39 citations · 41 across the 8 of their papers we have counts for
8 papers
Faster Distributed Inference-Only Recommender Systems via Bounded Lag Synchronous Collectives
Kiril Dichev, Filip Pawlowski, Albert-Jan Yzelman
Recommender systems are enablers of personalized content delivery, and therefore revenue, for many large companies. In the last decade, deep learning recommender models (DLRMs) are…
HiCR, an Abstract Model for Distributed Heterogeneous Programming
Sergio Miguel Martin, Luca Terracciano, Kiril Dichev +3
We present HiCR, a model to represent the semantics of distributed heterogeneous applications and runtime systems. The model describes a minimal set of abstract operations to enabl…
Resource-Efficient Deep Learning: A Survey on Model-, Arithmetic-, and Implementation-Level Techniques
JunKyu Lee, Lev Mukhanov, Amir Sabbagh Molahosseini +6
Deep learning is pervasive in our daily life, including self-driving cars, virtual assistants, social network services, healthcare services, face recognition, etc. However, deep ne…
Implementing Efficient Message Logging Protocols as MPI Application Extensions
Kiril Dichev, Dimitrios S. Nikolopoulos
Message logging protocols are enablers of local rollback, a more efficient alternative to global rollback, for fault tolerant MPI applications. Until now, message logging MPI imple…
Energy-efficient localised rollback after failures via data flow analysis
Kiril Dichev, Kirk Cameron, Dimitrios Nikolopoulos
Exascale systems will suffer failures hourly. HPC programmers rely mostly on application-level checkpoint and a global rollback to recover. In recent years, techniques reducing the…
Dependency-Aware Rollback and Checkpoint-Restart for Distributed Task-Based Runtimes
Kiril Dichev, Herbert Jordan, Konstantinos Tovletoglou +4
With the increase in compute nodes in large compute platforms, a proportional increase in node failures will follow. Many application-based checkpoint/restart (C/R) techniques have…