In 1998, in commemoration of the 25th anniversary of ISCA, Guri Sohi put together a volume of selected articles and author retrospectives from 1973 to 1995 (ISCA-1 to ISCA-22) with the help of several program chairs and program vice chairs from ISCA-15 to ISCA-21. As we celebrate ISCA’s 50th birthday this year, we wanted to compile a volume that picks up where the first volume left off. We borrowed some metrics from Guri Sohi’s original effort: we excluded papers from the most recent two years and assembled a small team of judges made up of program chairmen from the last few ISCA conferences: David Brooks, Fred Chong, Lieven Eeckhout, Babak Falsafi and Hillery Hunter. (Although Lizy John was the program director for ISCA 2021, as co-editor of this volume, she only participated in review coordination, but not in paper selection.)
We have not set out to select the « best » or « most influential » newspapers. Instead, we have created a collection of significant and often memorable papers from 1996 to 2020 that tell a story of how research at the ISCA has progressed over those twenty-five years, each accompanied by a retrospective of the authors. The resulting volume is necessarily an imperfect compilation, missing many significant contributions to the ISCA and sister conferences. However, we believe it tells an exciting and meaningful story.
We guided the card selection as follows: We assigned two judges to select cards from each five-year interval (a different pair for each interval). Thus, each arbitrator was paired with another arbitrator for a five-year interval and with an additional arbitrator for a different five-year interval. For each five-year interval, we asked the responsible couple to set aside any records where either of them had a conflict of interest, more or less as defined by the ISCA-50 rules. We also excluded SIGARCH/TCCA Influential ISCA Paper winners from the selection process, as they would automatically be included in the retrospective.
In round 1, each arbitrator tentatively selected fifteen papers from each of their allotted five-year intervals. Then, in round 2, the judges met independently with their assigned partner for each five-year interval, discussed their individual choices, and came up with a joint selection of up to twenty papers. In round 3, we assigned each of the cards set aside as conflicts to a referee who didn’t have a conflict of interest and who hadn’t been assigned the five-year range to which the card belonged. Papers that one of us had a conflict over were assigned by the other; there were no documents in the list that we both had conflicts over. Finally, in round 4, we repeated the process for the documents selected in round 3.
In all, each article (1,077 in total) was considered by two judges who had no conflicts of interest, and each selected article was chosen by both judges, either by discussion (rounds 1-2) or sequentially (rounds 3-4). ). The final number of articles shortlisted was 98. This is more than double the number of articles included in the Guri Sohi retrospective, which is mainly a reflection of the growth of our community in the intervening years. We then contacted the authors and asked them to write a retrospective of their papers.
We are incredibly grateful to David, Fred, Lieven, Babak and Hillery for the tremendous effort they have put into this endeavor, and to the authors for their enthusiasm and dedication in preparing their retrospectives. We hope that the resulting collection will be exciting reading for both the older and younger generations of computer architects!
José F. Martínez served as a member of the SIGARCH Executive Committee from 2019 to 2023 and Lizy K. John served as a member of the TCCA Executive Committee during the same time period. They are at Cornell University and the University of Texas at Austin, respectively.
Disclaimer:These posts are written by individual contributors to share their thoughts on the Computer Architecture Today blog for the benefit of the community. Any views or opinions represented in this blog are personal, belong solely to the blog author, and do not represent those of ACM SIGARCH or its parent organization, ACM.