On September 22, 2011, the final match and award ceremony of 2011 Taiwan CUDA Programming Contest, co-hosted by NCHC, NVIDIA, and ACCTA, took place at NCHC headquarters in Hsinchu. The championship of this contest was won by “Miracle” team from NTHU.
GPU Computing, the Rising New Star in High Performance Computing
GPU computing has already become the top choice for graphics and floating point computation, the CUDA (Compute Unified Device Architecture) programming language is the key for R&D personnel to easily resolve computing problems by using GPU. CUDA can be used to do GPU computing and speed up the processing of many complex computing jobs. It could be applied to a wide range of applications, such as 3D graphics computing, medical analysis, seismic computing, bio-chemical simulation, weather and climate computing, and financial risk management, which can all have the accelerated processing via GPU. Therefore, through this contest, we hope to encourage domestic talents to absorb more knowledge of the CUDA programming language.
NCHC, One of the Pioneers in the Development of GPU Accelerated Computing
From its inception in 1991, the NCHC's primary mission has continuously been to provide users with quality HPC environments. In response to the international trend of GPU application and development, NCHC has incorporated GPU into the HPC environment since 2007. It also organized a GPU Computing Team to conduct related research and applications. In 2009, NCHC established four CPU+GPU systems consisting of 10 PCs and a test platform of 16-node GPU cluster for academia first to use GPU on scientific research. Furthermore, in 2010, NCHC integrated cluster techniques to build a 48-node GPU+CPU cluster for users to accelerate the output of research outcomes.
CUDA Programming Contest, Established GPU foundation in Taiwan
In addition to building GPU clusters, considering the high technical threshold of GPU computing, NCHC has aggressively integrated the industry and academia to conduct research on GPU applications and offered CUDA programming courses to build a steady foundation for the use of GPU. NVIDIA also has been actively worldwide promoting the computing power of GPU and systematically training talents. This was the first time, NCHC and NVIDIA co-hosted the 2011 Taiwan CUDA Programming Contest. We hope not only to foster and uncover domestic GPU talents, but also to encourage more students to join the R&D of GPU. Through the contest, we hope to interact with participants and to help more people enhance their grasp of CUDA.
This year’s contest had no limitations on contestants and subjects. It attracted 33 teams to participate. All participating entries were evaluated on four key aspects – creativity, practicality, professionalism, and interest by experts from industry, academia, and NCHC. Total of thirteen finalists advanced to the finals, took place at NCHC on September 22nd.
In the future, we hope that talent from all fields will be able to refine GPU techniques in order to contribute more to domestic technology applications and industry development.
Champion: NTHU – Li-ting Chiu, Chieh-hsing Chen, Ching-yang Kuo, and Chung-che Wang (Contest Subject Title: Miracle)
Second place: NTU – Chung-yao Hung, Pai-shun Ting, Cheng-hsuan Li, and Yu-jung Chen (Contest Subject Title: Multi-View Stereo Reconstruction)
Third place: NCKU – Wei-tsuo Chen, Wen-yu Su, Shuang-yu Chen, and Wei-hsiang Liao (Contest Subject Title: Molecular Dynamic Simulation on a GPU Cluster)
Teams of Excellence:
NTNU and NTHU – Cheng-hung Lin, Chen-hsiung Liu, Lung-sheng Chien, and Shi-chieh Chang (Contest Subject Title: PFAC: A Library for String Matching on NVIDIA GPUs)
CGU and CCU – Yu-hsiang Lin, Yi-chi Liu, and Sheng-ta Li (Contest Subject Title: Crack RSA)
NTHU –Sheng-yuan Lin, Ching-yi Chen, Cheng-hung Lin, and Chen-hsiung Liu (Contest Subject Title: Fast Feature-point based Image Morphing)
¡½Left: The Miracle team of NTHU receiving the champion trophy and the prize from NCHC Director Kuo-ning Chiang (first from the right)
¡½Right: Winning team members in a group photo with judges and superiors