Revolutionary Performance for 2013
Tesla K20 GPUs and Xeon Phi Coprocessors Are Here!
Happy New Year! NVIDIA (Tesla "Kepler" K20 GPU Accelerators), Intel (Xeon Phi Coprocessors), and AMD (Opteron 6300 Series CPUs) have each made major product releases recently. All are available for Microway's fully-integrated clusters, servers and WhisperStations. Whether you need a budgetary quote, immediate delivery or technical advice, we are here for you.
NVIDIA Tesla "Kepler" K20 and K20X GPU Accelerators
NVIDIA's new Tesla GPUs based upon the "Kepler" architecture are now shipping. The Tesla K20 GPUs come in both active and passive cooling variants. Additionally, Tesla K20X is the passively cooled performance model designed for rackmount servers only.
Tesla K20 and K20X are huge upgrades over Fermi solutions with impressive new HPC features. Shipping today!
NVIDIA Tesla K20
- 2496 CUDA cores provide 1.17 TFLOPS double- and 3.52 TFLOPS single-precision floating point performance
- 5GB GDDR5 memory with optional ECC provides memory bandwidth of 208 GB/sec
- Matches the thermal envelope of Tesla C2075 cards: 225W TDP
NVIDIA Tesla K20X
- 2688 CUDA cores provide 1.33 TFLOPS double- and 3.95 TFLOPS single-precision floating point performance
- 6GB GDDR5 memory with optional ECC provides memory bandwidth of 250 GB/sec
- 235W TDP and passive heatsink for specially designed rackmount servers
January GTC Pass Promotion for US Tesla K20 Customers
Purchase a Tesla K20 or K20X enabled solution before January 27, and you'll receive a FREE full conference pass to the GPU Technology Conference in March! Offer available to US customers only
Test the Power of Tesla K20 on Microway's Benchmark Cluster
Our Tesla GPU benchmark cluster has been updated with Tesla K20 GPU Accelerators. If you are interested in verified hardware performance, we're happy to let you remotely login and test preinstalled applications or your own custom code.
Test Drive Cluster Specifications
- Four Microway NumberSmasher 1U GPU Nodes
- Two NVIDIA Tesla K20 GPUs per node (NVIDIA Tesla K10 available on request)
- Two Intel Xeon E5-2600 series "Sandy-Bridge" CPUs in each node
- GPU accelerated versions of AMBER, NAMD, GROMACS, Quantum Espresso preinstalled along with CUDA 5.0
- Over 28 TFLOPS Single and 9.36 TFLOPS Double Precision GPU performance
- QDR InfiniBand HCAs and switching
Intel Xeon Phi Coprocessors Released!
We're excited to say that Intel's MIC architecture, a many-core x86 coprocessor, is launching as the Xeon Phi 5110P Coprocessor. Explore the possibilities of a single x86 code base and standard parallel based programming.
Passively cooled Xeon Phi 5110P coprocessors will be shipping soon.
- 60 cores @ 1.053 GHz in a ring architecture
- Over 1 TFLOP double precision floating point performance
- 8 GB GDDR5 memory and 320 GB/sec of memory bandwidth
- 225W TDP and passive cooling for Xeon 5110P
- Compatible with most of our rackmount chassis already designed for GPUs
Intel will also debut an actively cooled model for workstations in Q1 2013. Those considering a Xeon Phi 5110P purchase are urged to place an order soon, as the new cards will be delivered on a first-come, first-served basis.
Intel Makes Many Xeon Phi Software Support Announcements
Microway HPC Tech Tip
Software Support for Intel Xeon Phi Coprocessors
Abstract: With Intel's release of the Xeon Phi coprocessor cards, HPC users must ask themselves how much performance they need and how they plan to achieve it. Will resources be devoted towards fast new hardware, re-writing/optimizing software or some balance of the two?
The resurgence of coprocessor and acceleration cards over the last few years has caused many HPC users to re-evaluate their software applications. Supporting a hardware architecture other than traditional x86 CPUs takes some amount of effort. Intel hopes that their Xeon Phi coprocessors, which use x86 cores, will be very attractive.
AMD Opteron 6300 Series Processors (Piledriver Core Architecture)
AMD's latest CPUs offer up to a 15% performance gain over the Opteron 6200 Series. Now available in our 2- and 4-socket Navion WhisperStations, servers and clusters:
- Higher clock speeds than Opteron 6200 Series
- Up to 64 cores per server with 8, 12, or 16 processor cores per socket
- DDR3-1866 memory support in select 2P/4P rackmount configurations
- New support for FMA3, F16c, BMI, and TBM instructions
AMD Opteron systems are an excellent price-performance story for HPC users with parallel codes. A 1U SMP server/compute node with 32 processor cores, 64GB memory and two 1TB enterprise drives is just $4,099! 64 cores for $6,100!
Accelerate your research!
Take advantage of our expertise: we can recommend the most effective compute solutions, help you plan a painless installation, and deliver fully-tested/fully-integrated clusters, servers and workstations.
Eliot Eshelman at 508-732-5534
Brett Newman at 508-732-5542
Ted Fitts at 508-732-5516
GSA Schedule GS-35F-0431N