10 Nov 2017 Austin - The International Conference for High Performance Computing, Networking, Storage and Analysis - also known as Supercomputing 2017 or SC17 - kicked off on November 12th in Denver, Colorado, bringing together the international supercomputing community.
The Texas Advanced Computing Center will be there (Booth #1343), showcasing its world-class high-performance computing systems, emerging technologies, and computing expertise, which are powering world-changing discoveries throughout science and engineering.
Below are a few highlights from TACC's offerings:
In July 2017, TACC dedicated Stampede2, the 12th most powerful supercomputer in the world and the largest supercomputer at any university in the U.S., giving USA's research access to its 4,200 Intel Knights Landing (KNL) nodes.
With the release of Intel's new Skylake processors this summer, TACC completed phase 2 of the system, which brings Stampede2 to roughly 18 petaflops in peak performance.
Early users have begun using the Skylake components of the system, running quantum computing simulations, aerospace design optimizations and large-scale deep learning training.
TACC will be showcasing Stampede2 and its KNL and Skylake processors in the booth.
Machine Learning (ML) and Deep Learning (DL) represent new approaches to data analysis and prediction, particularly in areas where no overarching theory exists.
Just like traditional HPC, training neural networks or running machine learning algorithms requires extremely large numbers of computations (quintillions).
Efforts at TACC have shown that HPC systems can overcome bottlenecks in distributed performance by using well-optimized kernels and libraries, employing hyper-threading, and sizing the batches of training data properly. In fact, researchers have been using TACC supercomputers to apply ML/DL to a range of problems, ranging from health to transportation.
You can hear about TACC's Deep Learning efforts from Weijia Xu, manager of the data mining and statistics group, at the TACC booth on November 14 from 11:30am-12pm. Or stop by the Dell EMC booth on November 14 at 5pm to hear about machine learning from Niall Gaffney, TACC's Director of Data Intensive Computing.
Focused on broadening participation and creating a more diverse, inclusive HPC workforce, the three-and-a-half-day long Advanced Computing for Social Change Institute, themed around social action, will engage 30 diverse undergraduate and graduate students from Colorado during SC17.
The second Advanced Computing for Social Change Institute will capture the imagination of students by guiding them in a computing challenge that uses computation, data analysis and visualization to take a data-centric view of immigration.
Organized by TACC, XSEDE and SC17, the project is unique in its application of the world's most cutting-edge technologies to address social change.
TACC - an expert in scientific visualization - is working to develop virtual reality, augmented reality and software-defined ray tracing algorithms to help researchers develop new insights from their data.
TACC will run demos in the TACC booth of a number of VR and AR systems adapted for scientific visualization. These include the HTC Vive running VR simulations of brain MR scans; a Microsoft Hololens showing plasma simulations; an Oculus rift showcasing molecular models; and a demonstration of GraviT, a scalable ray tracing framework for visualization, running on TACC systems.
TACC vis experts will be presenting their insights at the Intel HPC Developer Conference and show their work in the Scientific Visualization & Data Analytics Showcase.
TACC doesn't just build and operate massive machines, it also develops the tools that allow researchers with all levels of expertise to use them.
Some of the best-known tools, including LMOD and XALT, have been adopted by HPC centres around the world. Others, like Sanitytool, Remora, idev, and launcher are less known but are helping TACC users achieve their goals with less effort.
Most recently, TACC developed and released a new, open-source, multi-factor authentication tool, OpenMFA, to help supercomputing center provide secure access to their resources.
You can learn about OpenMFA in the TACC booth on November 14 at 2pm, hear Robert McLay discuss LMOD and XALT on November 15 at 10am or listen to a presentation about TACC-made HPC tools on November 15 at 11am.
You've heard about petascale and exascale computing, but how about Texascale? That's where TACC is heading, with powerful systems that solve grand challenges with real-world impacts.
TACC is now home to more than 15 advanced computing systems - several in the TOP500 in the world - serving thousands of researchers nationwide.
From supercomputers specializing in genomic analysis, Cloud computing and visualization to ones specializing in distributed machine learning, FPGA processing, and web-based science, whatever large-scale computational science needs, TACC provides.
To see a full schedule of booth talks and presentations by TACC staff, you can visit TACC's SC17 microsite.