In April 14th, the 2015 ASC student supercomputer challenge (ASC15) officially announced the 16 finalists, including the Tsinghua University, Massachusetts Green Team, Chinese University of Hong Kong, Nanyang Technology University, four of the world’s top 50 universities.
At the same time, the organizing committee also announced the 6 questions of the finals. The questions are supercomputer benchmark HPL, SKA (Square Kilometre Array telescope) data processing software Gridding, nano scale molecular dynamics software NAMD, the air quality numerical model WRF-CHEM, the computational fluid dynamics software Palabos, and one mystery question (will unveiled on the scene). Among these 6 questions, the Gridding will become ASC15’s “e Prize computing Challenge Award”, 16 elite teams will try to resolve this problem, bothering many scientists.
For the first time, adopting the international research project into the game is also a highlight of ASC15. The SKA project, initiated by scientists from many countries, is the largest astronomy cooperation project. The project’s goal is to build the world’s largest radio telescope array, in order to explore the origin of the universe, dark matter, and the mysteries of the universe. The amount of data per second SKA can gather is greater than 12Tb, equivalent to 3.5 times to the 2013 China international Internet bandwidth. Almost 50% of the massive astronomical data were needed to be processed by Gridding software, of which a part of data need to be calculated in real time. That is why the requirements of computing speed is very high. At present, many astronomy experts have been working on the Gridding software optimization, but there is no satisfactory version. Therefore, the Gridding question of the ASC final gives much creative zest to the finalists, and the results remail full of suspense.
In fact, the more challenging questions – “e Prize computing challenge”, is not only the part of ASC competition characteristic, but also one of the most attractive point for the college students. For those teams (average age ~20), “challenge” seems to have the magic, so that they are crazy about it, and the process from confused to click into place is the more valuable harvest to grow up. The innovation spirit in the challenge, advocated by the ASC, as well exhibited at one glance.
As shared by the name, “e Prize Award” more focus on “challenge”, the goal is to improve the degree of parallelism of the supercomputer applications, then to really realize the value of the large supercomputer. In the last year for the first time, ASC set up a “e Prize computing Challenge Award” for a petroleum exploration software 3D-EW, Shanghai Jiao Tong University migrated it from serial to parallel, ported it from CPU to CPU+MIC, eventually scale up to 1,024 nodes (200K cores) with near linear speedup.
In addition to Gridding, WRF-CHEM is also a very interesting question. WRF-CHEM adopts the most advanced air quality forecasting model, can predict changes in the smag and PM2.5 particles, PM10 particles, sulfur dioxide and other harmful ingredients in the air. If the model’s resolution doubled, the computing quantity will show a geometric growth. The optimization from those ASC15 teams of this software, can help to reach the smaller scale and higher accuracy air quality forecast.
It is not difficult to see that the challenge goes through entire ASC15. The participating college students from different countries, different nationalities, will how to demonstrate their knowledge and innovation? All answer will be unveiled in ASC15 finals (Taiyuan University of Technology, 2015 May).