BOINC@AUSTRALIA FORUM

  • December 15, 2019, 04:58:48 AM
  • Welcome, Guest
Please login or register.

Login with username, password and session length
Advanced search  

News:

We have a Twiiter Account for the Team https://twitter.com/BOINCAUSTRALIA

Pages: [1]   Go Down

Author Topic: Project News  (Read 3289 times)

Cruncher Pete

  • Guest
Project News
« on: May 10, 2009, 05:05:19 PM »

From Forum Mod of Superlink@Technion:

10 May 09:

Friends, here's the long awaited report on how the last batches performed.

First, as always, I am grateful for the great participation and the interest in the project!

Let me split this update into two parts. First, as for the "crunching" ability of the system - we've got truly great results. We didn't break our own record in terms of the throughput, but it is less important. What important is that the throughput was quite stable, there were no glitches during the run, and the server worked well. This is the good news. The bad news is that the two batches ( the one before the last, and the last used to confirm the results) produced different results. As it turned out, out of 2.5 mln jobs, there were 3 jobs returned by DIFFERENT computers, that passed our rather strict sanity checks, but in fact were COMPLETELY wrong. As you know we're not using redundancy since we cannot afford wasting the CPU power. And during the last year we have been seeing several cases of incorrect results, but all these were filtered out by our sanity checks without using the redundancy. The sad conclusion is to add another level of sanity checks and, sometimes, redundancy, by that increasing the total amount of computations.

Now to the biological perspective. The batch produced great results. It confirmed the suspected region on a chromosome with the mutation causing ARCC (Age-Related Corticol Cataract). This narrows down the region and will hopefully allow geneticists to find the mutation. If found, we will post the reference to the respective paper, but it may take a while.. Actually, the statistical analysis carried out by Superlink@Technion is the easiest part.

Currently we are not expecting any large batches, maybe by the beg. of June, the earliest.
Logged
Pages: [1]   Go Up
 

Page created in 0.045 seconds with 18 queries.