Drug sim nanoscale




Advanced search

Message boards : Number crunching : Drug sim nanoscale

Author Message
Werinbert
Send message
Joined: 25 May 13
Posts: 10
Credit: 10,065
RAC: 0
Message 266 - Posted: 5 Jun 2013 | 18:03:28 UTC

Currently this is application set for quorum of 1 and replication of 2. This setting is wasteful as there are two runs being processed while only one run is needed/used.

You should either have settings of Q=1 and R=1 to reduce extra, unneeded runs; or Q-2 and R=2 if you want to do validation checks between two separate runs, and check against cheating.

Thanks

Profile fsanz
Project administrator
Project developer
Volunteer developer
Project scientist
Send message
Joined: 23 Sep 09
Posts: 84
Credit: 1,071,256
RAC: 0
Message 270 - Posted: 6 Jun 2013 | 15:30:44 UTC - in response to Message 266.
Last modified: 6 Jun 2013 | 15:33:03 UTC

You're right,we have right now Q=1 and R=2, let me explain that.

If you have a look at http://registro.ibercivis.es/jobs.php, and you select biosoft, you can see the status of the last job (groups of workunits) 641494, it's at ~90%.

One or two weeks ago, we have for this application Q=1 and R=1, and the remaining 10% of the jobs was sended to the clients, and as you know, we have about 5 months of delay_bound for this app.

So, a lot of these last results, probably, are sended to some hosts that aren't going to reconect to our server. With other words, there are clients that are not going to reconect to the Ibercivis server, and these clients have pending results to compute and send. So the server needs to wait all the delay_bound to create a new result. This is the tail problem, that is common to all BOINC projects.

To solve that, some projects are re-sending these last results to the cloud, we're trying to solve this issue duplicating the numbers of results for each workunit (R=2). We can see the evolution of the job in http://registro.ibercivis.es/jobs.php, biosoft
Regards

Francisco

Werinbert
Send message
Joined: 25 May 13
Posts: 10
Credit: 10,065
RAC: 0
Message 274 - Posted: 8 Jun 2013 | 7:13:09 UTC

Thank you for the explanation.

By the way, the jobs.php page shows only the drop-down menus and nothing else, so I am unable to see any of the job information.

Profile fsanz
Project administrator
Project developer
Volunteer developer
Project scientist
Send message
Joined: 23 Sep 09
Posts: 84
Credit: 1,071,256
RAC: 0
Message 278 - Posted: 8 Jun 2013 | 15:08:28 UTC - in response to Message 274.

It's working for me under linux+chrome.

What os, navigation are you using?

Thanks for the feedback

Werinbert
Send message
Joined: 25 May 13
Posts: 10
Credit: 10,065
RAC: 0
Message 282 - Posted: 8 Jun 2013 | 16:18:18 UTC

Windows 7
Internet Explorer 9

Profile fsanz
Project administrator
Project developer
Volunteer developer
Project scientist
Send message
Joined: 23 Sep 09
Posts: 84
Credit: 1,071,256
RAC: 0
Message 286 - Posted: 13 Jun 2013 | 14:07:48 UTC - in response to Message 282.

Hello

Can you test it now?

Regards

Werinbert
Send message
Joined: 25 May 13
Posts: 10
Credit: 10,065
RAC: 0
Message 290 - Posted: 13 Jun 2013 | 23:08:06 UTC - in response to Message 286.

It is working now, thanks.

Profile Cesium_133*
Avatar
Send message
Joined: 26 Jan 09
Posts: 5
Credit: 185,930
RAC: 0
Message 302 - Posted: 25 Jun 2013 | 5:23:53 UTC

Any reason why the whole Ibercivis suite has no work? I was running Drug Sim Nanoscale and got the "no work" message... I like the points you folks give out, and I like drugs... the beneficial kind, i.e. :) Can someone give me some 4-1-1 on this? Thx, Cesium...
____________
Think of the DC-10 crash at O'Hare in Chicago in 1979. Worst US air disaster ever involving 1 plane. If we don't get a grip on disease... we're gonna end up like those 273 people who perished-


Post to thread

Message boards : Number crunching : Drug sim nanoscale