Skip to content

Command line program. Comparing histogram computing time on GPU (CUDA) and CPU.

Notifications You must be signed in to change notification settings

eMKa007/CUDA-Histogram

Repository files navigation

GPU vs CPU - histogram computing time.

Command line program. Main idea was to compare computing time of histogram on GPU (CUDA) and CPU.

CUDA uses shared memory to avoid bottlenecks in atomic operations.

CPU uses simple loop through all the pixels in the image.

NOTE: This program is intended to be use with nVidia GPU devices. Be sure you are able to run CUDA C code on your machine.

Inside this repository there is example image "atol.jpg". It is pretty big ( 5160x2696px ), but it is good candidate as a testing image.

USAGE:

Histogram_Compare.exe <imageName.jpg> <NumberOfExecutions>
Where: 
	<imageName.jpg> - path to input image.
    <NumberOfExecutions> - Number of tests computing time.
    
Tips: 
	It is good to locate input image at the same place as *.exe file. 
    Bigger NumberOfExecutions <int> cause more computing time. Optimal amount
    is about 250-500. Max value = 1000. Can not be < 0. 

Example Output:

alt text

About

Command line program. Comparing histogram computing time on GPU (CUDA) and CPU.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published