Winners of ASC17 Finals

 

Champion


Tsinghua University

 

Silver Medal Winner


Beihang University

 

e Prize


Tsinghua University

 

Highest LINPACK


Weifang University

 

Application Innovation


Ural Federal University

 

Application Innovation


Shanghai Jiao Tong University

 

Application Innovation


Northwestern Polytechnical University

 

Application Innovation


National Tsing Hua University

 

Best Popularity


Saint Petersburg State University

Zhengzhou University

 

First Class


National Tsing Hua University

Sun Yat-sen University

Shanghai Jiao Tong University

University of Warsaw

Dalian University of Technology

Hong Kong Baptist University

Northwestern Polytechnical University

Weifang University

Taiyuan University of Technology

Southeast University

University of Erlangen-Nuremberg

Zhengzhou University

Ocean University of China

Ural Federal University

Huazhong University of Science and Technology

The PLA Information Engineering University

Saint Petersburg State University

University of Miskolc

 

ASC Student Supercomputer Challenge 2017
Final Competition Notification

    Date Time Content (1) Content (2) Content (3)
    April 23   Check-in
    April 24 08:00-20:00 Announcement of contest rules, cluster building and testing The runtime power consumption of the entire cluster should be less than 3000W MASNUM application debugging MASNUM application shall be optimized on Sunway KNL platform setup KNL
    application shall be
    optimized on
    KNL
    platform
    April 25 08:00-20:00 Cluster building and testing MASNUM application debugging KNL platform setup
    April 26 08:00-18:00 Performance testing of HPL, HPCG, Mystery Application, Falcon Performance testing of MASNUM application Performance testing of LAMMPS on KNL
    April 27 08:00-18:00 Performance testing of PaddlePaddle Performance testing of MASNUM application Performance testing of LAMMPS on KNL
    (08:00-15:00)
    April 28 08:00-13:00 Team Presentation
    14:30-18:30 The Awards Ceremony

    * subject to change

 

Rules of the final stage
1. Optimization methods that are only applicable to specific parameters or input data are strictly prohibited.
2. If there are any modifications on the algorithm, the new algorithm must be mathematically equivalent to the original one.
3. If any rules given above are violated, a score of zero will be given for the corresponding task.
Note: when in doubt, a team needs to submit a query to the contest committee before the competition on whether a specific optimization method violates the rules, and a decision will be made by the evaluation committee before the competition. Otherwise, the team will have no chance to provide further explanations when its optimization method is ruled out by the evaluation committee during the competition.

-- Power Consumption and Hardware Platform Restrictions

Restriction of power consumption and hardware platform

  1. Except the KNL platform, the runtime power consumption of every team must be under 3000W, otherwise the current task result becomes invalid. Within the limit of the power consumption, the team should design the system to achieve the best performance of test applications.
  2. All teams should build their design based on the Inspur NF5280M4 server. The components listed in the table below will be provided by Inspur. The teams can also choose to use other components (except the server itself) at their own costs. (If you use GPUs, please prepare the GPU power cables as needed yourself, the NF5280M4 has two PCI-E 8PIN power cables for each GPU. The NF5280M4 server can serve two GPUs at most.). During the final contest, the system platform cannot be rebooted, or changed.

  3. Item Name Configuration
    Server Inspur
    NF5280M4
    CPU: Intel Xeon E5-2680v4 x 2,2.4Ghz,14 cores
    Memory: 16G x8,DDR4,2133Mhz
    Hard disk: 1TB SATA x 1
    Estimated power consumption:
    E5-2680v4 TDP 120W, memory nominal 7.5W, hard disk
    nominal 10W
    HCA card FDR InfiniBand Mellanox Connect X®-3 HCA card, single port QSFP,
    FDR IB MCX353A-FCBT
    Estimated power consumption:9W
    Switch GbE switch 10/100/1000Mb/s,24 ports Ethernet switch
    Estimated power consumption:30W
    FDR-IB switch Switch X™ FDR InfiniBand switch, 36 QSFP port
    Estimated power consumption:130W
    Cable Gigabit CAT6 cables CAT6 copper cable, blue, 3m
    InfiniBand cable InfiniBand FDR cable, QSFP port, working with the InfiniBand switch

    Tips of power consumption: the power consumption of a single server above (excluding accelerators) is about 380W when running with full workload.

  4. During the final contest, each team will be provided one Inspur NF6248 KNL server for LAMMPS. This server will be NOT placed in the team’s cabinet. The team can connect their KNL server with local network. And there is NO 3000W power limit for LAMMPS.

    Item Name Configuration
    KNL platform Inspur
    NF6248
    CPU: Intel Xeon Phi-7210, 1.3Ghz,64 cores
    Memory: 16G x6,DDR4,2133Mhz
    Hard disk: 1TB SATA x 1
    Estimated power consumption:
    No power limit.

-- Technical Regulation and Evaluation Criteria

ARestrictions
  1. HPL, HPCG, Mystery Application, Falcon and Traffic prediction with PaddlePaddle shall be run on each team’s onsite cluster:
    The power consumption must be under 3000W. Otherwise, no result will be accepted.
  2. MASNUM shall be optimized on Sunway
  3. LAMMPS shall be optimized on the KNL platform

BPerformance Optimization (90 points)
I. HPL performance optimization (7.5 points):

  1. Platform requirement: The runtime power consumption must be under 3000W. Otherwise, the current task result becomes invalid. 
  2. Goal: The highest performance is the goal while passing the correctness checking.
  3. Software downloading: http://www.netlib.org/benchmark/hpl/

II. Performance optimization of HPCG (7.5 points):

  1. Platform requirement: The runtime power consumption must be under 3000W. Otherwise, the current task result becomes invalid.
  2. About run time: HPCG (version 3.0) runs must be at least 1800 seconds (30 minutes) as reported in the output file. The Quick Path option is not allowed.
  3. Software downloading: http://www.hpcg-benchmark.org/software/index.html

III. Performance optimization of LAMMPS (15 points):

  1. Platform requirement: The test platform is KNL. Every team will be assigned ONE KNL node as your own from the committee with no power limitation. Results run from other platform will be invalid.
  2. Goal: The committee will announce several LAMMPS workload during the finals. Every team can only modify the options related to the parallel setting. Other modifications of the workloads are prohibited. The team needs to pass the correctness checking of each workload, and the goal is to achieve the shortest runtime of each workload.
  3. Software downloading: http://lammps.sandia.gov/download.html

IV. Performance optimization of Falcon (15 points):

  1. Platform requirement: The power restriction of the test platform is 3000W. If the power consumption of system exceeds 3000W during the contest, the current task result becomes invalid. 
  2. Goal: The committee will announce several Falcon workloads during the finals. Every team can only modify the options related to the parallel setting. Other modifications of the workloads are prohibited. The team needs to pass the correctness checking of each workload, and the goal is to achieve the shortest runtime of each workload
  3. Falcon source code downloading: https://github.com/PacificBiosciences/FALCON

V. Performance optimization of the Mystery Application (15 points):

  1. Platform requirement: The power restriction of the test platform is 3000W. If the power consumption of system exceeds 3000W during the contest, the current task result becomes invalid.  
  2. Goal: The committee will announce the Mystery Application software and the corresponding workloads on site to all the teams at the same time. Each team can then perform application compilation and optimization; every team can only modify the options related to the parallel setting. Other modifications of the workloads are prohibited. Every workload result is required to pass the correctness checking and to achieve the shortest runtime of all workloads.

VI. Performance optimization of MASNUM (15 points):

  1. Platform requirement: MASNUM application shall be tested and optimized on Sunway TaihuLight. Each team will be allowed to use at most 64 SW CPUs with 256 CGs.
  2. Goal: The committee will announce multiple MASNUM workloads during the finals. Every team can only modify the options related to the parallel setting. Other modifications of the workloads are prohibited. Every team is allowed to design and implement proper parallel algorithm optimization and many-core optimization for the MASNUM source code. Each team needs to pass the correctness checking of each workload, and the goal is to achieve the shortest runtime of each workload. All teams will be required to provide screen print log and all result files of each workload.

VII. Traffic prediction with PaddlePaddle (15 points):

  1. Platform requirement: The power restriction of the onsite test platform is 3000W. This question must be solved with PaddlePaddle on the onsite test platform. If the power consumption of system exceeds 3000W during the contest, the current task result becomes invalid.  
  2. Goal: The committee will announce new speeds.csv and graph.csv files during the finals. The graph.csv file contains the connection information of the traffic network which is different from the preliminary contest. Each team is required to predict each link’s traffic condition for a short time interval. The historical traffic conditions of each link will be provided in speeds.csv file.

C Evaluation method of the Applications:

    Applications Points Evaluation method
    Performance Optimization
    (90 points)
    HPL 7.5 Let  be the actual performance of each team in which  is the maximum, the score P1 will be given as:
    HPCG 7.5 P2 is calculated in the same way as P2 in HPL.
    MASNUM 15 , where N is the number of workloads,  is the full score of the ith workload, the score P3 will be given as:

    Where  is the runtime of the ith workload, and  is the minimum among all the participating teams.
    Falcon 15 P4 is calculated in the same way as P3 in MASNUM
    Mystery  Application 15 P5 is calculated in the same way as P3 in MASNUM
    Traffic prediction with PaddlePaddle 15
    Where E is the Root Mean Squared Error (RMSE) of the prediction, which is defined below,

    Xactual,i is actual condition at moment i, while Xprediction,i is the prediction at moment i. Emin is the minimum of E among the teams.
    LAMMPS 15 P7 is calculated in the same way as P3 in MASNUM
    Performance Total Points

DTeam Presentation (10 points)

  1. Each team should provide a presentation in the manner of PPT by the order decided in a draw. Both the text and the speech should be given in English with up to two student speakers.
  2. The presentation should be given within 10 minutes.


Contact Us
Technical Support Yu Liu, Weiwei Wang techsupport@asc-events.org
Media Jie He media@asc-events.org
Collaboration Yan Yan executive.director@asc-events.org
General Information Dongzhi Kong info@asc-events.org

 

Partners
Copyright 2017 Asia Supercomputer Community. All Rights Reserved