The International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), the premier forum for interdisciplinary systems research, recognized two teams from the Department of Computer Science with awards for their research papers.  

Clearing the clouds: a study of emerging scale-out workloads on modern hardware, with lead author Michael Ferdman, won the 2023 ACM SIGARCH/SIGPLAN/SIGOPS ASPLOS Influential Paper Award, the equivalent of a test-of-time award that recognizes ASPLOS papers from ten or more years ago that have had major influence on the field. The paper introduced CloudSuite, a benchmark suite of scale-out workloads to measure the efficiency of servers used in cloud computing.  Ferdman and researchers from Ecole Polytechnique Federale de Lausanne in Switzerland discovered that the predominant processor microarchitectures, such as the ones used by Intel and AMD, were inefficient for running cloud computing workloads. Through this research, they were able to identify the key source of area and power inefficiency in the organization of the instruction fetch, core microarchitecture, and memory system.

Also at the ASPLOS conference, Stony Brook University doctoral computer science student Evan West and computer science professor Michael Bender, along with researchers from five institutions, received the Distinguished Paper award for their paper Mosaic Pages: Big TLB Reach with Small Pages. The paper addresses the problem with translation lookaside buffer (TLB) where big data applications face bottleneck as the number of TLB entries is constrained and growing slower than the memory needs of the applications. The paper develops “mosaic pages” which compresses multiple address  translations into one TLB entry.

The ASPLOS annual event brings together researchers and practitioners in computer architecture, programming languages, and operating systems. In 2023, the 28th meeting was held in Vancouver, Canada.

- Kimberly Xiao