Climbing Mont Blanc – A Case Study in Challenging the Most Eager Students in a Large Programming Class
Description
Climbing Mont Blanc (CMB) is a system developed by students for students. Its aim is to inspire students to learn more about the energy efficiency of programs. CMB facilitates the evaluation of programs executed on modern heterogeneous multi-core processors such as the Exynos Octa chips used in, e.g., the Samsung Galaxy S5 and newer mobile devices. CMB evaluates execution time, energy efficiency, and a user-specified quality metric for problems where this has been defined. Student submissions can be ranked according to these different metrics, and the system can be used to host programming competitions. By arranging competitions we
encourage the participants to study and improve their own solutions in order to climb the scoreboard. This provides a strong incentive for the students to get a better understanding of energy efficient programming through exploring and testing different solutions.
In the last two years we have organized a fully optional competition as part of the course TDT4102 Procedural and Object-Oriented Programming at NTNU to challenge the most eager students in energy efficient programming. Seven new problems were developed for the CMB Challenge 2018, and 22 students took part. A total of 732 programs were submitted and 947 tests were performed with results collected and reported. It was a close race between the top three contestants and running the competition gave lots of useful feedback to the CMB-project regarding technical solutions and how to design problems. It also gave new insight among students and the course teacher about elements of energy efficient C++ coding.
We briefly present the CMB system, the problem sets developed for the competitions, and the main insights gained so far. CMB is a long-term project. The system is still under development, and is available for other interested teachers and researchers.
Notes
Files
CMB_2018_report.pdf
Files
(686.8 kB)
Name | Size | Download all |
---|---|---|
md5:6498ec4456d2cf041a6a26eb41ac7d58
|
686.8 kB | Preview Download |