Skip to content
forked from pkusys/Rummy

GPU-accelerated vector query processing system that supports large vector datasets beyond GPU memory.

License

Notifications You must be signed in to change notification settings

SangYeop-Lee/Rummy

 
 

Repository files navigation

0. Introduction

Rummy is a GPU-based vector query processing system for billion-scale datasets beyond GPU memory (with a single GPU). This repository contains one version of the source code. You can find more details about the system design in our NSDI'24 paper: "Fast Vector Query Processing for Large Datasets Beyond GPU Memory with Reordered Pipelining" [Paper].

1. Implementation and reproducing

The core implementation is in DIR and some other files with name *pipe*. The evaluation code and running commands is in <repo>/eval. Please note that all running scripts are in the main branch. You can refer to <repo>/Setup.md to set up the environment.

2. Contact

For any question, please contact zzlcs at pku dot edu dot cn.

The original documentation of Faiss is here.

About

GPU-accelerated vector query processing system that supports large vector datasets beyond GPU memory.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • C++ 51.0%
  • Cuda 29.0%
  • Python 16.3%
  • C 2.0%
  • CMake 1.0%
  • Shell 0.7%