Understanding Capacity-Driven Scale-Out Neural Recommendation Inference
Authors:
- Yavuz YetimYYYavuz Yetim
- Özgür ÖzkanÖÖÖzgür Özkan
- Zhuoran ZhaoZZZhuoran Zhao
- Shin-Yeh TsaiSTShin-Yeh Tsai
- Carole-Jean WuCWCarole-Jean Wu
Preprint
Abstract
Deep learning recommendation models have grown to the terabyte scale. Traditional serving schemes--that load entire models to a single server--are unable to support this scale. One approach to support this scale is with distributed serving, or distributed inference, which divides the memory requirements of a single large model across multiple servers. This work is a first-step for the systems research community to develop novel model-serving solutions, given the huge system design space. Large-scale deep recommender systems are a novel workload and vital to study, as they consume up to 79% of all inference cycles in the data center. To that end, this work describes and characterizes scale-out deep learning recommendation inference using data-center serving infrastructure. This work specifically explores latency-bounded inference systems, compared to the throughput-oriented training systems of other recent works. We find that the latency and compute overheads of distributed inference are largely a result of a model's static embedding table distribution and sparsity of input inference requests. We further evaluate three embedding table mapping strategies of three DLRM-like models and specify challenging design trade-offs in terms of end-to-end latency, compute overhead, and resource efficiency. Overall, we observe only a marginal latency overhead when the data-center scale recommendation models are served with the distributed inference manner--P99 latency is increased by only 1% in the best case configuration. The latency overheads are largely a result of the commodity infrastructure used and the sparsity of embedding tables. Even more encouragingly, we also show how distributed inference can account for efficiency improvements in data-center scale recommendation serving.Keywords
All Related Versions
This publication has 16 references indexed in Scilit:
- MLPerf Inference Benchmark2020
- Scaling description of generalization with number of parameters in deep learningJournal of Statistical Mechanics: Theory and Experiment, 2020
- Deep Interest Network for Click-Through Rate Prediction2018
- Graph Convolutional Neural Networks for Web-Scale Recommender Systems2018
- Canopy2017
- Neurosurgeon2017
- Wide & Deep Learning for Recommender Systems2016
- Deep Neural Networks for YouTube Recommendations2016
- Base-delta-immediate compression2012
- Matrix Factorization Techniques for Recommender SystemsComputer, 2009