Message boards :
Number crunching :
"In progress" means ??
Message board moderation
| Author | Message |
|---|---|
|
Send message Joined: 13 Jul 05 Posts: 456 Credit: 75,142 RAC: 0 |
On the home page we have
What exactly does "in progress" mean here? 27 WU that have not yet made quorum? 27 WU that have not yet been validated (but which might have the quorum and be waiting for the validator? Or 27 WU with results that are not yet returend or timed out? With redundant replication (ie producing more results in the initial replication than are needed to form quorum) a WU could be complete in the sense of having a canonical result to pass back to the scientists, but still not yet complete in that every result has completed. Just curious. Anyone know? River~~ 
|
 Ocean Archer Send message Joined: 13 Jul 05 Posts: 143 Credit: 263,300 RAC: 0 |
River -- I'm not an expert on these things, but my humble opinion is that there are 27 WUs out there still to be returned. Hopefully they will get back soon so we can move on to the next portion of the project ... If I've lived this long, I've gotta be that old |
|
Send message Joined: 4 Sep 05 Posts: 112 Credit: 1,864,470 RAC: 0  
|
I think that would include WU's that have a quorum but have one or more results still not returend or timed out. Click here to join the #1 Aussie Alliance on LHC. |
|
Send message Joined: 26 Sep 05 Posts: 85 Credit: 421,130 RAC: 0
|
|
|
Send message Joined: 17 Sep 05 Posts: 60 Credit: 4,221 RAC: 0 |
****************Duplicate Post**************** |
|
Send message Joined: 17 Sep 05 Posts: 60 Credit: 4,221 RAC: 0 |
I think River's question would be similiar to mine. I have been wondering for awhile how they count the workunits on the front page. Each workunit is sent out around 3-5 times. Does the count on the front page include each workunit being sent out multiple times? **Edit** To answer River's Question I would make the educated guess that "in progress" means that the workunit has been sent out but that the computer has not returned the workunit yet. (Just what Archer said) |
|
Keck_Komputers Send message Joined: 1 Sep 04 Posts: 275 Credit: 2,652,452 RAC: 0 |
"Workunits in progress" means the number of workunits that have not been validated yet. If it said "results in progress" then it would be counting the number sent to individual hosts and not returned yet. At this project there are normally 5 results per workunit. What that will say and mean when the new terms come out I am not sure. It may not change though since SETI is still using "results" and "workunits" on their website, they should have the current server package out. BOINC WIKI   BOINCing since 2002/12/8 |
|
Send message Joined: 17 Sep 05 Posts: 60 Credit: 4,221 RAC: 0 |
"Workunits in progress" means the number of workunits that have not been validated yet. If it said "results in progress" then it would be counting the number sent to individual hosts and not returned yet. At this project there are normally 5 results per workunit. ------------------------------------------------------- Wow, thanks Keck. That means when this project has 100,000 workunits there are 500,000 results to crunch (or a number pretty close to that). That is a mind boggling amount of work to fathom. |
|
Send message Joined: 13 Jul 05 Posts: 456 Credit: 75,142 RAC: 0 |
... That means when this project has 100,000 workunits there are 500,000 results to crunch (or a number pretty close to that). That is a mind boggling amount of work to fathom. not quite. If I understand Keck's post then 100,000 WU could be anything from 500,000 results out there if none of those WU have any returned results, down to just 100,000 if they are all waiting fo rthat last result to come home, or even 0 if they'd all been returned but the validator was asleep. That would explain why the counts do not reduce vary fast in the first day or so of a new batch - as results come back they go into the pending count, and the WU count is nt reduced. Later as results alow validation to happen, the pendng counts go down, and so do the WU counts on the front page Thanks to everyone for you answers to my question. R~~ |
 Send message Joined: 30 Jan 06 Posts: 10 Credit: 1,947 RAC: 0 |
When there is such a small number remaining why not just send these work units out to a few more computers to speed things up a bit? Or does it not really matter because the next batch isn't ready to be sent out to us anyway? Founder of BOINC group Objectivists: Rational people crunching data for science.  
|
|
Send message Joined: 1 Sep 04 Posts: 506 Credit: 118,619 RAC: 0 |
When there is such a small number remaining why not just send these work units out to a few more computers to speed things up a bit? Or does it not really matter because the next batch isn't ready to be sent out to us anyway? The problem with your suggestion is that it requires a manual intervention by CERN staff to do it. With a system as involved as BOINC, wading in to do the job by hand can cause more problems than it solves. This is why orphaned units tend to be left in the databases. There's plenty of other work for the CERN people to do, so a delay of a couple of days getting the results back isn't a problem. Gaspode the UnDressed http://www.littlevale.co.uk |
|
Ocean Archer Send message Joined: 13 Jul 05 Posts: 143 Credit: 263,300 RAC: 0 |
The other side to the coin is if there were "manual intervention" on behalf of certain Work Units, perhaps those WUs would be directed to "fat cat" crunchers whose results might skew the result. I remember a discussion thread a few months ago about how the various machines all deliver slightly different results due to the way the different programs are compiled. Such "manual intervention" could open the door to creating a result to fit the project, rather than one that was totally unbiased. I'd certainly hate to see one of the BOINC projects to which I donate time fail because of such actions ... If I've lived this long, I've gotta be that old |
|
Send message Joined: 26 Sep 05 Posts: 85 Credit: 421,130 RAC: 0
|
From the stantpoint of the project, a few days doesn't matter. As River and some others have pointed out, they aren't waiting for results... Case in point, no work is out there yet, and CERN is still not ready for the next batch of work. They have other things they also need to do (as they're doing right now), and so will get to the WUs when they're ready. Wasn't necessarily a matter of crunchers holding them up... As to the CPUs, it is true that Intel and AMD for instance have slightly different implementation, but this isn't based simply on the clock rate (actually performance, given the efficiency of 2 CPUs need not be the same). Both Intel and AMD have never gen processors, as well as older. And someone with a lattest gen dual proc P4 3.4+ GHz processor using the lattest core, and someone using an X2 or FX60! on the other hand, aren't exactly going to be hurting for crunch time... The slower comps would be more a PII vs P4, or a K62 vs. A64 type situation, then an Intel vs. AMD. Though there might be slight differences performance wise, the high end processors from each company aren't entirely non-competitive against each other, or anything of the sort... And true, perhaps the PII might get a bit of a different result then a P4 (different architecture), but as long as one can compare Intel to AMD, slight differences will show up (if this is what's wanted, unlike with Predictor which ties a WU to a given processor type to allow less variation). 
|
©2024 CERN