deltavee, there's a well known trick out there, which forces WU's such as these (i.e., with shorter deadlines than normal) to cut in and run ahead of all that's currently on their machine. It worked on mine (and has done for many-a-month), so personally, I'm not sorry. I'd rather see the Beta test complete as quickly as possible (after all, that's the primary concern), than have to be waiting for WU's to eventually turn up.

I got one on a P4-1.6GHz and it completed in 4.6 hours, really remarkable for that old machine that has just 512MB of RAM.

Way faster.
From 1hr.32min to 54 min. on my fastest box.
From 2hr.30min. to 1hr.21min. on my slowest.
Very much on par with Linux.
When do they go into full release.

This is unknown at this point. First we need these results back. Then we will be sending them to the researchers to compare with what they run on their cluster as well as compare it to the batch that finished in production on World Community Grid. To make sure the optimizations didn't change the results we needed these steps to verify.

The batch on World Community Grid has already run and sent back to the researchers. The batch on their cluster is currently running (I do not know the time it takes to run a batch on their cluster).

If it compares, then we will move it to production, but again that is the researchers call and they will let us know.

Hope this helps,
-Uplinger

PS: Thanks again to all for helping on this beta and your experiences you have noticed.

It's a noticeable improvement in speed. Even on an Intel with hyperthreading the results are finishing in under two hours. I do hope you leave at least one project with smaller sized work units such as this one. I have many older computers that are restricted to this project because they tend to error out on more demanding applications.