Flash-LLM: Enabling Cost-Effective and Highly-Efficient Large Generative Model Inference with Unstructured Sparsity

Summary: Flash-LLM: tensor-core framework using “Load-as-Sparse, Compute-as-Dense” SpMM to exploit unstructured sparsity by trading redundant compute to cut memory bandwidth and improve tensor-core utilization. Delivers 1.5–2.9× SpMM and up to 3.8× end-to-end tokens/sec vs Sputnik, SparTA, DeepSpeed, FasterTransformer. (summarized by gpt-5-mini on Feb 09 2026)

Paper ID: 13443
Venue: VLDB
Year: 2024
Pagerank: 4.8154191e-05
Overall Rank: 7,152 | 50.25%
DOI: 10.14778/3626292.3626303

Incoming Non-self Citations Over Time

Authors

Incoming Citations (Sorted by Pagerank)

Showing 3 of 3 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
8,520	mLoRA: Fine-Tuning LoRA Adapters via Highly-Efficient Pipeline Parallelism in Multiple GPUs	2025	VLDB	4.4937074e-05
9,677	Apt-Serve: Adaptive Request Scheduling on Hybrid Cache for Scalable LLM Inference Serving	2025	SIGMOD	4.3047774e-05
10,492	Malleus: Straggler-Resilient Hybrid Parallel Training of Large-scale Models via Malleable Data and Model Parallelization	2025	SIGMOD	4.1945683e-05

Outgoing Citations (Sorted by Pagerank)

Showing 11 of 11 cited papers.

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

Rank	Cited Paper	Year	Venue	Pagerank
411	PyTorch Distributed: Experiences on Accelerating Data Parallel Training	2020	VLDB	0.00023906921
517	Can Foundation Models Wrangle Your Data?	2023	VLDB	0.00021169035
2,290	TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Data	2022	VLDB	9.0934125e-05
2,352	MiCS: Near-linear Scaling for Training Gigantic Model on Public Cloud	2023	VLDB	8.9766205e-05
3,363	CROSSBOW: Scaling Deep Learning with Small Batch Sizes on Multi-GPU Servers	2019	VLDB	7.1731921e-05
4,934	From BERT to GPT-3 Codex: Harnessing the Potential of Very Large Language Models for Data Management	2022	VLDB	5.8198826e-05
5,333	Heterogeneity-Aware Distributed Machine Learning Training via Partial Reduce	2021	SIGMOD	5.5656575e-05
6,156	Optimizing Tensor Programs on Flexible Storage	2023	SIGMOD	5.1802603e-05
6,377	Galvatron: Efficient Transformer Training over Multiple GPUs Using Automatic Parallelism	2023	VLDB	5.0911095e-05
8,607	Harmony: Overcoming the Hurdles of GPU Memory Capacity to Train Massive DNN Models on Commodity Servers	2022	VLDB	4.4855009e-05
8,808	FlexMoE: Scaling Large-scale Sparse Pre-trained Model Training via Dynamic Device Placement	2023	SIGMOD	4.4454035e-05

Semantically Similar Papers

Overall Rank	Paper	Year	Venue	Pagerank
10,167	FlashANNS: GPU-Driven Asynchronous I/O Pipelining for Eliminating Storage-Compute Bottlenecks in Billion-Scale Similarity Search	2026	SIGMOD	4.1945683e-05
10,020	HotPrefix: Hotness-Aware KV Cache Scheduling for Efficient Prefix Sharing in LLM Inference Systems	2026	SIGMOD	4.1945683e-05
7,301	Randomized Algorithms Accelerated over CPU-GPU for Ultra-High Dimensional Similarity Search	2018	SIGMOD	4.768971e-05
10,122	TranSQL+: Serving Large Language Models with SQL on Low-Resource Hardware	2026	SIGMOD	4.1945683e-05
8,808	FlexMoE: Scaling Large-scale Sparse Pre-trained Model Training via Dynamic Device Placement	2023	SIGMOD	4.4454035e-05
10,595	Optimized Batch Prompting for Cost-effective LLMs	2025	VLDB	4.1945683e-05
7,339	SpareLLM: Automatically Selecting Task-Specific Minimum-Cost Large Language Models under Equivalence Constraint	2025	SIGMOD	4.7579469e-05
13,138	Database Perspective on LLM Inference Systems	2025	VLDB	-
10,222	RetroInfer: A Vector Storage Engine for Scalable Long-Context LLM Inference	2026	VLDB	4.1945683e-05
9,805	MEMO: Fine-grained Tensor Management For Ultra-long Context LLM Training	2025	SIGMOD	4.2805224e-05