Pathways: Asynchronous Distributed Dataflow for ML
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:LFOP2BZDrecord.jsonopen to challenge →
read the original abstract
We present the design of a new large scale orchestration layer for accelerators. Our system, Pathways, is explicitly designed to enable exploration of new systems and ML research ideas, while retaining state of the art performance for current models. Pathways uses a sharded dataflow graph of asynchronous operators that consume and produce futures, and efficiently gang-schedules heterogeneous parallel computations on thousands of accelerators while coordinating data transfers over their dedicated interconnects. Pathways makes use of a novel asynchronous distributed dataflow design that lets the control plane execute in parallel despite dependencies in the data plane. This design, with careful engineering, allows Pathways to adopt a single-controller model that makes it easier to express complex new parallelism patterns. We demonstrate that Pathways can achieve performance parity (~100% accelerator utilization) with state-of-the-art systems when running SPMD computations over 2048 TPUs, while also delivering throughput comparable to the SPMD case for Transformer models that are pipelined across 16 stages, or sharded across two islands of accelerators connected over a data center network.
This paper has not been read by Pith yet.
Forward citations
Cited by 5 Pith papers
-
Unifying Dynamical Systems and Graph Theory to Mechanistically Understand Computation in Neural Networks
Multi-hop graph analysis of RNNs reveals temporal information routing and motivates resolvent regularization that outperforms L1 by enforcing pathway-level sparsity aligned with task structure.
-
Unifying Dynamical Systems and Graph Theory to Mechanistically Understand Computation in Neural Networks
RNN computation is recovered from multi-hop graph pathways, and constraining these pathways via resolvent regularization yields improved temporal sparsity and task performance over standard L1.
-
CoCa: Contrastive Captioners are Image-Text Foundation Models
CoCa unifies contrastive and generative pretraining in one image-text model to reach 86.3% zero-shot ImageNet accuracy and new state-of-the-art results on multiple downstream benchmarks.
-
PaLM: Scaling Language Modeling with Pathways
PaLM 540B demonstrates continued scaling benefits by setting new few-shot SOTA results on hundreds of benchmarks and outperforming humans on BIG-bench.
-
Piper: A Programmable Distributed Training System
Piper decouples user-defined distributed training strategies from runtime execution using transformations on a unified global training DAG IR, achieving parity on ZeRO and gains on composed strategies like DualPipe.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.