A result from different operating systems and CPU families is (often) not entirely equal, hence no validation can be performed as to which result is correct. WCG sends as per the list provided by knreed the sets to equal platforms thus in a very very large number of results returned, quorum can be established. Those with error get additional copies send out for processing to find a match.

At another project, think it was D@H, had to go as far as also creating CPU groups to ensure results would be verifiable. A result from a AMD would not be equal to one from Intel (similar to the Mac Intel / PPC situation here).

Some explanation is given in this wiki page: http://www.boinc-wiki.info/index.php?title=Homogeneous_Redundancy
There are only 4 projects that are known to use this precision.

As shared, the 'defeat' meant that WCG went from 4 HR groups to 6, complicating the result distribution. Given the Mac share of work (like 4%), it will slow down batch completion for those HR streams. That previously got so bad that AC@H had to be taken off Mac as not enough results would come back timely and caused overall project delay.

For perspective, going from 4 to 6 HR groups meant that the distribution stream, excluding HPF2 had to be expanded from 4x4 to 4x6, meaning 24 way split of work distribution, a 50% increase. Any platform added to WCG presently adds 5 more science apps to code, test, maintain. That's a big stretch. There is no DC grid project like WCG that has to deal with 25 way distribution.

PS

If seeing an Acronym not explained you can find most in a Start Here forum FAQ with links to pages discussing the item. http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=11187

Can someone please explain why the HCC wu's are so large? I can't remember that they were so big before the outtage...

Any ideas on when the Linux situation will be resolved? I set up 6 Linux machines in my garage for the Cancer project. I don't mind that they are working on other projects currently but I would REALLY like to be doing something for cancer.

Thanks!

Lets get some perspective guys:

HCC relaunch was on Wednesday Nov.21 before Thanksgiving at which time a significant bug problem prevented the Launch on Linux.

The guys have come back on Monday and have been ultra silent even in the CA room, so best guess is that they are heads down in search of that reticent grail.

Keep tabs on e.g. this thread: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=17157
Or this forum: http://www.worldcommunitygrid.org/forums/wcg/listthreads?forum=2
by adding them to your Forum Watch threads, and email will be send if anything develops.

cheers

I am still encountering about a 50% error rate on the HCC runs. It appears that the task will stop at or near the 65% completion level. Anyone else having this issue?

Joel