The emergence of global-scale online services has galvanized scale-out software, characterized by splitting vast datasets and massive computation across many independent servers. In a paper appearing in ASPLOS 2012, Profs. Ailamaki and Falsafi and their teams identify the inefficiencies in modern server processors and memory systems when running emerging scale-out workloads (e.g., analytics, data serving, debugging as a service, video streaming and web) and advocate server chip architectures and hardware mechanisms that maximize silicon efficiency for these workloads. For more information see, Clearing the Clouds: A Study of Emerging Workloads on Modern Hardware by Ferdman et al., available as an EPFL Tech. Report.
- Edouard Bugnion to Play Key Role in EPFL-ICRC Alliance March 12, 2018
- Polisis: An AI Eye to Decipher Privacy Policies in Seconds February 26, 2018
- Building Efficient Causal Consistency Systems: A LABOS-LPD Joint Initiative Funded by EcoCloud February 8, 2018
- Scientists Develop 10X Faster Machine Learning Algorithm January 8, 2018
- Dynamic Safe Interruptibility: A Breakthrough in AI December 21, 2017