# OptiReduce: Resilient and Tail-Optimal AllReduce for Distributed Deep Learning in the Cloud

arXiv: https://arxiv.org/abs/2310.06993 · HTML: https://mlsanity.com/paper/2310.06993/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 10 Oct 2023.

## Built on by (most cited first, 8 of 8)

- [Quantizing With Randomized Hadamard Transforms: Efficient Heuristic Now Proven](https://mlsanity.com/paper/2605.06014/) (2026)
- [DynamiQ: Accelerating Gradient Synchronization using Compressed Multi-hop All-reduce](https://mlsanity.com/paper/2602.08923/) (2026)
- [Efficient AllReduce with Stragglers](https://mlsanity.com/paper/2505.23523/) (2025)
- [DBLP: Phase-Aware Bounded-Loss Transport for Burst-Resilient Distributed ML Training](https://mlsanity.com/paper/2605.01989/) (2026)
- [High-speed Networking for Giga-Scale AI Factories](https://mlsanity.com/paper/2605.21187/) (2026)
- [Tuning Collective Patterns to Alleviate Congestion in Shared AI Clusters](https://mlsanity.com/paper/2609.04417/) (2026)
- [The Life of a Token: from Words to Bits on the Wire](https://mlsanity.com/paper/2609.19924/) (2026)
- [TORQUE: Optimizing What (not) to Quantize Before and After Rotation](https://mlsanity.com/paper/2609.36032/) (2026)

## Abstract (arXiv)

We present OptiReduce, a new collective-communication system for the cloud with bounded, predictable completion times for deep-learning jobs in the presence of varying computation (stragglers) and communication (congestion and gradient drops) variabilities. OptiReduce exploits the inherent resiliency and the stochastic nature of distributed deep-learning (DDL) training and fine-tuning to work with approximated (or lost) gradients -- providing an efficient balance between (tail) performance and the resulting accuracy of the trained models. Exploiting this domain-specific characteristic of DDL, OptiReduce introduces (1) mechanisms (e.g., unreliable bounded transport with adaptive timeout) to improve the DDL jobs' tail execution time, and (2) strategies (e.g., Transpose AllReduce and Hadamard Transform) to mitigate the impact of gradient drops on model accuracy. Our evaluation shows that OptiReduce achieves 70% and 30% faster time-to-accuracy (TTA), on average, when operating in shared, cloud environments (e.g., CloudLab) compared to Gloo and NCCL, respectively.