Posts by Andre Kerstens

1)

Message boards : Number crunching : Docking statestics....

( Message 4273 )
Posted 3281 days ago by Profile Andre Kerstens
Hans,

the statistics are only updated once a day and therefore it might have seen you only had 3500 credits. If you check now, you will see that you have 13652 credits.

Cheers
Andre

Hello!
Thanks for the steady flow of works and the applications that just works; at least for me!
During the night I have "lost" 10k according to boinc staistics And other I have gained about 13000 credits at this time of writing, but according to the stats on Docking I only have about 3500 credits. Was it reset yesterday or something that I can't find elsewhere in the message boards?

Keep up the nice work !
Thanks!

With regards,
Hans Sveen
Oslo, Norway

2)

Message boards : Cafe Docking : Personal Milestones

( Message 4183 )
Posted 3297 days ago by Profile Andre Kerstens
No personal milestones for 401 days... that's too much so here is one:
My left wing is almost done :-) More details on http://rv7a.tauker.net
Cheers!
Andre
3)

Message boards : Number crunching : Cobblestones

( Message 4133 )
Posted 3330 days ago by Profile Andre Kerstens
Hmmm, that sounds interesting. Yes, please let the project team know if you find something on that. In the meantime, Arun could try commenting out the boinc calls in the charmm code and see if the time calls are still being made. If not, then that points in the direction you are thinking.

Andre


Do you know for sure if the problem still exists? Unfortunately, I don't remember the details but while docking was shut down there was a fix mentioned on the BOINC developers mailing list (might have been the forums) that sounded to me like it might have been causing a similar problem. It's been a long time so I don't recall if it was in the BOINC client or in the application framework that was distributed. Since it didn't affect most applications, it must have been in a support function or something. A polling loop with no delay in it that was calling the OS time of day function to check elapsed time was what it sounded like. Might have had something to do with a heartbeat function. I'll see if I can find it.

4)

Message boards : Number crunching : Cobblestones

( Message 4116 )
Posted 3331 days ago by Profile Andre Kerstens
I've emailed you the notes I could find. Hope they will be a little bit useful.

Andre


The problem of execution time differing between windows and linux need to be solved before we move on to fixed credit based on FLOPS. I will be working on this issue tomorrow. Andre, can you pass me the notes you have on this issue ?

cheers,
Arun

5)

Message boards : Number crunching : Cobblestones

( Message 4105 )
Posted 3332 days ago by Profile Andre Kerstens
Good point. No, I've not been able to figure out why these system calls to get the time of day on linux are made a gazillion times per run. I do think that this might cause the massive difference in runtime between linux and windows. The runtime difference issue is already on the project's to-do list, so I'll make sure that whatever notes I have on this will be passed on to the next person trying to crack this issue.

Cheers
Andre

Since the WUs will have variable length, credits based on FLOPS would be my choice.


I'd agree with FLOPS but when assigning the credit per flop, I'd keep in mind other resources being used. IIRC, charmm takes a lot more memory than some other projects and was very disk or OS call intensive. I'm not sure Andre ever found out what was going on with the heavy disk/OS activity. On Linux, it showed up as large amounts of CPU time spent in "System" space. ISTR that many quad core machines (Often Macs) experienced vastly increased time per WU as more cores were working on docking until it basically paralyzed the machine. That was being worked on when the project shut down for the move. It may even have been a bug that was subsequently fixed in the BOINC client. There seemed to be a massive number of calls to request the time being made to the OS. I don't recall if that was ever linked to re-reading the script that runs charmm and possible updating of the last-access time for the script file or if it turned out to be something else entirely.

6)

Message boards : Unix/Linux : Slow Process Times for Linux / Compared to Windows

( Message 4104 )
Posted 3332 days ago by Profile Andre Kerstens
I just checked with Michela and she said that this is on the long list of things to fix. I guess at least this means it has not been forgotten :-)
Cheers,
Andre

Are we still using the same work unit types from UTEP ??
There was a big slow down in both Linux and Mac before the move that no one could pin point and the Windows application sped up.
It looks like that is still the case, you will get more Windows users than any other when this is realised, mainly due to the lower output and credit on the non Windows platforms.

My AMD Opteron Windows 285 takes about 2 hours to complete a work unit.
My AMD Opteron Linux 285 has so far taken 2 hours 45 minutes is at 53% and still has an estimated 3 hours 11 minutes to go.
My AMD opteron Linux 275 has taken 4 hours 21 minutes to be at 76% and still has 1 hour 40 minutes to go.

7)

Message boards : Number crunching : Checkpointing

( Message 4103 )
Posted 3332 days ago by Profile Andre Kerstens
Hi David,

I think I've seen this behavior before when the project was still at utep. The percentage done is provided to boinc by the application (in this case the value of FDONE from the percentdone.txt file that Michela mentions below), but boinc itself is supposed to update the cpu time value. It's possible that the new boinc version does this different than the previous version we used at utep. Arun, you might want to check that out.

Cheers
Andre

PS The current checkpoint seems to work fine, it's just not visible yet, but Arun will correct that tomorrow if I understand it correctly.

One of the WU mentioned above has finished. It matched a result from a p4 on XP and got credit.

http://docking.cis.udel.edu/result.php?resultid=9978

Maybe you can tell if it really continued from the checkpoint or started over. Either way, something is wrong because when I restarted the BOINC client and the result restarted, accumulated CPU time went to zero but the percentage complete continued from approximately where it had been before I shut down the BOINC client to test it.

<core_client_version>5.10.45</core_client_version>
<![CDATA[
<stderr_txt>
Calling BOINC init.
Starting charmm run...
Calling BOINC init.
Starting charmm run...
SUCCESS - Charmm exited with code 0.
Resolving file charmm.out...
Calling BOINC finish.

</stderr_txt>
]]>
8)

Message boards : Web site : News

( Message 3913 )
Posted 3400 days ago by Profile Andre Kerstens
That's mostly because there is not much to tell about :-( Of course now that the postdoc (Arun) has started work on D@H, this will soon change I am sure. I am helping Michela as much as I can, but have been swamped by work the last weeks; and whenever I am not working for SGI I am in the workshop working on my plane. There's hardly time anymore for anything else. But again, as soon as Arun is getting up to speed, he will make sure this project gets a lot more active again. It's great that so many of you have stuck around and are waiting for work. That's definitely appreciated by the whole team!

Cheers!
Andre


The fact that news can't even be updated on a monthly basis, is very disappointing.

9)

Message boards : Number crunching : Test WUs

( Message 3911 )
Posted 3404 days ago by Profile Andre Kerstens
Just answered this in another thread.

Andre
10)

Message boards : Number crunching : Docking Schedule

( Message 3910 )
Posted 3404 days ago by Profile Andre Kerstens
Yes, Michela's postdoc has arrived in Delaware. Arun has started working on the D@H project last week and is getting up to speed on everything BOINC right now. This project should start to feel a little bit more alive very soon now :-)

Andre

PS Michela is in Miami this week for the IPDPS conference. I think she is also meeting up with David Anderson and people from the WCG project somewhere this week.

G'Day to the Docking project,

Any news on your progress ???

Are we there yet ???



Next 10 posts