A statistical regression-based GPU design space exploration tool
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Stargazer is a regression-based GPU design space exploration tool. It can be
used to efficiently survey extremely large hardware and software design spaces
of modern GPUs.
For more information, please visit [http://www.princeton.edu/~wjia/stargazer].
Stargazer 1.0, Wenhao Jia [wjia@princeton.edu], Princeton University, 2012

This is a list of important files included in the package. Scripts have
embedded comments which further explain their usage.

1. stargazer.R: The core stepwise regression method.
2. matrix.csv: Example input file from simulating a matrix multiply program.
3. matrix-training.csv: 60% of the matrix.csv file, randomly selected.
4. matrix-test.csv: 40% of the matrix.csv file, randomly selected.
5. gpgpusim/README: Directions for how to use Python scripts in this folder.
6. gpgpusim/simulate.py: Run GPGPU-Sim multiple times with varying parameters.
7. gpgpusim/collect.py: Collect simulation results and generate CSV files.
8. gpgpusim/split.py: Split a CSV file into a traing set and a test set.
9. gpgpusim/filter.py: Scan a CSV file s.t. only lines w/ certain values remain.
10. R/README: Directions for how to use R scripts in this folder.
11. R/accuracy.R: Measure model prediction accuracy w/ training and test sets.
12. R/size.R: Show how to obtain the prediction accuracy vs. sample size curve.
13. R/time.R: Demonstrates how to time the regression process.