Ethernet switch Infiniband switch
Infrastructure	Rack, PDU, LCD+keyboard	Rack, PDU, LCD+keyboard

Installed Software

The Sprinx Systems Tesla preconfigured cluster comes with preinstalled with

- SUSE Linux Enterprise Server 11, 64-bit

- CUDA 3.1 drivers, toolkit and SDK

- Clustering 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, and so on. 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.

More Options

Please contact HPC division leader to find out more about customizing these clusters to your needs. Also, don’t forget to ask about our special cluster configuration to enter the Top 500 Supercomputing List with Hybrid GPU Cluster: 2 x 42U racks & network infrastructure.

Hybrid Cluster Technology

NVIDIA TESLA S1070 1U Computing System
1U Twin™ SuperServer - Dual Socket nodes

Infiniband MT25204 20Gbps per node
Dual Gigabit Ethernet per node

NVIDIA TESLA S1070

With the world’s first teraflop many-core processor, the NVIDIA® Tesla™ S1070 computing system speeds the transition to energy-efficient parallel computing. With 960 processor cores and a standard C compiler that simplifies application development, Tesla S1070 scales to solve the world’s most important computing challenges—more quickly and accurately.

1U TWIN™ SERVERS

1U Twin™ SuperServers feature two Dual Processor server nodes in a single chassis and 90%+ high-efficiency power supply.This platform is an excellent choice for high-performance computing (HPC) clusters, server farms and other datacenters where space, cost, energy-efficiency and density are high priorities. SuperServers support up to four hot-swap drives (2 per node) and up to 96GB memory (48GB per server node). Each node also features a full-bandwidth PCI-Express x16 slot for high-performance expansion card support.