Additional WU

Questions and Answers : Preferences : Additional WU
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile marsinph
Volunteer tester

Send message
Joined: 7 Apr 01
Posts: 172
Credit: 23,823,824
RAC: 0
Belgium
Message 1819374 - Posted: 24 Sep 2016, 16:22:46 UTC
Last modified: 24 Sep 2016, 16:25:16 UTC

Hello,
Probably already asked in the past
I return each day about 90 WU. RAC about 8500 , one host/8cpu
My preference are set on "at least 10 days of work"
Also "additional work" set on 10 days.
If I can count, total 20 days.
But I only receive 200 WU. It will say about 3 days of work !
Who can explain ? I can not understand because some have a RAC about one hundred, but for one host receives 100's of WU !!! And of course those WU are never finished
Have a nice day
ID: 1819374 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 1819381 - Posted: 24 Sep 2016, 16:35:37 UTC - in response to Message 1819374.  
Last modified: 24 Sep 2016, 16:36:09 UTC

Seti has a maximum set of 100 tasks per CPU (doesn't matter how many cores) and 100 tasks per GPU. It could also be that it's 100 tasks for all GPUs combined, if one has more than one, I forget.

The maxima are set to keep the BOINC database at a manageable size, as even with the present max it's several terabyte. The project has in the past tried to give some leeway but this only resulted in a crashing database, which takes the whole project down.

Until they have better hardware/better database software/more people to work on that, there won't be any changes in this.
ID: 1819381 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22158
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1819391 - Posted: 24 Sep 2016, 18:20:20 UTC

If you want to maintain a full cache of 100 then (and I know this sounds contradictory) reduce the "additional days" to less than 1. This forces BOINC to check the number of tasks every time it contacts the servers and top up as required.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1819391 · Report as offensive
Profile marsinph
Volunteer tester

Send message
Joined: 7 Apr 01
Posts: 172
Credit: 23,823,824
RAC: 0
Belgium
Message 1819710 - Posted: 25 Sep 2016, 18:04:06 UTC - in response to Message 1819381.  

Tks Ageless for very clear explanation about 100 WU per CPU/GPU.
So in my case, it is logical. So I am not able to keep work for more than 3 days because I return about 70 or 80 WU each day !!! Of course depending the estimated time of calculation. Some WU about 3hr computation, some about one ...
But then how is it possible that a user with only one host have more than one thousand WU ??? And with a RAC of near zero !?
It is what I not understand.
ID: 1819710 · Report as offensive
Profile marsinph
Volunteer tester

Send message
Joined: 7 Apr 01
Posts: 172
Credit: 23,823,824
RAC: 0
Belgium
Message 1819711 - Posted: 25 Sep 2016, 18:07:30 UTC - in response to Message 1819391.  

Hello Bob, Tks for answer. I have change preferences ans put "additional work" to zero. After about six hours running, nothing change !
I repeat my host run 24/24 , 7/7 , 365/365, connection always on.
My profile is open, so you can check.
Kind regards
Philippe
ID: 1819711 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22158
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1819714 - Posted: 25 Sep 2016, 18:28:11 UTC
Last modified: 25 Sep 2016, 18:32:24 UTC

A quick look - you are sitting on the maximum number of tasks for your computer (100 for the CPU and 100 for the GPU) - which you weren't when you first asked. What will now happen is the cache will be more consistently full, rather than dropping down down a bit then filling you which is what you may have been seeing with 10+10.

Those with "engorged" caches have been indulging in some rather intensive cache management, often using some form of scripting and task shuffling to reach these high cache sizes. But it makes no odds given the normal weekly outage is about 6 hours - even my biggest rig just about manages to get through that without any such activity.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1819714 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 1819746 - Posted: 25 Sep 2016, 21:01:56 UTC - in response to Message 1819714.  

Those with "engorged" caches have been indulging in some rather intensive cache management, often using some form of scripting and task shuffling to reach these high cache sizes.

Some older 4.nn and 5.n.n clients primarily on Linux don't let limitations sit in their way either.
ID: 1819746 · Report as offensive
AMDave
Volunteer tester

Send message
Joined: 9 Mar 01
Posts: 234
Credit: 11,671,730
RAC: 0
United States
Message 1819760 - Posted: 25 Sep 2016, 22:01:55 UTC - in response to Message 1819374.  

For additional reference (and history).

ID: 1819760 · Report as offensive
Profile marsinph
Volunteer tester

Send message
Joined: 7 Apr 01
Posts: 172
Credit: 23,823,824
RAC: 0
Belgium
Message 1819932 - Posted: 26 Sep 2016, 16:43:17 UTC - in response to Message 1819374.  

Thanks a lot everyone about all your explanations and references.

So by limitations (100WU/GPU), I can receive about 2 1/2 work (turnaround about 38 min so 40x100WU /60/24 = 2.6days)

By CPU : about 3hr turnaround (x8 cores) = about 64 each day ! Total about 1 1/2 day work. Of course there are CPU task about 1hour turnaround. So totally, estimated WU for only one day.

But I understand the reasons about limitations. I hope there will be no any outage longer than 2 days ! My oldest WU from my 100CPU+100GPU is now 2 days old. Of course it are GPU WU

Have a nice day from Belgium !
ID: 1819932 · Report as offensive
Profile BilBg
Volunteer tester
Avatar

Send message
Joined: 27 May 07
Posts: 3720
Credit: 9,385,827
RAC: 0
Bulgaria
Message 1820395 - Posted: 29 Sep 2016, 6:04:49 UTC - in response to Message 1819932.  

You use the term "turnaround" wrong, what you call "turnaround" in your post is in fact "Run time".
if "I can receive about 2 1/2 work" - that is your turnaround - task received now will be finished and returned/reported to server after ~2.5 days.

You can see here "Average turnaround time 2.14 days":
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7072294

And here "Average turnaround time" for every app:
http://setiathome.berkeley.edu/host_app_versions.php?hostid=7072294
 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 
ID: 1820395 · Report as offensive

Questions and Answers : Preferences : Additional WU


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.