Saved in:
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2508.04701 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909971487129600 |
|---|---|
| author | Yogatama, Bobbi Yang, Yifei Kristensen, Kevin Sarda, Devesh Kim, Abigale Cockcroft, Adrian Teng, Yu Patterson, Joshua Kimball, Gregory McKinney, Wes Gong, Weiwei Yu, Xiangyao |
| author_facet | Yogatama, Bobbi Yang, Yifei Kristensen, Kevin Sarda, Devesh Kim, Abigale Cockcroft, Adrian Teng, Yu Patterson, Joshua Kimball, Gregory McKinney, Wes Gong, Weiwei Yu, Xiangyao |
| contents | The era of GPU-powered data analytics has arrived. In this paper, we argue that recent advances in hardware (e.g., larger GPU memory, faster interconnect and IO, and declining cost) and software (e.g., composable data systems and mature libraries) have removed the key barriers that have limited the wider adoption of GPU data analytics. We present Sirius, a prototype open-source GPU-native SQL engine that offers drop-in acceleration for diverse data systems. Sirius treats GPU as the primary engine and leverages libraries like libcudf for high-performance relational operators. It provides drop-in acceleration for existing databases by leveraging the standard Substrait query representation, replacing the CPU engine without changing the user-facing interface. Sirius achieves 8.3x and 7.4x better cost efficiency on TPC-H and ClickBench, respectively, when integrated with single-node DuckDB, and delivers up to 12.5x speedup when integrated with Apache Doris distributed engine. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_04701 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Rethinking Analytical Processing in the GPU Era Yogatama, Bobbi Yang, Yifei Kristensen, Kevin Sarda, Devesh Kim, Abigale Cockcroft, Adrian Teng, Yu Patterson, Joshua Kimball, Gregory McKinney, Wes Gong, Weiwei Yu, Xiangyao Databases The era of GPU-powered data analytics has arrived. In this paper, we argue that recent advances in hardware (e.g., larger GPU memory, faster interconnect and IO, and declining cost) and software (e.g., composable data systems and mature libraries) have removed the key barriers that have limited the wider adoption of GPU data analytics. We present Sirius, a prototype open-source GPU-native SQL engine that offers drop-in acceleration for diverse data systems. Sirius treats GPU as the primary engine and leverages libraries like libcudf for high-performance relational operators. It provides drop-in acceleration for existing databases by leveraging the standard Substrait query representation, replacing the CPU engine without changing the user-facing interface. Sirius achieves 8.3x and 7.4x better cost efficiency on TPC-H and ClickBench, respectively, when integrated with single-node DuckDB, and delivers up to 12.5x speedup when integrated with Apache Doris distributed engine. |
| title | Rethinking Analytical Processing in the GPU Era |
| topic | Databases |
| url | https://arxiv.org/abs/2508.04701 |