ovgrm48???

Message boards : Number crunching : ovgrm48???

To post messages, you must log in.

AuthorMessage
Profile mdoerner
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 30 Jul 08
Posts: 202
Credit: 6,998,388
RAC: 0
Message 1319 - Posted: 9 Nov 2009, 3:18:50 UTC

Is this a new message or another snippet of the old one we've had problems cracking?

Mike D
ID: 1319 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,297,604
RAC: 381,257
Message 1320 - Posted: 9 Nov 2009, 8:00:19 UTC

M4 project doesn't have work available (actually it does, but it keeps sending the same last workunits until it receives everything back), so while waiting for Stefan to switch the server to another text, I split the full message to serveral short parts and set the BOINC server to run them in parallel, each with only 3 walks through the keyspace (that's around 35k workunits).

Btw, the first batch was bugged, there were around 3-4k workunits with download errors and another 3-4k with validate errors (for all these which used any CPU time credits were granted, because the problem was caused by my mistake).

M4 Project homepage
M4 Project wiki
ID: 1320 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Orion7

Send message
Joined: 2 Oct 09
Posts: 2
Credit: 228,508
RAC: 0
Message 1321 - Posted: 9 Nov 2009, 12:02:21 UTC

Okay, this "ovgrms" need a much longer deadline!

Or is it just me, seeing ruined schedules for the work resent of enigma?

Greetings

Orion7
seti.germany
ID: 1321 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,297,604
RAC: 381,257
Message 1322 - Posted: 9 Nov 2009, 12:08:39 UTC

The deadline is the same as for other workunits, could you explain what's wrong ?

M4 Project homepage
M4 Project wiki
ID: 1322 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Orion7

Send message
Joined: 2 Oct 09
Posts: 2
Credit: 228,508
RAC: 0
Message 1323 - Posted: 9 Nov 2009, 12:18:27 UTC

Today I got lots of Wus with deadlines between 11. or 14. November 2009 to 21. November 2009.

http://www.enigmaathome.net/results.php?hostid=20026&offset=20&show_names=0&state=0

But sorry, thought this was only at ovgrms. After better investigation, I see there are several normal enigma-wus between.

Hmm strange thing too ...

Greetings

Orion7
seti.germany
ID: 1323 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,297,604
RAC: 381,257
Message 1324 - Posted: 9 Nov 2009, 12:23:42 UTC

I think it's just the way how BOINC works: you have a fast PC with a short turnaround time (1.1 day is a lot lower than average), so the server sends you a lot of resends, some of them with shorter deadlines.

M4 Project homepage
M4 Project wiki
ID: 1324 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [AF>France>Aquitaine>Cote-Adour-et-Gaves]Bernard 64250

Send message
Joined: 24 Sep 09
Posts: 18
Credit: 4,428,367
RAC: 0
Message 1334 - Posted: 11 Nov 2009, 18:42:44 UTC

I think to have the same problem than Orion 7 from Germany.
One of my PCs just started an estimated 27h workunit received november 10th at 15:50 and to be returned before november 13th at 09:37. That may be a little difficult to reach the deadline.

Bernard du 40 from L'Alliance Francophone
ID: 1334 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,297,604
RAC: 381,257
Message 1340 - Posted: 12 Nov 2009, 19:34:14 UTC

Feel free to abort any workunit you don't like, but I think that it is just a problem with estimated time.
I've seen the same problem on my machine, longest workunits have their estimated time at around 35-40 hours while they take 8-9h maximum.
I have no idea how it could happen, I guess I made a mistake when I calculated the fpops_est. This is quite strange however, because I followed the same exact procedure I used for all the previous workunits, that always worked (benchmarked the tasks on my own PC and compared the time to a task with known fpops_est). Anyway, I'll check everything again and eventually I'll fix the templates, it has to wait a few hours though, because I'm in the middle of upgrading/fixing the workunit generator.

M4 Project homepage
M4 Project wiki
ID: 1340 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [AF>France>Aquitaine>Cote-Adour-et-Gaves]Bernard 64250

Send message
Joined: 24 Sep 09
Posts: 18
Credit: 4,428,367
RAC: 0
Message 1343 - Posted: 13 Nov 2009, 2:11:17 UTC - in response to Message 1340.  
Last modified: 13 Nov 2009, 2:32:04 UTC

My concern wasn't the duration itself, but the deadline : initially it was 66h for a 27h WU, corresponding to 10h of CPU time per day.
By lack of chance, I already had other Enigma WUs in queue, so it started 24 hours later and the required CPU time had grown to a huge 15.5h/day.
In order to reach it, I didn't switch OFF my PC during the night and it ended fine.

New Workunits seem to have turned back to normal, but I still have a 70h WU running with a 14-day deadline (i.e. only 5h of CPU time per day, that looks a little bit more reasonable). It's already at 50%, and it will probably take 60 hours to complete, not so far from the estimated duration.
ID: 1343 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 69,297,604
RAC: 381,257
Message 1346 - Posted: 13 Nov 2009, 12:48:57 UTC - in response to Message 1343.  

Super short deadline was a bug and it's fixed now (however, there's a small number of affected workunits still available in the server queue).
Wrong time estimation is another thing, for now I have no idea why it happens. I have to take a closer look at the runtime averages and fpops count.


M4 Project homepage
M4 Project wiki
ID: 1346 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : ovgrm48???




Copyright © 2017 TJM