Skip to content

Dreeseaw/TSNE-CUDA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

14 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

TSNE-CUDA

A CUDA implementation of the t-SNE clustering algorithm

A quick, simple application for quickly creating t-SNE representations of data on NVIDIA GPUs. Meant for use by researchers on Penn State's ACI-ICS high performance compute cluster.

Requirements

Developed and tested using an NVIDIA Quadro K4000 with CUDA 9.1 installed on the PSU ACI-ICS cluster. For more info on the cluster, please visit https://ics.psu.edu/computing-services/.

Download and Use

After downloading to an enviroment with an NVIDIA GPU, make sure you have nvcc findable in your path and make the executable. Running the application requires 3 arguments: the path to the csv file from the current executable, the amount of data points, and the amount of variables in each point (not including label). For example,

./tsne data/mnist_test 1000 784

This will produce an output.txt in your current directory. To produce the final image, run the python script in your directory

python graph.py

Files Included

TSNE-CUDA
|-> data
|---> fashion_mnist-test.csv 
|---> mnist_test.csv
|-> tsne (executable)
|-> tsne.cu (source code)
|-> graph.py (produce graphs using matplotlib and generated output)
|-> Makefile 

Speed

alt text

This project was done with a partner for CMPSC 450, High Performance Computing

About

A CUDA implementation of the t-SNE clustering algorithm

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published