HiPC 2015 Student Parallel Programming Challenge Supported by Intel and NVIDIA
Download Parallel Programming Challenge PDF Flyer
HiPC 2015 will host the first parallel programming challenge for students in India and abroad. The event will be geared towards students and will provide an introduction to parallel programming concepts and implementations. The purpose of this challenge is to highlight the value of code parallelism, parallel algorithms, and the role of right mathematics in solving real-world problems.
The top three student teams will be selected based on the evaluation criterion outlined below and will be invited to present their solution at the HiPC 2015 conference.
||Intel Xeon Phi Track
||NVIDIA GPU track
|Challenge problem released and Team Registration opens.
|Final Registration deadline for teams.
|Functional evaluation submission for Phase 1
|Phase 2: Team given access to cluster for self-evaluation. Mentor forum opens.
|Final Submission Deadline.
- Students groups consisting of 1 to 3 members must register as a team before the registration deadline.
- Each member must be a registered full time student.
- The submission should strictly adhere to the input/output format specified in the challenge problem.
- A set of approved frameworks will be provided along with the challenge problem. Any submissions that do not follow the given guidelines will be disqualified.
- All submissions must be team’s own work. Plagiarism will be dealt with utmost seriousness and will have serious consequences. We will follow the IEEE rules that apply to papers including adding authors to IEEE Prohibited Authors List Database. The decisions of track and program chairs in any such case will be final. All submissions will be vetted for plagiarism using MOSS MIT tool. Submissions which are found to be plagiarized will be rejected.
- Functional correctness of the result is a must to be considered for further evaluation.
- Submission Scoring/Ranking: All submissions will be ranked based on the summation of best runtimes of all datasets.
- The contest will have two parallel and independent tracks. First track will be for Intel Xeon Phi based node and second track will consists of a high end node with a NVidia GPU. More details will be released soon along with the problem statement.
- Student teams can submit to one or both tracks
- The submissions to each tracks will be evaluated independently and there will be no cross track comparison.
The challenge problem for the Intel track has been released and can be accessed through the forum. Students may choose one or more problems to complete the challenge. They may deploy and test their submission locally (on their machine or a cluster provided by their institute). After initial testing, each participating group will be given access to the cluster to deploy and may test their solutions in an environment similar to the one that will be used for final evaluations.
To participate in the challenge, one member from the team should register here. To do so, first create an account on the registration website, and then click “HiPC 2015 Parallel Programing Challenge” link on the left navigation bar to begin team registration.
- Participation in the programming challenge is free and does not incur any registration fees.
- After the final notification, each member of the team selected for presentation must register for the conference individually at regular conference rates. See here for details. The registration fees and other expenses will be reimbursed on-site.
Each participating team should submit the following:
- Source code and associated makefiles
- Instructions on compiling
- Output report of functionality correctness tests
- Performance measurement table (template will be provided)
- Technical report detailing software architecture, specific performance optimizations, etc. (4 pages, 2 column IEEE format)
All submissions will be vetted for plagiarism using MOSS MIT tool. Submissions which are found to be plagiarized will be rejected. Submission link will be posted soon.
A team of experts in high performance computing and parallel programming will be available to the students to guide them through the process of building scalable, robust parallel solutions. All questions should be posted in a public forum visible to all the participating teams and no private conversations will be allowed. All teams should adhere to the forum guidelines (to be posted soon); failure to do so might lead to disqualification.
The top three submission teams will be invited to the HiPC 2015 conference in Bengaluru, India where they will have the opportunity to present their approach and solutions. In addition, these teams will be presented with prizes at the HiPC conference from our sponsors.
Student Travel Support
The top three teams will be invited to present at the HiPC 2015 conference. Partial travel support will be provided for students to cover their traveling expenses. Please visit this page later for more details.
For further details contact firstname.lastname@example.org