Summary of A Data-driven Approach to Dataflow-aware Online Scheduling For Graph Neural Network Inference, by Pol Puigdemont et al.
A Data-Driven Approach to Dataflow-Aware Online Scheduling for Graph Neural Network Inference
by Pol Puigdemont, Enrico Russo, Axel Wassington, Abhijit Das, Sergi Abadal, Maurizio Palesi
First submitted to arxiv on: 25 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Hardware Architecture (cs.AR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed data-driven framework for dataflow-aware latency prediction in Graph Neural Networks (GNNs) tackles the variability in performance due to differing dataflows and graph properties, a limitation in the adaptability of GNN accelerators. By training regressors to predict the latency of executing specific graphs on particular dataflows using simulations on synthetic graphs, the framework achieves up to 91.28% accuracy and a Mean Absolute Percentage Error (MAPE) of 3.78%. The online scheduling algorithm introduced in this paper uses these regressors to enhance scheduling decisions, resulting in significant speedups in mean completion time and execution time compared to the best feasible baseline across all datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The researchers developed a way to predict how long it takes for special computer chips (GNN accelerators) to process different types of data. They tested their method on many different kinds of data and found that it was very accurate, predicting the right answer over 90% of the time! This is important because these chips need to be able to work with lots of different types of data, but they don’t always do a good job. The new approach helps the chips make better decisions about how to process the data and can make things run faster. |
Keywords
* Artificial intelligence * Gnn
