​

• 2D convolution(Image Convolution) is an important concept majorly used in image processing. 2D convolution takes one input image and generate an output image.  One input image acts as the input matrix. Other matrix(called as kernel) acts as a filter on the input image.In the 2D convolution, we slide the filter matrix(kernel matrix) over each pixel of the input image matrix. 

​

• Implemented 2D convolution in CUDA with any two input matrices. For example: Image matrix is 2048 x 2048 elements and Kernel matrix  is 7 x 7 elements.

       

• Compared the theoretical maximum attainable performance(in Giga Floating point operations per second) of a) Kernel Code b) Algorithm on Intel Core i7 920.