AI Paper Atlas

Research map / Graph learning / Graph algorithms and clustering

Efficient Algorithms for Device Placement of DNN Graph Operators

Jakub M Tarnawski, Amar Phanishayee, Nikhil Devanur, Divya Mahajan, Fanny Nina Paravecino

NeurIPS 2020 · Graph learning › Graph algorithms and clustering · approach: Neural combinatorial optimization on graphs · 15 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Allow non-contiguous rather than contiguous splits (ablation)Pipelined inference and training device placement · Evaluated DNN workloadsThroughput: average ∼10%; BERT-3, 20-27%Improved
Allow non-contiguous rather than contiguous splits (ablation)BERT-3 operator graph, pipelined inference · BERT-3 operator graph, pipelined inferenceTPS (time-per-sample): 27.92 -> 21.91Improved
Allow non-contiguous rather than contiguous splits (ablation)BERT-3 operator graph, pipelined training · BERT-3 operator graph, pipelined trainingTPS (time-per-sample): 65.30 -> 54.21Improved
Allow non-contiguous rather than contiguous splits (ablation)ResNet50 operator graph, pipelined inference · ResNet50 operator graph, pipelined inferenceTPS (time-per-sample): 124.35 -> 124.35No clear effect
Use linearization in DP to restrict the search space (ablation)DPL · BERT-6 operator-graph trainingThroughput: 9% lossHurt
Use linearization in DP to restrict the search space (ablation)DPL · InceptionV3 layer-graph trainingThroughput: 1% lossHurt
Optimize the original operator graph rather than its contracted layer graph (ablation)DP contiguous · BERT-12 operator graph, pipelined inferenceTPS (time-per-sample): 159.43 -> 147.48; 8%Improved
Optimize latency directly rather than using max-load DP splits (ablation)Latency IP · Single-query inference workloadsLatency: up to 42% lower; 17% on averageImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy