layout | title |
---|---|
default |
Schedule |
- ByzML: A Byzantine Machine Learning Library
Arsany Guirguis, EPFL - BFT Replication with Network Ordered and Attested History: NOAH
Mingliang Jiang, National University of Singapore - Millenial: Modular Microservices Macrobenchmarks
Vaastav Anand, Max Planck Institute for Software Systems - Building Trustworthy Many-Core Systems
Nathan Rutherford, Royal Holloway University of London - Hydra: In-Network Ordering with Multiple Sequencers
Inho Choi, National University of Singapore - Towards Generation of Attack Trees using Machine Learning
Kacper Sowka, Coventry University - Extend Rubik for Fluently Programming Network Stacks
Gungda Sun, National University of Singapore
- Mega-RPCs: Hardware Assisted Efficient Memory Management for Processing Mega-sized RPCs
Norwich Mungkalaton, University of Sidney - An OS to Improve the Programming of Heterogeneous Systems in the Data Center
Karim Manaouil, University Of Edinburgh - Machine Learning Input Data Processing as a Service
Dan Graur, ETH Zurich - Adaptive and dynamic edge gateways deployment for mission-critical applications
Nina Santi, Inria - Mitigating Excessive Virtual CPU Spinning in VM-agnostic Hypervisors
Kenta Ishiguro, Keio University - Tardis: User-level System Design in the context of Network Processing
Yihan Yang, National University of Singapore - RDMA Support for Online Data Intensive Services
Alireza Sanaee, Queen Mary University of London - Implementing Data Structures for Network Function Virtualisation on FPGA
Yunfan Li, National University of Singapore
- Service Boosters: Library Operating Systems for the Data Center
Henri Demoulin, University of Pennsylvania - A Black-box Approach for Scaling OS Kernels
Ankit Bhardwaj, University of Utah - Towards Pushing the Performance and Cost Envelope For Next-Generation Cloud Platforms
Jashwant Raj Gunasekaran, The Pennsylvania State University - Toward Workload-Aware State Management in Streaming Systems
Showan Asyabi, Boston University - Simplifying heterogeneous migration between x86 and ARM machines
Nikolaos Mavrogeorgis, University of Edinburgh - Transparent and Low Overhead Tracing and Fault Injection
Jun Zhang, UC Santa Cruz
- Goldilocks Fault Tolerance for Actor
Audrey Cheng, UC Berkeley - Trusted Execution for High-Performance Computing
Ayaz Akram, UC Davis - Locality Optimizations for Data Center Applications
Tanvir Ahmed Khan, University of Michigan - Secure, Distributed Sparse Matrix Multiplication for Machine Learning
Samyukta Yagati, UC Berkeley - Architectural Implications of Graph Neural Networks for Recommendation
Samuel Hsia, Harvard University - Autoscaling consensus
David Chu, UC Berkeley - Finding Crash Consistency Bugs in Persistent Memory File Systems
Hayley LeBanc, University of Texas at Austin - Wendy the Good Little BFT
Neil Giridharan, UC Berkeley