Welcome to MilkyWay@home

convolution searches

Message boards : Number crunching : convolution searches
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 1070 - Posted: 14 Dec 2007, 21:04:33 UTC

I'm starting up some convolution searches, let me know how they run.

gs_50 ... gs_54 are running convolution
gs_40 ... gs_44 are running non-convolved

both are running the hybrid simplex/genetic search optimization method.

the convolution workunits should take a lot longer (mabye 30+ minutes), but should grant the same amount of credit per time as unconvolved work units. let me know if anything is off.
ID: 1070 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
pieface

Send message
Joined: 7 Nov 07
Posts: 13
Credit: 20,143,762
RAC: 0
Message 1074 - Posted: 14 Dec 2007, 23:08:55 UTC

Hmmmm.... if a 400 sec wu gets 2 'credits, then I don't think the proportion is correct when an 1200 sec wu gets 4.86?

400 sec and
1200 sec

but hey.... there's snow out on the hill's...
ID: 1074 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 1076 - Posted: 14 Dec 2007, 23:33:06 UTC - in response to Message 1074.  

Hmmmm.... if a 400 sec wu gets 2 'credits, then I don't think the proportion is correct when an 1200 sec wu gets 4.86?

400 sec and
1200 sec

but hey.... there's snow out on the hill's...


yeah i noticed this myself, i'm checking out whats going wrong. after the already generated workunits are out of the system the new convolution workunits should grant the right amount of credit.
ID: 1076 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 1078 - Posted: 14 Dec 2007, 23:47:46 UTC - in response to Message 1076.  

newly generated work units for the convolution should be granting a correct amount of credit. let me know if anything else is fishy. (they should be granting 6.25 credits)
ID: 1078 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jayargh
Avatar

Send message
Joined: 8 Oct 07
Posts: 289
Credit: 3,690,838
RAC: 0
Message 1079 - Posted: 14 Dec 2007, 23:59:11 UTC

Just had a weird one I aborted. 1194348 It was showing .118 progress going up in .001 increments after 18 minutes.Would have taken more than a day to complete at that rate.Was not the thing were after 100% it counts to 5 and uploads.

Thanks Travis also for the credit adjustment per new wu and the stats update :)
ID: 1079 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Irishgeezah
Avatar

Send message
Joined: 10 Nov 07
Posts: 37
Credit: 11,855,733
RAC: 0
Message 1081 - Posted: 15 Dec 2007, 0:28:38 UTC
Last modified: 15 Dec 2007, 0:31:04 UTC

The credit awarded for this workunit seems to be out by a little bit : Result

Probably one of the older ones with errant credit but even awarding 6.xx will still leave it about 25% lower than ps workunits are awarded for the same time crunching.


ID: 1081 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
cwhyl

Send message
Joined: 11 Nov 07
Posts: 41
Credit: 1,000,181
RAC: 0
Message 1082 - Posted: 15 Dec 2007, 0:43:19 UTC

A little funny behavior: when progress goes to 99% it jumps back to 0.001% but then quickly goes up again and finishes fine, no add on cpu time.
ID: 1082 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [B^S] Acmefrog
Avatar

Send message
Joined: 28 Aug 07
Posts: 49
Credit: 556,559
RAC: 0
Message 1084 - Posted: 15 Dec 2007, 3:06:55 UTC - in response to Message 1079.  
Last modified: 15 Dec 2007, 3:07:20 UTC

Just had a weird one I aborted. 1194348 It was showing .118 progress going up in .001 increments after 18 minutes.Would have taken more than a day to complete at that rate.


Mine go to 100% complete then back to 0% and starts back up at .001 increments as well. But mine will count to about .140-.150 then complete and upload. All have been successful---at least given credit for ;)
ID: 1084 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jayargh
Avatar

Send message
Joined: 8 Oct 07
Posts: 289
Credit: 3,690,838
RAC: 0
Message 1088 - Posted: 15 Dec 2007, 15:18:14 UTC

Got a computation error on this wu 1208218 . Process got signal 11.

The wu I aborted yesterday must have been normal and I caught it after it completed 100% but before uploading as what I saw looks like the norm on these convoloution results.
ID: 1088 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jord
Avatar

Send message
Joined: 30 Aug 07
Posts: 125
Credit: 207,206
RAC: 0
Message 1089 - Posted: 15 Dec 2007, 15:27:07 UTC - in response to Message 1078.  
Last modified: 15 Dec 2007, 15:29:40 UTC

newly generated work units for the convolution should be granting a correct amount of credit. let me know if anything else is fishy. (they should be granting 6.25 credits)

Something's fishy. ;-)

this little piggy went to the market and after haggling for 1275 seconds he got 4.38 after bartering for 3.42
this little piggy went to the market and after haggling for 1290 seconds, was ripped off with 6.50 when bartering for 3.46

Will check what the next one says in 5 minutes.
Jord.

The BOINC FAQ Service.
ID: 1089 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile [AF>Slappyto] popolito

Send message
Joined: 19 Nov 07
Posts: 1
Credit: 37,184
RAC: 0
Message 1090 - Posted: 15 Dec 2007, 15:28:53 UTC

You're lucky !
http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=1153167
:(
ID: 1090 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jord
Avatar

Send message
Joined: 30 Aug 07
Posts: 125
Credit: 207,206
RAC: 0
Message 1091 - Posted: 15 Dec 2007, 15:31:33 UTC - in response to Message 1090.  

You're lucky !
http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=1153167
:(

You're still getting 6.50 ... when apparently the high court had decided it was 6.25 .. perhaps that inflation hit here as well? Me goes checking what the dollar's worth. ;-)
Jord.

The BOINC FAQ Service.
ID: 1091 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Jayargh
Avatar

Send message
Joined: 8 Oct 07
Posts: 289
Credit: 3,690,838
RAC: 0
Message 1092 - Posted: 15 Dec 2007, 15:32:58 UTC - in response to Message 1089.  
Last modified: 15 Dec 2007, 15:35:07 UTC

newly generated work units for the convolution should be granting a correct amount of credit. let me know if anything else is fishy. (they should be granting 6.25 credits)

Something's fishy. ;-)

this little piggy went to the market and after haggling for 1275 seconds he got 4.38 after bartering for 3.42
this little piggy went to the market and after haggling for 1290 seconds, was ripped off with 6.50 when bartering for 3.46

Will check what the next one says in 5 minutes.



Jord-Travis changed the credit granted mid-stream from 4.38 to 6 (front page news)It is set so the credit is self adjusted at the server based on star volume.
ID: 1092 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jord
Avatar

Send message
Joined: 30 Aug 07
Posts: 125
Credit: 207,206
RAC: 0
Message 1093 - Posted: 15 Dec 2007, 15:34:42 UTC - in response to Message 1092.  

Good good. I was just reacting to his claim of 6.25 <> 6.50

Maybe that the server inhaled snowflakes.
Jord.

The BOINC FAQ Service.
ID: 1093 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile LiborA
Avatar

Send message
Joined: 15 Sep 07
Posts: 15
Credit: 9,818,265
RAC: 0
Message 1094 - Posted: 15 Dec 2007, 19:09:39 UTC

For this my comp is still credit system 6,5 cr per WU about 10% worse as for gs_3x WU.
For gs_3x I got about 20 cr/hour
For gs_5x I got about 18 cr/hour
ID: 1094 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 1095 - Posted: 15 Dec 2007, 20:03:10 UTC - in response to Message 1094.  

For this my comp is still credit system 6,5 cr per WU about 10% worse as for gs_3x WU.
For gs_3x I got about 20 cr/hour
For gs_5x I got about 18 cr/hour


i started a few with 6.25, but this didn't seem high enough, so they shoul be awarding 6.5 now.

i'll look into the errors in the code, it might be something nate put in. as to the progress going back down to 0 then back up really quickly it seems that for whatever reason the first time the progress goes to 100 that's for the integration over the volume, and the second time is when this integral values are used to calculate the fitness over all the stars. that should be fixed in the next version and it's nothing to worry about.
ID: 1095 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
=Lupus=

Send message
Joined: 27 Aug 07
Posts: 5
Credit: 4,252,441
RAC: 477
Message 1154 - Posted: 21 Dec 2007, 22:32:05 UTC - in response to Message 1094.  

For this my comp is still credit system 6,5 cr per WU about 10% worse as for gs_3x WU.
For gs_3x I got about 20 cr/hour
For gs_5x I got about 18 cr/hour


WTF? You only crunch 3 gs5's an hour with a dualcore Athlon64 4000+?
My Athlon64 3500+ single-core does the same. And it is not OC'd.
Or is it just 18cr per hour per core?
ID: 1154 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile LiborA
Avatar

Send message
Joined: 15 Sep 07
Posts: 15
Credit: 9,818,265
RAC: 0
Message 1155 - Posted: 22 Dec 2007, 6:34:27 UTC - in response to Message 1154.  

WTF? You only crunch 3 gs5's an hour with a dualcore Athlon64 4000+?
My Athlon64 3500+ single-core does the same. And it is not OC'd.
Or is it just 18cr per hour per core?


Yes this is per core
ID: 1155 · Rating: 1 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : convolution searches

©2024 Astroinformatics Group