12 papers selected.
Runtime Performance Anomaly Diagnosis in Production HPC Systems Using Active Learning
Burak Aksar, Efe Sencan, Benjamin Schwaller, Omar Aaziz et al.
TL;DR — An active-learning framework automatically diagnoses runtime performance anomalies in production HPC systems by querying targeted job profiles to minimize labeling effort.
Why notable — It bridges ML-based anomaly detection and operational HPC monitoring, demonstrating scalable root-cause identification on real supercomputer workloads.
AutoDDL: Automatic Distributed Deep Learning With Near-Optimal Bandwidth Cost
Jinfan Chen, Shigang Li 0002, Ran Guo, Jinhui Yuan et al.
TL;DR — AutoDDL automatically searches for the distributed DNN training strategy that minimizes communication bandwidth cost while meeting performance targets.
Why notable — Combining Hoefler’s communication-model expertise with automatic strategy search, this paper is essential reading for practitioners scaling DNN training across large clusters.
PeakFS: An Ultra-High Performance Parallel File System via Computing-Network-Storage Co-Optimization for HPC Applications
Yixiao Chen, Haomai Yang, Kai Lu 0002, Wenlve Huang et al.
TL;DR — PeakFS co-optimizes compute, network, and storage layers of a parallel file system to deliver ultra-high I/O throughput for HPC workloads.
Why notable — Its holistic co-design perspective sets a new performance baseline for HPC storage and provides actionable insights for next-generation parallel file system architects.
Formal Definitions and Performance Comparison of Consistency Models for Parallel File Systems
Chen Wang 0004, Kathryn M. Mohror, Marc Snir
TL;DR — This paper formalizes consistency models used by parallel file systems and provides the first systematic empirical comparison of their performance trade-offs.
Why notable — Rigorous formal treatment from Snir and Mohror clarifies long-standing ambiguities in HPC storage semantics, making it an important reference for storage system designers.
Malleability in Modern HPC Systems: Current Experiences, Challenges, and Future Opportunities
Ahmad Tarraf, Martin Schreiber 0001, Alberto Cascajo, Jean-Baptiste Besnard et al.
TL;DR — A comprehensive survey of dynamic resource malleability in HPC, covering runtime systems, job schedulers, and application-level support with lessons from production systems.
Why notable — As energy-aware and burst-resilient HPC scheduling becomes critical, this broad community-driven survey is the definitive starting point for research on malleable HPC runtimes.
Pyxis: Scheduling Mixed Tasks in Disaggregated Datacenters
Sheng Qi, Chao Jin, Mosharaf Chowdhury, Zhenming Liu et al.
TL;DR — Pyxis is a scheduler for disaggregated datacenters that jointly manages latency-sensitive and batch tasks by exploiting flexible resource pooling across the disaggregated fabric.
Why notable — It tackles one of the central open problems in cloud scheduling—multi-tenancy under disaggregation—with rigorous analysis and demonstrated gains on real workloads.
Swift: Expedited Failure Recovery for Large-Scale DNN Training
Yuchen Zhong, Guangming Sheng, Juncheng Liu, Jinhui Yuan et al.
TL;DR — Swift dramatically reduces checkpoint and recovery overhead for large-scale DNN training by combining lightweight in-memory snapshots with selective recomputation.
Why notable — As training runs on hundreds of GPUs grow longer and failures become inevitable, Swift’s fault-tolerance approach directly addresses a practical bottleneck in modern deep-learning infrastructure.
FastLoad: Speeding Up Data Loading of Both Sparse Matrix and Vector for SpMV on GPUs
Jinyu Hu, Huizhang Luo, Hong Jiang 0001, Guoqing Xiao 0001 et al.
TL;DR — FastLoad optimizes the memory-access pattern for loading both the sparse matrix and the dense vector in SpMV on GPUs, yielding significant throughput improvements.
Why notable — SpMV is a foundational kernel for scientific computing and graph analytics; this paper’s memory-access analysis and optimizations benefit a wide class of GPU applications.
KLNK: Expanding Page Boundaries in a Distributed Shared Memory System
Yiwei Ci, Michael R. Lyu, Zhan Zhang 0002, De-Cheng Zuo et al.
TL;DR — KLNK extends distributed shared memory page granularity to reduce false sharing and improve throughput for irregular access patterns.
Why notable — It addresses a classic but unsolved bottleneck in DSM systems with a practical, page-table-level mechanism applicable to emerging disaggregated memory architectures.
Enabling Efficient Erasure Coding in Disaggregated Memory Systems
Qiliang Li, Liangliang Xu, Yongkun Li 0001, Min Lyu et al.
TL;DR — This paper designs an erasure-coding scheme tailored to disaggregated memory, exploiting its unique bandwidth topology to achieve fault tolerance with low overhead.
Why notable — Fault tolerance in disaggregated memory is an open problem of growing importance; this work provides concrete mechanisms and strong performance results.
Simple, Fast and Widely Applicable Concurrent Memory Reclamation via Neutralization
Ajay Singh 0002, Trevor Alexander Brown, Ali José Mashtizadeh
TL;DR — Neutralization is a new mechanism for safe memory reclamation in lock-free data structures that is simpler, faster, and more portable than prior approaches.
Why notable — Safe memory reclamation is a pervasive challenge in concurrent programming; this algorithm’s breadth of applicability and performance improvements make it highly reusable.
DeepTM: Efficient Tensor Management in Heterogeneous Memory for DNN Training
Haoran Zhou, Wei Rang, Hongyang Chen 0001, Xiaobo Zhou 0002 et al.
TL;DR — DeepTM dynamically manages tensor placement across DRAM and NVM during DNN training to reduce memory pressure and improve throughput.
Why notable — As model sizes outpace GPU memory, heterogeneous memory management becomes critical; DeepTM provides a practical, training-aware solution with measurable benefits.