Building resources for teaching computer architecture through electronic peer review

Abstract

Electronic peer review is a concept that allows students to get much more feedback on their work than they normally do in a classroom setting. Students submit assignments to the system, which presents them to other students for review. Reviewer and author then communicate over a shared Web page, and the author has a chance to submit revised versions in response to reviewer comments. At the end of the period, the reviewer gives the author a grade. Each author gets reviews from several reviewers, whose grades are averaged. At the end of the review period, there is a final round when students grade each other's reviews. Their grade is determined by the quality of both their submitted work and their reviewing. This paper reports on our use of peer review in two computer architecture courses, a microarchitecture course and a parallel-architecture course. Students in these courses engaged in a variety of peer-reviewed tasks: Writing survey papers on an aspect of computer architecture, making up homework problems over the material covered in class, creating machine-scorable questions on topics covered during the semester, animating and improving graphics in the lecture presentations and annotating the lecture notes by inserting hyperlinks to other Web documents. Students generally found these exercises beneficial to their learning experience and they have provided resources that can be used to improve the course. In fact, with such a system large classes are actually a blessing, since they produce better and more copious educational materials to be used in subsequent semesters.