The following animated gif shows the algorithm in action. The left side is the generated Mandelbrot fractal and the right side is a visualization of the activity detection algorithm. The algorithm is choosing how to zoom into the fractal in real-time (when the image was recorded).

I’ve also posted the Matlab code if you are interested in seeing how it works.

I’ve tried some alternative non-CAPTCHA filters, but none of them seem very reliable (some very simple tests were giving false positives).

On a (non-false) positive note, since this is my spring break, I’m trying to get some of my recent projects together and post them this week

For those of you who don’t know, PyMW is a Master-Worker computing framework in Python. It wraps several other Master-Worker frameworks such as MPI, Condor, BOINC or even just using multi-core processors, and exposes them as a simple and elegant API.

The way it works is, you create tasks and submit them to a Master and the master uses an interface/wrapper (BOINC, Condor, MPI, etc) to process those tasks. The master distributes the tasks out to compute nodes (or processor cores), called Workers, and the results get sent back to the Master. This allows you to debug using the multi-core interface (a single machine) and then, by changing one switch on the command line, you can run the same code on thousands of machines using BOINC or any other supported interface.

My proposal was to improve BOINC integration with PyMW by 1) eliminating the startup script and the need to compile C code; 2) adding pure-Python support for BOINC Assimilators and Validators; and 3) by adding a new checkpointing mechanism for long running jobs (optional).

I completed (2) very quickly by virtue of some old Python code I found in BOINC, but (1) took a lot of sweat and tears — the existing BOINC interface was functional, but in need of some serious work. It was running under Mac and Linux, however the Windows client was crashing on every task. To get it working under Windows I ended up creating a C++ launcher application to avoid using batch scripts.

In addition to my proposal, I also added a few other tasks. When working with BOINC, the existing interface assumed that Python was already installed and on the system PATH environment variable. This is not a very safe assumption and so I also crated a portable Python interpreter integrated with the BOINC API. PyMW will now install this interpreter as your BOINC application so that clients no longer need to have Python installed to run PyMW-BOINC compute jobs. Along the way I also created a new logo and graphic design for the PyMW web site and setup WordPress, check it out.

Sadly goal (3) was not completed, however, it was originally proposed as an optional part of the project. I actively chose to pursue the BOINC-Python interpreter over (3) because I felt it was more important to the BOINC interface. In the end, checkpointing can always be done manually, but sending the Python interpreter is a considerably harder task.

I am still wrapping up a few odds and ends, but for the most part, I feel very happy with the state of the BOINC interface. If you are looking for a distributed or parallel processing framework for a future project, please consider PyMW.

If you have any questions or comments, I would love to hear them!

The problem is that when running an application on BOINC, you have no guarantee of what libraries will be available, so you must either distribute the libraries you need or compile them statically. I chose the later, which also caused lots of issues with the Python standard library. It’s mostly working now, with the exception of the sqlite module.

I moved the repo over to my bit bucket account since Nicolas is busy with other projects. The latest code can be found here.

The project will likely be incorporated into the BOINC trunk, but the current code is available on bitbucket as PyBOINC if you are interested in seeing how it works.

The Python developers made embedding the interpreter incredibly easy from C/C++ and the BOINC API is readily available from C/C++ as well, so there really isn’t much code. However, packaging the Python standard library and compiling it so it runs on multiple platforms are still going to be challenging.

This has again exposed the BOINC file immutability issue and I can see now that there is no hacking my way around it — I am going to have to change the interface so that all executables and WorkUnits are uniquely named.

This hasn’t been an issue up until now, mainly because I was running the PyMW example applications and not modifying them between executions. However, with the introduction of PyMW data bundles, this problem has become painfully obvious. Since data bundles are given a temporary file name and this file name is dynamically embedded into the body of the executable, the executable is now changing its contents on every run.

The fact that the file is changing and the file name remains the same means that BOINC keeps only one copy of the file (because of file name immutability/versioning). The end result is that when a work unit executes on a worker machine, it tries to open the first data bundle file name that was ever created because that first file was cached and never updated.

To fix this, I’ve added some code into the BOINC interface that deletes all work unit related files from the BOINC “download” directory on every execution. This has fixed the problem for now, but the interface should rename all files to a unique name before execution. This is one of my goals for the next iteration of the BOINC interface.

This is really no different from handling exceptions in non-distributed code, but I know all to well how exceptions are normally handled (hint: they aren’t). If you are interested, you can read my full post on the PyMW blog.

The interface is working well now, but BOINC work unit failures still require manual user intervention. I would like to automate as much failure recovery as possible, so I will be focusing on this for the next few days.