mlsanityData to 7 Oct 2026

Heterogeneous Acceleration Pipeline for Recommendation System Training

arXiv 2204.05436 ↗First posted 11 Apr 2022cs.AR cs.AI cs.LGcited by 13

Not on the map. Why?

Abstract (arXiv)

Recommendation models rely on deep learning networks and large embedding tables, resulting in computationally and memory-intensive processes. These models are typically trained using hybrid CPU-GPU or GPU-only configurations. The hybrid mode combines the GPU's neural network acceleration with the CPUs' memory storage and supply for embedding tables but may incur significant CPU-to-GPU transfer time. In contrast, the GPU-only mode utilizes High Bandwidth Memory (HBM) across multiple GPUs for storing embedding tables. However, this approach is expensive and presents scaling concerns. This paper introduces Hotline, a heterogeneous acceleration pipeline that addresses these concerns. Hotline develops a data-aware and model-aware scheduling pipeline by leveraging the insight that only a few embedding entries are frequently accessed (popular). This approach utilizes CPU main memory for non-popular embeddings and GPUs' HBM for popular embeddings. To achieve this, Hotline accelerator fragments a mini-batch into popular and non-popular micro-batches. It gathers the necessary working parameters for non-popular micro-batches from the CPU, while GPUs execute popular micro-batches. The hardware accelerator dynamically coordinates the execution of popular embeddings on GPUs and non-popular embeddings from the CPU's main memory. Real-world datasets and models confirm Hotline's effectiveness, reducing average end-to-end training time by 2.2x compared to Intel-optimized CPU-GPU DLRM baseline.

Key facts. Cited by 13 papers in the map (data through 7 Oct 2026). Builds on 32 of its 77 references that are in the map. First posted on arXiv 11 Apr 2022.

Builds on

32 of 77 references.

References · 32

  1. 2020Language Models are Few-Shot Learnerscited by 32,199
  2. 2017Neural Collaborative Filteringcited by 1,108
  3. 2017In-Datacenter Performance Analysis of a Tensor Processing Unitcited by 610
  4. 2019Deep Learning Recommendation Model for Personalization and Recommendation Systemscited by 327
  5. 2020Embedding-based Retrieval in Facebook Searchcited by 155
  6. 2015Compressing Neural Networks with the Hashing Trickcited by 150
  7. 2019The Architectural Implications of Facebook's DNN-based Personalized Recommendationcited by 83
  8. 2019Analysis of Large-Scale Multi-Tenant GPU Clusters for DNN Training Workloadscited by 79
  9. 2021Software-Hardware Co-design for Fast and Scalable Training of Deep Learning Recommendation Modelscited by 78
  10. 2019Compositional Embeddings Using Complementary Partitions for Memory-Efficient Recommendation Systemscited by 68
  11. 2020Distributed Hierarchical GPU Parameter Server for Massive Scale Deep Learning Ads Systemscited by 59
  12. 2021TT-Rec: Tensor Train Compression for Deep Learning Recommendation Modelscited by 58
20 more
  1. 2020DeepRecSys: A System for Optimizing End-To-End At-scale Neural Recommendation Inferencecited by 54
  2. 2019RecNMP: Accelerating Personalized Recommendation with Near-Memory Processingcited by 49
  3. 2020Understanding Training Efficiency of Deep Learning Recommendation Models at Scalecited by 43
  4. 2019TensorDIMM: A Practical Near-Memory Processing Architecture for Embeddings and Tensor Operations in Deep Learningcited by 42
  5. 2019Mixed Dimension Embeddings with Application to Memory-Efficient Recommendation Systemscited by 41
  6. 2020Learning to Embed Categorical Features without Embedding Tables for Recommendationcited by 38
  7. 2020Deep Learning Training in Facebook Data Centers: Design of Scale-up and Scale-out Systemscited by 37
  8. 2021Understanding Data Storage and Ingestion for Large-Scale Deep Recommendation Model Trainingcited by 28
  9. 2022RecShard: Statistical Feature-Based Memory Optimization for Industry-Scale Neural Recommendationcited by 28
  10. 2021Accelerating Recommendation System Training by Leveraging Popular Choicescited by 22
  11. 2020Centaur: A Chiplet-based, Hybrid Sparse-Dense Accelerator for Personalized Recommendationscited by 21
  12. 2020Mixed-Precision Embedding Using a Cachecited by 18
  13. 2020Optimizing Deep Learning Recommender Systems' Training On CPU Cluster Architecturescited by 17
  14. 2018Bandana: Using Non-volatile Memory for Storing Deep Learning Modelscited by 14
  15. 2020Developing a Recommendation Benchmark for MLPerf Training and Inferencecited by 12
  16. 2020Time-based Sequence Model for Personalization and Recommendation Systemscited by 11
  17. 2020A Generic Network Compression Framework for Sequential Recommender Systemscited by 10
  18. 2022BagPipe: Accelerating Deep Recommendation Model Trainingcited by 10
  19. 2018In-RDBMS Hardware Acceleration of Advanced Analyticscited by 8
  20. 2019Saec: Similarity-Aware Embedding Compression in Recommendation Systemscited by 7

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 2; 2023: 5; 2024: 2; 2025: 3; 2026: 1.

13 papers.

Sort:

By area:

  1. 2024Keyformer: KV Cache Reduction through Key Tokens Selection for Efficient Generative Inferencecited by 124
  2. 2022A Comprehensive Survey on Trustworthy Recommender Systemscited by 32
  3. 2025Characterizing the Efficiency of Distributed Training: A Power, Performance, and Thermal Perspectivecited by 5
  4. 2022A Frequency-aware Software Cache for Large Recommendation System Embeddingscited by 4
  5. 2023Systems for Parallel and Distributed Large-Model Deep Learning Trainingcited by 4
  6. 2024Integrated Hardware Architecture and Device Placement Searchcited by 4
  7. 2023InTune: Reinforcement Learning-based Data Pipeline Optimization for Deep Recommendation Modelscited by 3
  8. 2023Ad-Rec: Advanced Feature Interactions to Address Covariate-Shifts in Recommendation Networkscited by 3
  9. 2025VectorLiteRAG: Latency-Aware and Fine-Grained Resource Partitioning for Efficient RAGcited by 3
  10. 2023Saturn: An Optimized Data System for Large Model Deep Learning Workloadscited by 2
  11. 2023Evaluating and Enhancing Robustness of Deep Recommendation Systems Against Hardware Errorscited by 1
  12. 2025Faster Distributed Inference-Only Recommender Systems via Bounded Lag Synchronous Collectivescited by 0
1 more
  1. 2026NestPipe: Large-Scale Recommendation Training on 1,500+ Accelerators via Nested Pipeliningcited by 0