CUDA_Image_Convolution

Orig Author: Alan Reiner Date: 01 September, 2010 Email: [email protected]

This is my first stab 2D convolution using CUDA. It's pretty good, it does a 4096x4096 array of floating point (grayscale) values with an arbitrary 15x15 PSF in about 125 ms (plus 85ms of memory copies).

There is no interface to use it. It must be used by modifying the main() code and recompiling for each use. At some point in the future I will organize it into a library that can be linked by other projects.

Supported Hardware:

This code was designed for NVIDIA devices of Compute Capability 2.0+ which is any NVIDIA GTX 4XX series card (Fermi). The code should compile and run on 1.X devices, but the code is not optimized for them so there may be a huge performance hit. Maybe not. (see note below about compiling for 1.X GPUs)

I believe NVIDIA 8800 GT is the earliest NVIDIA card that supports CUDA, and it would be Compute Capability 1.0.

CUDA was created by NVIDIA, for NVIDIA. It will not work ATI cards. If you want to take up GPU-programming on ATI, the only two options I know are OpenGL and OpenCL. However, the maturity of those programming interfaces are unclear (but at least such programs can be run on both NVIDIA and ATI)

Installing and running:

This directory should contain everything you need to compile the image convolution code, besides the CUDA 3.1 toolkit which can be downloaded from the following website:

  http://developer.nvidia.com/object/cuda_3_1_downloads.html

In addition to installing the toolkit, you might need to add libcutil.a and libshrutil.a to your LD path. In linux, it is sufficient to copy them to your /usr/local/cuda/lib[64] directory.

I personally work with space-separated files for images, because they are trivial to read and write from C++. Additionally, I use MATLAB to read/write these files too, which makes it easy create input for the CUDA code, and verify output.

There is no reason this code won't work on Windows, though I've never tried it.

I strongly urge anyone trying to learn CUDA to download the CUDA SDK. It contains endless examples of every CUDA feature in existence. Once you know what to do, you can get the syntax from there, which more often than not is very ugly.

To work with older NVIDIA cards (8800+ GT, 9XXX GT, GTX 2XX):

Open commom/common.mk and around line 150, uncomment the GENCODE_SM10 line, which enables dual-compilation for CUDA Compute Capability 1.X devices. The reason I commented this out is that 1.X devices don't support printf(), which is my main method for debugging. Therefore, leaving this line in the Makefile prevents the code from compiling when I'm attempting to debug.

Planned Updates

• My current project requires operation on images represented by floating-point values, hence why all the methods are for floats. HOWEVER, most image processing is done on 8-bit data, which would use 1/4 the bandwidth, and integer math is a bit faster than floating point operations

• Bilateral filter is only a few steps away, just have to verify the operations and memory allocations

• I need to break out separate files and functions for unit testing, instead of just modifying main() for each function

My test image is salt4096.txt, which is a binary input image I've been using to test the convolution routines (isolated dots make it obvious whether your PSF was applied correctly). However, because of its size, I have not committed it. I added salt512.txt just for this.