Michele Ferrero
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Michele Ferrero has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Gabriele Castellano
Huawei Technologies Co. Ltd, Paris, France
2 shared publications - Massimo Gallo
Huawei Technologies Co. Ltd, Paris, France
2 shared publications - Paolo Medagliani
Independent Researcher, Paris, France
2 shared publications - Baosen Zhao
Politecnico di Milano, Milan, Italy
1 shared publication - Davide Palmiotti
Politecnico di Milano, Milano, Italy
1 shared publication - Francesco Musumeci
Politecnico di Milano, Milan, Italy
1 shared publication
Latest Publications
- Open Access
- 19 November 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Networking
- Vol. 4 (CoNEXT4) , 1-25
MCCL: Measurement-based Collective Communication Library
Collective communication is a critical performance bottleneck in large-scale distributed training. Existing libraries and synthesis frameworks compute open-loop schedules once, assuming a static topology and peak link capacities. In production, however, persistent gray failures can silently degrade link throughput and latency for hours, causing pre-computed schedules to lose much of their efficiency during execution. In this paper, we present Measurement-based Collective Communication Library (MCCL), a collective communication library that closes the loop between runtime network measurement and dynamic schedule synthesis. MCCL passively measures GPU-to-GPU throughput, detects persistent degradation, re-synthesizes the affected communication-intensive schedules with warm-started synthesizers, and hot-swaps them without interrupting training. We validate the complete loop on a multi-node hardware prototype and use simulations to evaluate cluster-scale scenarios. In our experiments, MCCL incurs negligible overhead (<2%) while maintaining high measurement accuracy (<1.2% error) and low schedule hot-swap time (≈ 12 ms). Notably, under gray failures, MCCL reduces training time by up to 4.99x over static schedules on simulated training workloads for state-of-the-art LLMs such as Llama and DeepSeek.
- 19 November 2026
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Networking
- Vol. 4 (CoNEXT4) , 1-25
Open AccessMCCL: Measurement-based Collective Communication Library
Collective communication is a critical performance bottleneck in large-scale distributed training. Existing libraries and synthesis frameworks compute open-loop schedules once, assuming a static topology and peak link capacities. In production, however, persistent gray failures can silently degrade link throughput and latency for hours, causing pre-computed schedules to lose much of their efficiency during execution. In this paper, we present Measurement-based Collective Communication Library (MCCL), a collective communication library that closes the loop between runtime network measurement and dynamic schedule synthesis. MCCL passively measures GPU-to-GPU throughput, detects persistent degradation, re-synthesizes the affected communication-intensive schedules with warm-started synthesizers, and hot-swaps them without interrupting training. We validate the complete loop on a multi-node hardware prototype and use simulations to evaluate cluster-scale scenarios. In our experiments, MCCL incurs negligible overhead (<2%) while maintaining high measurement accuracy (<1.2% error) and low schedule hot-swap time (≈ 12 ms). Notably, under gray failures, MCCL reduces training time by up to 4.99x over static schedules on simulated training workloads for state-of-the-art LLMs such as Llama and DeepSeek.
- Open Access
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 2189-2191
POSTER: Beyond Probabilistic Data Structures for AI/ML Workload Monitoring
The rapid growth of AI models is placing unprecedented pressure on switch ASIC memory allocated for telemetry and flow monitoring. In this poster, we argue that the predictability of AI/ML traffic patterns can be exploited by perfect hashing techniques for accurate flow- and packet-tracking with minimal memory overhead.
- 11 August 2026
- Published by Association for Computing Machinery (ACM)
- p. 2189-2191
Open AccessPOSTER: Beyond Probabilistic Data Structures for AI/ML Workload Monitoring
The rapid growth of AI models is placing unprecedented pressure on switch ASIC memory allocated for telemetry and flow monitoring. In this poster, we argue that the predictability of AI/ML traffic patterns can be exploited by perfect hashing techniques for accurate flow- and packet-tracking with minimal memory overhead.
- Open Access
- 30 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 21-27
SCALE-CCL: A Scalable Collective Communication Library for Wide-Area Distributed Training
- 30 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 21-27
Open AccessSCALE-CCL: A Scalable Collective Communication Library for Wide-Area Distributed Training