Pith. sign in

REVIEW 5 cited by

FALCON: Pinpointing and Mitigating Stragglers for Large-Scale Hybrid-Parallel Training

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.12588 v1 pith:ERSWGXXA submitted 2024-10-16 cs.DC cs.OS

FALCON: Pinpointing and Mitigating Stragglers for Large-Scale Hybrid-Parallel Training

classification cs.DC cs.OS
keywords fail-slowsfalconclustereffectivelytrainingdetecthybrid-parallellarge-scale
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Fail-slows, or stragglers, are common but largely unheeded problems in large-scale hybrid-parallel training that spans thousands of GPU servers and runs for weeks to months. Yet, these problems are not well studied, nor can they be quickly detected and effectively mitigated. In this paper, we first present a characterization study on a shared production cluster with over 10,000 GPUs1. We find that fail-slows are caused by various CPU/GPU computation and cross-node networking issues, lasting from tens of seconds to nearly ten hours, and collectively delaying the average job completion time by 1.34%. The current practice is to manually detect these fail-slows and simply treat them as fail-stops using a checkpoint-and-restart failover approach, which are labor-intensive and time-consuming. In this paper, we propose FALCON, a framework that rapidly identifies fail-slowed GPUs and/or communication links, and effectively tackles them with a novel multi-level mitigation mechanism, all without human intervention. We have applied FALCON to detect human-labeled fail-slows in a production cluster with over 99% accuracy. Cluster deployment further demonstrates that FALCON effectively handles manually injected fail-slows, mitigating the training slowdown by 60.1%.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. StageFrontier: Synchronization-Aware Stage Accounting for Distributed ML Training

    cs.DC 2026-06 conditional novelty 7.0

    StageFrontier computes an exact additive accounting of exposed step time in distributed training by taking the frontier of per-rank coarse stage durations reported with unsynchronized CPU wall clocks.

  2. From Detection to Recovery: Operational Analysis on LLM Pre-training with 504 GPUs

    cs.DC 2026-05 unverdicted novelty 7.0

    Production logs from a 504-GPU LLM training cluster show 100% failure detection via multi-metric analysis, NFS saturation limiting bandwidth to 1.4-10.4% of link speed, and auto-retry achieving 33.3% success versus 12...

  3. From Detection to Recovery: Operational Analysis on LLM Pre-training with 504 GPUs

    cs.DC 2026-05 unverdicted novelty 5.0

    Production-scale empirical study of a 63-node 504-GPU cluster reports multi-signal failure detection needs, low checkpoint bandwidth utilization, heavy-tailed node exclusions, and 2.7x higher success for auto-retry chains.

  4. PRISM: Probabilistic Runtime Insights and Scalable Performance Modeling for Large-Scale Distributed Training

    cs.DC 2025-10 unverdicted novelty 5.0

    PRISM introduces a probabilistic performance modeling framework that quantifies guarantees on training time for large-scale distributed systems under runtime variability.

  5. TSGuard: Automated User-Centric Incident Diagnosis for AI Workloads in the Cloud

    cs.SE 2025-06 unverdicted novelty 5.0

    TSGuard builds domain knowledge bases offline from historical incidents and applies online multi-agent structured reasoning to diagnose AI workload failures, delivering 19.8% higher accuracy and 63.4% lower verificati...