News from site!

We are rapidly nearing the 100% mark! Since things are running very well and are yielding excellent results, after hitting this initial goal, the project will continue with a new-and-improved algorithm.

In the (very) short term, we will keep the current algorithm running for a while. Even though we will soon have reached our initial goal, the theoretical range up to which the current code can run is still a ways off, so we are still finding new triples.

Since around october 2008 we have been hard at work designing a new algorithm, incorporating new mathematical ideas and performance tricks. The result is an algorithm that is significantly faster than the current one.

The actual calculation code for this next run is almost complete and mainly requires some final performance tweaks and making sure it runs on a wide range of computers. I'm currently working on updating our backend code to generate and process the new type of workunit, and embedding the calculation code into a BOINC client application.

After that is done, our plan is to re-open ABC@home-Beta to do live testing of the new code. Since the basic infrastructure will be largely the same, I hope we won't need a very long testing phase. Assuming we won't run into large problems during testing, we will then distribute the new application to the main ABC@home project. This should take place sometime in february.

I'll keep you informed on the progress we're making in moving to the new algorithm. Many thanks for your continued help, and a slightly belated happy 2009!

Thanks for the update, JPLiz.