What do you get with 137,000 square feet of exhibit space and 343 industry and research organizations? You get Supercomputing, the largest conference for HPC experts.
With over 13,000 attendees (the largest attendance in the history of the show) the exhibit floor was buzzing with chatter about current trends, new products, and innovative ideas for HPC in 2016. Not only was this a big year for the Supercomputing community but also it was the launching pad for our newest product CodeDynamics and a fresh product release for TotalView for HPC.
What’s new with TotalView for HPC?
Showing off the latest advances in TotalView for HPC was a success at this year’s event. Many of our current customers were happy to see our newest improvements in the 8.15.10 release including:
• Support for Linux PowerLE: TotalView now supports Linux PowerLE (Little Endian) systems. All major debugging support is provided including memory debugging, Remote Display Client, CUDA, and MPI debugging.
• Improvements for C++11: TotalView now supports C++11 features for the GNU compiler.
• Added support for NVIDIA CUDA 7.5.
• Additional compiler support: GCC 5.2, Intel 16.0.
Whether people were familiar or new to the features and functionality of TotalView (or they just wanted to ogle our Alienware laptops we use to show CUDA debugging), one thing is for sure, they were impressed with the ability of the product to simultaneously debug many processes and threads in one single window. Their eyes lit up with excitement—most likely imagining how much this product could improve their code while decreasing development cycle times.
Our newest family member CodeDynamics launches
While TotalView for HPC continues to deliver support for HPC applications, CodeDynamics brings parallel, multithreaded testing power to commercial applications. With the use cases for large-scale data modeling and analytics grows, it’s driving the adoption of technologies traditionally reserved for HPC into new enterprise and industrial applications. The need is the same: extremely fast systems, highly optimized code, and innovative algorithms – all requirements for computing solutions in financial, energy, science, government, and other commercial services.
Attendees thought that the fresh looking GUI of CodeDynamics looked very modern, user friendly, and more advanced than other enterprise level tools of the same nature.
Some other unique features of this product include:
• The ReplayEngine which allows users to step backwards through code to isolate the exact location of a fault.
• The ability to breakpoint, stop, inspect, and start individual threads, making it easier to understand execution interactions and memory dependencies.
• A type transformation facility that allows easy inspection of complex STL collection class and Boost library objects.
• Quick start of testing sessions on a target executable or an application that’s already running.
Big data matters too!
Another hit at our booth this year was our Java Numerical Library (JMSL) product which is the only data analytics and mining engine native to Java. People were interested to see how this product could help in their scientific, academic, and data mining product, which are the key verticals in the HPC marketplace. JMSL boosts machine models, provides algorithms that support Hadoop, and helps organizations make better business decisions with predictive analytics.
2016 will be another big year for HPC
Supercomputing was a big event for Rogue Wave this year with product improvements to current solutions and a new product introduction for the commercial marketplace. This event continues to push innovation, creativity, and modernization in the HPC marketplace. This year, more than any other year proved that HPC and big data matter and will continue for many years to come.
See you in Salt Lake City in 2016 at booth 2425!
Looking to learn more? Check out these materials:
• Deterministically troubleshooting network distributed applications: Learn how debugging can help troubleshoot issues while also showing you how your other programs are working just fine.
• Debugging numerical simulations on accelerated architectures: TotalView for OpenPOWER, CUDA, and OpenMP: This talk also reviews the collaboration between Lawrence Livermore National Laboratory (LLNL) and Rogue Wave Software towards the TotalView debugger supporting the OpenPOWER stack on LLNL’s future Sierra system.
• JMSL Numerical Library architecture: See the building blocks that Rogue Wave developed to perform higher mathematics and statistics in Java.