Accelerate into Spring
NVIDIA Maximus with "Kepler" GPUs, Intel Xeon Phi Coprocessors and Large Datasets
Harness newly launched acceleration technologies. It may be time to examine what speedups are available for your projects. Microway's Maximus WhisperStations offer accelerated visualization and simulation now that NVIDIA has launched "Kepler" Quadro cards. See if your application is supported!
Xeon Phi coprocessors are shipping in our servers and compute clusters. Available fully-integrated with Intel's compilers, tools, libraries and MPSS, you can get to work right away.
Storage is growing, too. Whether you need low-cost backups, high-speed database access, or an integrated computational storage server, we provide customizable options. Send us your wishlist and find out what we can design for you!
New NVIDIA Quadro "Kepler" GPUs Provide Superior Performance for Professional Graphics
NVIDIA has launched a number of new Quadro Kepler GPUs for mid-level and entry level professional users. On professional applications like Adobe CS6, these GPUs deliver up to 2X the performance of the previous generation (ex: K4000 vs. Quadro 4000, Adobe Premiere Pro).
NVIDIA Quadro K4000, K2000, K600 GPUs
Quadro K4000: 3GB GDDR5 memory, 768 SMX CUDA Cores and 134GB/sec memory bandwidth in 80W TDP
Quadro K2000: 2GB GDDR5 memory, 384 SMX CUDA Cores and 64GB/sec memory bandwidth in 51W TDP
Quadro K600: 1GB GDDR5 memory, 192 SMX CUDA Cores and 29GB/sec memory bandwidth in 41W TDP
All Quadro Kepler GPUs have DisplayPort 1.2 connectors capable of 3840x2160 resolutions at 60Hz. Multi-monitor support for up to 3 directly connected displays is available for Quadro K4000/K2000/K600 (4 displays with DisplayPort hubs).
Also available is the NVIDIA Quadro K5000 with 4GB GDDR5, 1536 SMX CUDA cores
and support for up to 4 displays directly connected.
NEW: Next Generation Maximus 2.0 WhisperStations Now Shipping
NVIDIA Maximus technology combines the visualization capability of NVIDIA Quadro GPUs with the compute power of NVIDIA Tesla GPUs for amazing speedups of certified applications. Rendering, engineering simulation, design, and other applications benefit from this speed: users gain the ability to design and render/simulate at the same time.
Our Next Generation Maximus systems combine the new Quadro Kepler GPUs with Tesla K20 GPUs for compute offload. Performance far exceeds the previous generation!
NVIDIA Quadro Kepler GPU for visualization; NVIDIA Tesla GPU for Maximus certified compute offload.
1x Xeon E5-1600 or 2x Xeon E5-2600 Series CPUs with 4-8 cores and high clock speeds
16GB-128GB of DDR3-1600Mhz memory, as best suited for each application
Microway HPC Tech Tip
5 Easy First Steps on GPUs - Accelerating Your Applications
Abstract: This week NVIDIA provided a tutorial outlining first steps for GPU acceleration using OpenACC and CUDA. This was offered as part of the "GPUs Accelerating Research" week at Northeastern University and Boston University. After attending, it seemed appropriate to review and boil it down to the essentials. There's no way we can cover everything, but sometimes getting started is the hardest part.
Set aside half an hour, four hours, eight hours; whatever you can spare. The steps below will get your code up and running on GPUs. The resources at the end will be necessary for those seriously digging into acceleration.
New Intel Xeon Phi Events and Programming Resources Available
Intel is adding Xeon Phi programming events and webinars throughout this spring.
Passively cooled Xeon Phi 5110P coprocessors are shipping now. We anticipate shipping actively cooled Xeon Phi coprocessors immediately after launch this year
Free Xeon Phi Book Promo, April 26-May 26
To support the Xeon Phi programmers ordering Microway systems, we are offering a book promo for the next 30 days. Order a Xeon Phi coprocessor or Microway NumberSmasher server configured with Xeon Phi from April 25-May 25, and we'll include the new Intel Xeon Phi Coprocessor High Performance Programming
book for free! Limit one book per customer (a $60 value, US only).
Enterprise Storage Servers Now with 4TB Drives: Over 128TB in 4U!
4TB enterprise-class disks are now available from Seagate, Hitachi, and Toshiba. Microway Storage Servers are now qualified with these drives for increased capacity and density. A typical 4U storage server includes:
128TB usable storage from 36 hot-swappable 4TB drives in RAID 60
Areca, LSI, or 3Ware RAID controllers with high performance, hot-spare support, and multiple SAS connections per backplane
2 Xeon E5-2600 series CPUs, and up to 512GB of memory!
2 fixed SSD OS drives
Optional QDR/FDR InfiniBand uplinks or 10/40 Gigabit Ethernet
We're happy to customize a storage quote to meet your needs. Let us know your desired capacity, footprint, or performance requests and we'll configure an affordable solution that meets your requirements.
GPU Computing Accelerating Innovation, Research and Discovery Event Report
Northeastern University and Boston University hosted an event for Boston-based GPU computing end-users April 24-25 in cooperation with NVIDIA. Day 1 included sessions and invited talks from local CUDA Centers of Excellence (Northeastern, Harvard University, Boston University), medical customers (MGH/Partners), and commercial users. Day 2 included a hands-on CUDA programming workshop.
Over 60 GPU computing experts attended on Day 1, and NVIDIA presented sessions on the Role of GPUs in Research Computing (Kim Powell, included future hardware roadmap) and the Future of CUDA (Mark Ebersole, included new programming updates). Microway supported the event with a WhisperStation- Tesla and staff presence. We'll forward along presentation links as they become available. Below are a few highlights of presentations by institutions with Microway hardware:
Valar: A Benchmark Suite to Study the Dynamic Behavior of Heterogeneous Systems - Northeastern University, Perhaad Mistry
Valar provides a diverse set of heterogeneous computing benchmarks/workloads that help users compare the strengths and weaknesses of various compute architectures. These workloads help demonstrate combined system performance (CPU+GPU or combined APU) in a way that realistically simulates the type of well coded workloads end-users might run. Helps highlight the nuances of your particular hardware.
Development of a GPU-based Monte Carlo dose calculation package for proton radiotherapy - MGH, Jan Schuemann
A very early adopter of GPUs, MGH uses NVIDIA Tesla to help calibrate the least damaging dosages for patient treatment with proton radiotherapy. Leveraging a Monte Carlo simulation model, the gPMC project, and other GPU accelerated functions/packages, the team at MGH can calculates the appropriate dosage and aims protons accordingly while the patient is on the treatment table. They consequently minimize the damage to healthy tissue. These calculations would take hours on a traditional cluster, and therefore would not be feasible to perform at all without GPU technology.
Faster 3D CT Reconstruction using CUDA and OpenCL - Northeastern University Reconfigurable and GPU Computing Laboratory, Saoni Mukherjee
The team at NEU evaluated a number of parallel computing standards(C/C++, OpenMP, OpenCL, CUDA) and architectures (multicore CPU, Radeon GPU, Tesla GPU) to help speed up 3D CT Reconstruction. They analyzed their compute job consisting of weighting, filtering, and backprojection and determined backprojection was best candidate for speedup with massive parallelism. GPU acceleration with CUDA on Tesla C2070 most effectively sped up this portion of their code. The group hopes to improve their ability to work with larger datasets in the future.
Microway HPC Tech Tip
Hadoop isn't just for Web 2.0 Big Data Anymore. Hadoop is for HPC.
Abstract: In 2004 Google released a white paper on their use of the MapReduce framework to perform fast and reliable executions of similar processes / data transformations & queries at terabyte scale. Yahoo then began the Hadoop project to support their search product. As a result of this, Apache elevated Hadoop, their MapReduce and DFS (distributed file management system) initiative out of Nutch, their open source search project.
What about Hadoop for High Performance Computing with scientific applications? It certainly has its place and a basic understanding of Hadoop helps you to understand where you can take advantage of Hadoop in HPC.
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.
Opt Out of Future Email