rpi_logo
New Separation Runs
New Separation Runs
log in

Advanced search

Message boards : News : New Separation Runs

Author Message
Profile Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 501
Credit: 34,647,251
RAC: 338

Message 67714 - Posted: 21 Aug 2018, 14:14:28 UTC

Hey Everyone,

I put up a few new separation runs. You may notice that these have slightly different run times. The credit awarded should reflect the run time.

The run names are:

de_modfit_sim19fixed_bundle4_4s_NoContraintsWithDisk260_1
de_modfit_sim19fixed_bundle4_4s_NoContraintsWithDisk260_2
de_modfit_sim19fixed_bundle4_4s_NoContraintsWithDisk260_3
de_modfit_sim19fixed_bundle6_2s_NoContraintsWithDisk140_1
de_modfit_sim19fixed_bundle6_2s_NoContraintsWithDisk140_2
de_modfit_sim19fixed_bundle6_2s_NoContraintsWithDisk140_3

If you have any questions or issues, let me know.

Jake

aad
Send message
Joined: 30 Mar 09
Posts: 59
Credit: 364,973,149
RAC: 421,578

Message 67729 - Posted: 27 Aug 2018, 11:17:41 UTC - in response to Message 67714.
Last modified: 27 Aug 2018, 11:25:27 UTC

The credit awarded should reflect the run time.

If you have any questions or issues, let me know.

Jake

de_modfit_sim19fixed_bundle6_2s_NoContraintsWithDisk140

take a few seconds longer, but get 203.92 credit instead of 227.62 credits....
This seems wrong to me.
Anyone else see this behavior...?

aad
Send message
Joined: 30 Mar 09
Posts: 59
Credit: 364,973,149
RAC: 421,578

Message 67731 - Posted: 27 Aug 2018, 21:31:44 UTC

As I investigated further;
Seems that only with ATI-cards the 'de_modfit_sim19fixed_bundle6_2s_NoContraintsWithDisk140' take longer to process and get less credit for it...

Profile Jake Weiss
Volunteer moderator
Project developer
Project tester
Project scientist
Send message
Joined: 25 Feb 13
Posts: 501
Credit: 34,647,251
RAC: 338

Message 67739 - Posted: 29 Aug 2018, 16:17:35 UTC

Hey aad,

That's strange, sorry about that. I will double check the credit calculations before the next set of runs go up.

Thanks,

Jake

aad
Send message
Joined: 30 Mar 09
Posts: 59
Credit: 364,973,149
RAC: 421,578

Message 67742 - Posted: 29 Aug 2018, 17:11:38 UTC
Last modified: 29 Aug 2018, 17:19:05 UTC

You can check it with this machine;
http://milkyway.cs.rpi.edu/milkyway/show_host_detail.php?hostid=399761
Thanks for looking in to this.

I think you don't have to bother with calculating the credit.
There is (with this GPU)only 5 or 6 seconds between the 6-2 and 4-4 versions.

Aad

bluestang
Send message
Joined: 13 Oct 16
Posts: 41
Credit: 140,265,265
RAC: 372,156

Message 67812 - Posted: 12 Sep 2018, 18:25:00 UTC

Well, if you run 3 or more tasks at the same time there is enough of a difference that the 6_2 bundles can take 20-40 seconds longer and affect daily points.

If the 6_2 units take longer then they should get more credit than the 4_4 units no?

Or just stick with what was working well in the 4_4 bundles :)


Post to thread

Message boards : News : New Separation Runs


Main page · Your account · Message boards


Copyright © 2018 AstroInformatics Group