Evaluating Multi-GPU Sorting with Modern Interconnects

Summary: Evaluates multi-GPU sorting across PCIe/NVLink/NVSwitch; proposes a P2P GPU-only sort and a heterogeneous sort, benchmarked on three modern platforms. Reports up to 35x higher P2P throughput with NVSwitch, up to 14x CPU radix-sort speedup (P2P) and 9x (HET); on fast interconnects P2P beats HET by ~1.65x, and copy/compute overlap does not hide transfer bottlenecks. (summarized by gpt-5-nano on Feb 09 2026)

Paper ID: 6356
Venue: SIGMOD
Year: 2022
Pagerank: 5.7777896e-05
Overall Rank: 6,775 | 53.52%
DOI: 10.1145/3514221.3517842

Incoming Non-self Citations Over Time

Authors

1. Tobias Maltenberger (Hasso Plattner Institute; University of Potsdam)
2. Ivan Ilic (Hasso Plattner Institute; University of Potsdam)
3. Ilin Tolovski (Hasso Plattner Institute; University of Potsdam)
4. Tilmann Rabl (Hasso Plattner Institute; University of Potsdam)

BibTeX Citation

@inproceedings{maltenberger_sigmod22,
        title = {{Evaluating Multi-GPU Sorting with Modern Interconnects}},
        author = {Maltenberger, Tobias and Ilic, Ivan and Tolovski, Ilin and Rabl, Tilmann},
        series = {{SIGMOD} '22},
        booktitle = {Proceedings of the {ACM} {SIGMOD} International Conference on Management of Data},
        publisher = {Association for Computing Machinery},
        doi = {10.1145/3514221.3517842},
        url = {https://dl.acm.org/doi/10.1145/3514221.3517842},
        year = {2022}
}

Incoming Citations (Sorted by Pagerank)

Showing 9 of 9 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
4,535	Triton Join: Efficiently Scaling to a Large Join State on GPUs with Fast Interconnects	2022	SIGMOD	6.6418298e-05
5,423	GPU Database Systems Characterization and Optimization	2024	VLDB	6.2244466e-05
6,066	Vortex: Overcoming Memory Capacity Limitations in GPU-Accelerated Large-Scale Data Analytics	2025	VLDB	5.9894416e-05
7,059	BOSS - An Architecture for Database Kernel Composition	2024	VLDB	5.7153841e-05
7,904	Terabyte-Scale Analytics in the Blink of an Eye	2026	VLDB	5.5180253e-05
8,443	Analyzing Vectorized Hash Tables Across CPU Architectures	2023	VLDB	5.4242976e-05
9,333	Efficiently Joining Large Relations on Multi-GPU Systems	2025	VLDB	5.2886781e-05
9,511	DPDPU: Data Processing with DPUs	2025	CIDR	5.2572834e-05
10,580	GPU Acceleration of SQL Analytics on Compressed Data	2026	VLDB	5.0935618e-05

Outgoing Citations (Sorted by Pagerank)

Showing 18 of 18 cited papers.

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

Rank	Cited Paper	Year	Venue	Pagerank
217	SIMD-Scan: Ultra Fast in-Memory Table Scan using on-Chip Vector Processing Units	2009	VLDB	0.00024465286
237	Amazon Redshift and the Case for Simpler Data Warehouses	2015	SIGMOD	0.00023698605
252	Multi-Core, Main-Memory Joins: Sort vs. Hash Revisited	2014	VLDB	0.00023240918
389	One Trillion Edges: Graph Processing at Facebook-Scale	2015	VLDB	0.00019386061
423	Massively Parallel Sort-Merge Joins in Main Memory Multi-Core Database Systems	2012	VLDB	0.00018724714
428	HYRISE—A Main Memory Hybrid Storage Engine	2011	VLDB	0.00018631602
575	Efficient Transaction Processing in SAP HANA Database – The End of a Column Store Myth	2012	SIGMOD	0.00016256534
678	Fast Sort on CPUs and GPUs: A Case for Bandwidth Oblivious SIMD Sort	2010	SIGMOD	0.00015059987
1,177	A Comprehensive Study of Main-Memory Partitioning and its Application to Large-Scale Comparison- and Radix-Sort	2014	SIGMOD	0.00011807196
1,466	A Study of the Fundamental Performance Characteristics of GPUs and CPUs for Database Analytics	2020	SIGMOD	0.00010689254
1,504	Self-Tuning, GPU-Accelerated Kernel Density Models for Multidimensional Selectivity Estimation	2015	SIGMOD	0.00010556246
2,567	Pump Up the Volume: Processing Large Data on GPUs with Fast Interconnects	2020	SIGMOD	8.4115337e-05
2,668	A Memory Bandwidth-Efficient Hybrid Radix Sort on GPUs	2017	SIGMOD	8.2755141e-05
3,134	Efficient Join Algorithms For Large Database Tables in a Multi-GPU Environment	2021	VLDB	7.7229903e-05
3,226	MG-Join: A Scalable Join for Massively Parallel Multi-GPU Architectures	2021	SIGMOD	7.6216779e-05
4,022	PARADIS: An Efficient Parallel Algorithm for In-place Radix Sort	2015	VLDB	6.9494096e-05
4,177	SIMD- and Cache-Friendly Algorithm for Sorting an Array of Structures	2015	VLDB	6.8491535e-05
6,836	GPU-accelerated data management under the test of time	2020	CIDR	5.7584085e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	3,134	Efficient Join Algorithms For Large Database Tables in a Multi-GPU Environment	2021	VLDB
2	827	High-Performance Sorting on Networks of Workstations	1997	SIGMOD
3	712	Efficient Implementation of Sorting on Multi-Core SIMD CPU Architecture	2008	VLDB
4	7,370	Efficient Top-K Query Processing on Massively Parallel Hardware	2018	SIGMOD
5	6,836	GPU-accelerated data management under the test of time	2020	CIDR
6	305	GPUTeraSort: High Performance Graphics Co-processor Sorting for Large Database Management	2006	SIGMOD
7	678	Fast Sort on CPUs and GPUs: A Case for Bandwidth Oblivious SIMD Sort	2010	SIGMOD
8	2,567	Pump Up the Volume: Processing Large Data on GPUs with Fast Interconnects	2020	SIGMOD
9	9,333	Efficiently Joining Large Relations on Multi-GPU Systems	2025	VLDB
10	2,668	A Memory Bandwidth-Efficient Hybrid Radix Sort on GPUs	2017	SIGMOD