Dec. 2 — Once again, the SC Awards Ceremony was an entertaining event and one of the many highlights of conference week. This year, the SC15 Award Co-Chairs were Padma Raghavan and Franck Cappello who helped open the SC15 Awards Ceremony.
According to Cappello, “The awards are very important for the SC conference series. They celebrate the best and the brightest of High Performance Computing [HPC]. The selection of the finalists and winners follow a very strict and codified process, involving many members of the program committee.”He continued, “Awards are not just plaques or certificates. They define excellence. They set the bar for the years to come. They are a very powerful inspiration for young and senior researchers.”
Raghavan added that even being an award finalist at SC15 is already an achievement in itself. Following is a brief summary of some of the SC15 award winners.
SC15 Student Cluster Competition
The Student Cluster Competition is comprised of two awards: the first for the Highest LINPACK and the 2nd for the Overall Winner. These awards showcase student expertise in a friendly yet spirited competition. It’s a non-stop, real-time, 48-hour challenge where students race to assemble a small cluster at SC15 to demonstrate the greatest sustained performance across a series of applications.
The students partner with vendors to design and build a cutting-edge cluster from commercially available components, not to exceed a 3120-watt power limit and work with application experts to tune and run the competition codes. The Student Cluster competition is supported by Allinea and Schlumberger.
The first Student Cluster Competition award – Highest LINPACK Benchmark – went to Team TUMuch Phun, Technische Universität from München, Germany.
The Overall Student Cluster Competition winner was determined based on a combined score for correctly completed workload, benchmark performance, demonstrated understanding of architecture and performance through profiling and analysis, and interviews. This year’s winner was Team Diablo from Tsinghua University in China.
Technical Papers Program
The Technical Papers Program is one of SC’s most valued components. Each year SC receives submissions of original research that introduce new ideas to the field and stimulate future trends in HPC. As in previous years, it was a tough competition, with 361 paper submissions covering a wide range of research interests and topics.
The committee met for two days in June to review the Papers. At the conclusion of the meeting, the committee had accepted 78 papers, which is a 22 percent acceptance rate. Of the 78 papers, nine have been selected as finalists for the Best Paper and Best Student Paper awards.
Please note, all technical papers are available at the ACM Digital Library and can be accessed by clicking here.
Best Student Paper
Patrick Flick’s and Srinivas Aluru’s winning paper proposes a novel algorithm for the suffix array problem. The performance results are very impressive and demonstrate a hundred fold improvement over existing methods. The algorithm itself has a potential broad impact on the field of genomics.
The three other finalists were:
- Adaptive and Transparent Cache Bypassing for GPUs – with Ang Li from Eindhoven University of Technology as the lead author.
- A Case for Application-Oblivious Energy – Efficient MPI Runtime- with Akshay Venkatesh – Ohio State University as the lead author.
- GraphReduce: Processing Large-Scale Graphs on Accelerator – Based Systems – with Dipanjan Sengupta – Georgia Institute of Technology as the lead author
Best Overall Paper
Xu Liu’s and Bo Wu’s winning paper proposes methods and tools that isolate the causes of memory bottlenecks in both hardware and software in advanced computing systems. The committee felt the fundamental contributions address the memory wall problem and could have wide applicability in software development at Exascale.
The other finalists were:
- Massively Parallel Phase-Field Simulations for Ternary Eutectic Directional Solidification – with Martin Bauer – FAU Erlangen Nuremberg as the lead author.
- Efficient Implementation of Quantum Materials Simulations on Distributed CPU-GPU Systems – with Raffaele Solcà – ETH Zurich as the lead author.
- Adaptive and Transparent Cache Bypassing for GPU’s – with Ang Li – Eindhoven University of Technology as lead author.
- Exploiting Asynchrony from Exact Forward Recovery for DUE in Iterative Solvers – with Luc Jaulmes – Barcelona Supercomputing Center as the lead author.
- PGX.D: A Fast Distributed Graph Processing System – with Sungpack Hong – Oracle Corporation as the lead author.