[ImageJ-devel] Fwd: OpenCL plugin

[Rick Lentz]

Hi Pol,

  Thank you for your feedback regarding OpenCL in ImageJ.  Per your second
question, there are some very minor changes needed to get your started on
your development of ImageJ OpenCL plugins.  I have updated the source
documentation to be reflective of the current state of the source available
in the loci repository (
http://dev.loci.wisc.edu/svn/software/branches/maven/projects/opencl-decon).

  If your refresh your download of the opencl-decon source, you may notice
the addition of a README file.  Here is the text from that file:

( Please ensure installation of OpenCL prior to running these programs )
The source at
http://dev.loci.wisc.edu/svn/software/branches/maven/projects/opencl-deconwas
developed to demonstrate successful leveraging of OpenCL within ImageJ
using JOCL both locally and remotely (as a binary web service).

The folder structure of the source consists of the following folders:
src - Java and OpenCL source files (extension .cl)
sourcedata - PSF and 3D data used as a small sample data set for the FHT3D
Example.
lib - libraries needed for classes using JOCL, ImageJ, and Hessian 4.0.7
native - contains the dynamic libraries needed for runtime execution of
OpenCL code on various operating systems

Background:
To use OpenCL from Java in ImageJ we leverage JOCL.  JOCL uses JNI to make
calls into the OpenCL API.  The OpenCL code you write can also leverage JOCL
to accelerate execution of ImageJ plugins from Java.

Notice the files fht.cl and sobel.cl in the source (src) directory.  The
Java code in provided in the examples compile these two files for execution
on the OpenCL enabled device that is programmatically chosen at runtime.  It
is runtime compilation of OpenCL that allows execution on any potential
OpenCL enabled device.

Start exploring the examples by viewing the developer comments in the file
SobelFilterExample.java.  Notice the Main() method calls run() which use an
awt.Image type as an input parameter.  Modify and run the Main() method as a
Java application with the VM Arguments -Xmx1024m.

  Without modification, SobelFilterExample.java loads an image from a web
server, process it locally using OpenCL, and displays the results.  There is
nothing novel about this example.  It simply allows runtime testing
of several system configuration steps to ensure working configuration of
JOCL and OpenCL native libraries.  Modify this example to suite your needs,
but please ensure proper JOCL and OpenCL configuration before proceeding.

Working within ImageJ: If developing an ImageJ plugin using OpenCL realize
that programatic control is passed to your plugin inside the PlugIn (or
PluginFilter) run() method.  An example of this can be found in
src.demos.OpenCL_SobelFilter.java.  For this plugin to run within ImageJ,
the JOCL jars and native libraries respective to the target platform will
need to be available by the ImageJ classloader.   The supporting JOCL native
libraries can be copied into the plugin directory within ImageJ to allow
plugin implementations using OpenCL from Java to reference the installed
native libraries provided by the OpenCL installation.

Now that you have demonstrated use of OpenCL from Java and within ImageJ,
you may wish to see an a compute intensive example demonstrating
modification of an existing Java implementation that delegates a portion of
its implementation to OpenCL.  Take a look at the developer comments in the
FHT3D_3D_Deconvolution.java example to see what steps are used for brokering
of data between Java and OpenCL between steps within an algorithm's
implementation.

Finally, the need may arise where the GPU compute capabilities you wish to
leverage are not on the same computer that ImageJ is running on.  The
FHTEJBService and Iterative_Deconvolve_3D_WS classes demonstrate how to
remotely serve up the functionality leveraging standard, open source J2EE
technologies.

I will comment in a separate email regarding your first question.

I am interested in following your progress.  Please feel free to contact me
via Google chat, Skype or email if you run into any more configuration
problems.

Best Regards,

Rick Lentz



Hello,
I am a PhD student working on image segmentation by texture analysis with an
application on remote sensing images (Montpellier, France). So i have to
process very large images (e.g. 15000*8000).
All methods I developed are formed as plugin under ImageJ API that I really
like, but not yet applicable on this sort of image.
Recently, I tough to use finally my graphic card (Nvidia GTS240) which I
think will help me a lot in my process. So after several search on the web,
I found this (OpenCL plugin you wrote) great ! it's precisely what i need.
So I decided to follow your tutorial to setup the environment but met some
problems... :

When launching makefile in "~/NVIDIA_GPU_Computing_SDK/C/ " and
"~/NVIDIA_GPU_Computing_SDK/OpenCL" I had following error "/usr/bin/ld:
cannot find -lGL". Plus, I dont have the same demo files, maybe because of
version ? (cudatoolkit_3.2.16_linux_64, gpucomputingsdk_3.2.16 and
devdriver_3.2_linux_64_260.19.26 on Ubuntu 10.10). On the other side demos I
have work perfectly (ooclDCT8x8, oclHistogram,...)
Main problem is that the repository you indicate to get the example project
(http://www.loci.wisc.edu/svn/decon) doesn't work... "
http://dev.loci.wisc.edu/svn/software/branches/maven/projects/opencl-decon"
repository works well. Is is the same ?

Thanks a lot if you find time to answer me (and sorry for my english).
And thanks you for work you are doing on ImageJ.
Best regards,
Pol Kennel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://imagej.net/pipermail/imagej-devel/attachments/20110111/8fde4862/attachment.html>