Skip to content

architecture-research-group/ae-asplos25-iks

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

25 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ae-asplos25-iks

Cycle-approximate simulator

Directory: cycle_approx_iks. Details on the simulator are provided in the README of this directory.

GPU implementation:

Directory: gpu_faiss.

For comparison with IKS, we run exact search on up to 8 GPUs for various corpus sizes. See gpu_faiss. Run ./setup.sh to install Conda and FAISS with GPU support. After this, run with python main.py. The maximum number of GPUs can be set with --max_gpus.

CPU implementation

Directory: ae-aslpos25-iks-faiss. This is a submodule, so run git submodule init and git submodule update.

The default parameters of FAISS are suboptimal on many current systems. To address this, we use Intel MKL as the backend, and increase the MKL block size to 16384. A source file for a program generating the values for Fig.9 is located in the ae_asplos25 directory. Run ./ae_setup.sh /path/to/mkl/latest/lib to run Cmake with the relevant options, and run ./ae_fig_9.sh to build and run the program for figure 9.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors