We're fresh back from the Life Sciences Technology Summit, the AMD Fusion Developers Summit, and we've launched a new HPC Tech Tip feature. Read more:
New Feature: Microway HPC Tech Tips
Tip #1: Take Care When Updating Your Cluster
Although modern Linux distributions have made it very easy to keep your software up-to-date, there are some pitfalls you might encounter when managing your compute cluster.
Cluster software packages are usually not managed from the same software repository as the standard Linux packages, so the updater will unknowingly break compatibility. In particular, changing the version of the Linux kernel on your cluster may require manual re-configuration - particularly for systems with large storage, InfiniBand and/or GPU compute processor components. These types of systems usually require that kernel modules or other packages be recompiled against the new kernel.
Please keep in mind that updating the software on your cluster may break existing functionality, so don't update just for the sake of updating! Plan an update schedule and notify users in case there is downtime from unexpected snags. You may always contact Microway technical support before you update to find out what problems you should expect from running a software update.
In our next issue: Linux software RAID monitoring and management
Microway Earns Attendee Choice Award for Best New Technology at Life Sciences Technology Summit
Microway HPC Sales Specialist Brett Newman was a guest on The Pulse Network's ITEC This Week for a video interview. This 20 minute program includes information about Microway GPU Computing solutions with NVIDIA Tesla GPUs, GPU computing performance, and comments on the Life Sciences Technology Summit.
Microway sponsored the Life Sciences Technology Insights Summit in Boston on 6/14-6/15. Attendees included 50 Life Sciences IT decision makers. Two Microway customers presented case studies at the event:
Dr. Preston Moore of University of the Sciences in Philadelphia highlighted his GPU accelerated research in a presentation on: Using NVIDIA Tesla GPUs to Understand Physicochemical Properties of
Dr. John Wolfgang of Massachusetts General Hospital demonstrated his success with Microway and NVIDIA Tesla solutions: SHADIE: A Quantitative, GPU-based Volume Rendering Solution for Radiation Oncology
Microway and NVIDIA Tesla GPU computing solutions were awarded the Attendee Choice Award for the Best New Technology by a ballot of the conference participants. Microway was also nominated for the Attendee Choice Award for Best Case Study.
AMD’s Announcement of a New APU/GPU Architecture for 2012:
New architecture represents a shift for AMD from a VLIW model similar to Intel’s MIC/Knight’s platform towards a SMT approach like NVIDIA’s Fermi platform
Integration of SMT, SIMD, MIMD technologies into new “Compute Unit” with 4 scalar pipelines each
New GPUs will start with current equivalent CU counts of 24 and may scale up to 100 CUs
2-level read/write cache with APUs sharing L2 between CPU and GPU
Support for x86/x64 Virtual Memory including memory page locking and page faults
C++ programmability features including virtual functions in GPU code
All future AMD GPUs will support double precision, and the ratio of double to single issue rates will vary from 1:16 to 1:4 to 1:2 by product
Microway also sponsored the HPC + GPU Meetup group of Seattle held concurrently with the Summit. Microway provided the door prize of an AMD Radeon 6870 at the meetup.
More OpenCL Programming Resources
AMD's OpenCL Forums
Developers should read and post OpenCL questions, coding tips, and other information to these extensive forums. Lead GP-GPU developer Michael Fried also contributes his own tips to this valuable resource.
Coming Soon: Heterogeneous Computing with OpenCL Textbook
Many attendees to the Fusion Developer Summit received a preview copy of a new textbook entitled Heterogeneous Computing with OpenCL. Written by Microway customers at Northeastern University, it is set for release in August. Topics include:
Introduction to Parallel Programming and Introduction to OpenCL
OpenCL Device Architectures
Basic OpenCL Examples, Understanding OpenCL's Concurrency and Execution Model, Dissecting a CPU/GPU OpenCL Implementation
OpenCL Case Studies: Convolution, Video Processing, Histogram, Mixed Particle Simulation
OpenCL Profiling and Debugging
Tesla M2090 Purchase Program for 1U GPU Server Customers
Customers who have purchased 1U GPU servers from Microway may now purchase M2090s and an upgrade kit for their servers. The kit includes the extra fans, air shrouds, and holding brackets required to install M2090s. The kit is only available to existing Microway 1U GPU server customers. New customers are encouraged to purchase the 1U GPU Server with Tesla M2090s installed, stress tested, and fully integrated.