DBScholar

Back to papers

FastFlow: Accelerating Deep Learning Model Training with Smart Offloading of Input Data Pipeline

Summary: FastFlow automatically mitigates CPU-side input-pipeline bottlenecks by smartly offloading preprocessing to remote CPUs and jointly leveraging local+remote resources to maximize GPU utilization. Integrated into TensorFlow, its performance-driven offloading policy yields 1–4.5× throughput gains vs TensorFlow/tf.data.service and up to 2.06× vs DALI. (summarized by gpt-5-mini on Feb 09 2026)

Paper ID
he6fabaddc51fdf23
Venue
VLDB
Year
2023
Pagerank
6.9931619e-05
Overall Rank
3,800 | 74.53%
DOI
10.14778/3579075.3579083
PDF
Download (CC BY-NC-ND 4.0)

Incoming Non-self Citations Over Time

Authors

BibTeX Citation

@article{um_vldb23,
        title = {{FastFlow: Accelerating Deep Learning Model Training with Smart Offloading of Input Data Pipeline}},
        author = {Um, Taegeon and Oh, Byungsoo and Seo, Byeongchan and Kweun, Minhyeok and Kim, Goeun and Lee, Woo-Yeon},
        journal = {PVLDB},
        series = {{VLDB} '23},
        volume = {16},
        number = {5},
        pages = {1086--1099},
        doi = {10.14778/3579075.3579083},
        url = {https://doi.org/10.14778/3579075.3579083},
        year = {2023}
}

Incoming Citations (Sorted by Pagerank)

Showing 9 of 9 citing papers.

Rank Citing Paper Year Venue Pagerank
3,223 How Large Language Models Will Disrupt Data Management 2023 VLDB 7.5043163e-05
4,910 Apt-Serve: Adaptive Request Scheduling on Hybrid Cache for Scalable LLM Inference Serving 2025 SIGMOD 6.3468554e-05
7,120 DLRover-RM: Resource Optimization for Deep Recommendation Models Training in the Cloud 2024 VLDB 5.580601e-05
7,848 FusionFlow: Accelerating Data Preprocessing for Machine Learning with CPU-GPU Cooperation 2024 VLDB 5.4236801e-05
9,097 TensorSocket: Shared Data Loading for Deep Learning Training 2026 SIGMOD 5.208561e-05
9,099 Scheduling Data Processing Pipelines for Incremental Training on MLP-based Recommendation Models 2025 SIGMOD 5.208561e-05
10,016 MEMO: Fine-grained Tensor Management For Ultra-long Context LLM Training 2025 SIGMOD 5.0825892e-05
10,040 cedar: Optimized and Unified Machine Learning Input Data Pipelines 2025 VLDB 5.0786422e-05
11,314 GPEmu: A GPU Emulator for Faster and Cheaper Prototyping and Evaluation of Deep Learning System Research 2025 VLDB 4.9605343e-05
Previous Page 1 / 1 Next

Outgoing Citations (Sorted by Pagerank)

Showing 4 of 4 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank Cited Paper Year Venue Pagerank
1,472 Analyzing and Mitigating Data Stalls in DNN Training 2021 VLDB 0.00010529036
2,062 tf.data: A Machine Learning Data Processing Framework 2021 VLDB 9.0820315e-05
3,598 Where Is My Training Bottleneck? Hidden Trade-Offs in Deep Learning Preprocessing Pipelines 2022 SIGMOD 7.1596688e-05
6,318 Progressive Compressed Records: Taking a Byte out of Deep Learning Data 2021 VLDB 5.7964881e-05
Previous Page 1 / 1 Next

Semantically Similar Papers

# Overall Rank Paper Year Venue
1 5,410 GoldMiner: Elastic Scaling of Training Data Pre-Processing Pipelines for Deep Learning 2023 SIGMOD
2 5,011 Galvatron: Efficient Transformer Training over Multiple GPUs Using Automatic Parallelism 2023 VLDB
3 9,429 MemFlow: Memory-Aware Distributed Deep Learning 2020 SIGMOD
4 10,415 ZenFlow: Enabling Stall-Free Offloading for LLM Training 2027 SIGMOD
5 3,770 CROSSBOW: Scaling Deep Learning with Small Batch Sizes on Multi-GPU Servers 2019 VLDB
6 10,513 DFLOP: A Data-driven Framework for Multimodal LLM Training Pipeline Optimization 2026 SIGMOD
7 9,097 TensorSocket: Shared Data Loading for Deep Learning Training 2026 SIGMOD
8 2,062 tf.data: A Machine Learning Data Processing Framework 2021 VLDB
9 3,598 Where Is My Training Bottleneck? Hidden Trade-Offs in Deep Learning Preprocessing Pipelines 2022 SIGMOD
10 7,848 FusionFlow: Accelerating Data Preprocessing for Machine Learning with CPU-GPU Cooperation 2024 VLDB