Automated Hot Text and Huge Pages: An Easy-to-adopt Solution Towards High Performing Services

International Conference on Web Services (ICWS)

Abstract

Performance optimizations of large scale services can lead to significant wins on service efficiency and performance. CPU resource is one of the most common performance bottlenecks, hence improving CPU performance has been the focus of many performance optimization efforts. In particular, reducing iTLB (instruction TLB) miss rates can greatly improve CPU performance and speed up service running.

At Facebook, we have achieved CPU reduction by applying a solution that firstly identifies hot-text of the (software) binary and then places the binary on huge pages (i.e., 2MB+ memory pages). The solution is wrapped into an automated framework, enabling service owners to effortlessly adopt it. Our framework has been applied to many services at Facebook, and this paper shares our experiences and findings.

Latest Publications

Sustainable AI: Environmental Implications, Challenges and Opportunities

Carole-Jean Wu, Ramya Raghavendra, Udit Gupta, Bilge Acun, Newsha Ardalani, Kiwan Maeng, Gloria Chang, Fiona Aga Behram, James Huang, Charles Bai, Michael Gschwind, Anurag Gupta, Myle Ott, Anastasia Melnikov, Salvatore Candido, David Brooks, Geeta Chauhan, Benjamin Lee, Hsien-Hsin S. Lee, Bugra Akyildiz, Max Balandat, Joe Spisak, Ravi Jain, Mike Rabbat, Kim Hazelwood

MLSys - 2022