Pengcheng Zhang
where else to find them
no profile links on file
what they talk about
all 6 in the filters
where they keep going back
no series more than once — all 4 in the filters
on a stage with
all 73 in the filters
who talks about the same things
nobody close enough yet
| title | conference | date |
|---|---|---|
| TrainMover: An Interruption-Resilient Runtime for ML Training | 20th USENIX Symposium on Operating Systems Design and Implementation (OSDI '26) | 2026-07-13 |
TrainMover is an ML training runtime that prepares standby GPUs ahead of interruptions by pre-recording output tensors and running communication-free sandboxed… | ||
| EROICA: Online Performance Troubleshooting for Large-scale Model Training | NSDI '26 | 2026-05-04 |
EROICA is an online profiler that triggers only on iteration slowdowns, profiles all workers simultaneously, extracts beta, mu, sigma fingerprints for… | ||
| SimAI: Unifying Architecture Design and Performance Tuning for Large-Scale Large... | NSDI '25 | 2025-04-28 |
SimAI claims a unified, high‑precision LLM training simulator that matches real runs 98.1% alignment by hijacking training frameworks (Megatron, DeepSpeed)… | ||
| Evolution of Aegis: Fault Diagnosis for AI Model Training Service in Production | NSDI '25 | 2025-04-28 |
Aegis is a production fault diagnosis system for AI model training that evolves from enhanced legacy tools to a customized collective communication library… | ||
| When Cloud Storage Meets RDMA | NSDI ’21 | 2021-04-12 |
Pangu integrates RDMA into production cloud storage and reports concrete wins: a user‑space stack (USS OS and USS file system) boosts CPU efficiency by over… | ||
| nothing matches | ||