rpi_logo
80 WU limit
80 WU limit
log in

Advanced search

Message boards : Number crunching : 80 WU limit

Author Message
Peter Hucker
Send message
Joined: 5 Jul 11
Posts: 138
Credit: 27,075,696
RAC: 0

Message 67399 - Posted: 24 Apr 2018, 22:55:02 UTC

Milkyway only gives me 80 WUs at a time. This only take 1 hour 15 minutes to process on my graphics card. I have BOINC set to have a 1-2 day buffer (in case of project server downtime etc), so it keeps going to get work from my backup project. Why do we have this small limit?

melk
Send message
Joined: 10 Dec 17
Posts: 47
Credit: 567,990,234
RAC: 435,453

Message 67404 - Posted: 25 Apr 2018, 12:25:05 UTC

The max for that range is 10 days. There is also the "plus additional days" setting below it. Do you have enough free disk space?

Peter Hucker
Send message
Joined: 5 Jul 11
Posts: 138
Credit: 27,075,696
RAC: 0

Message 67405 - Posted: 25 Apr 2018, 12:28:58 UTC - in response to Message 67404.
Last modified: 25 Apr 2018, 12:34:12 UTC

I have it set to 1 day + 1 additional day. The limit is the 80 WUs imposed by the Milkyway server, it doesn't matter how much I specify. I have loads of disk space. I just checked your computers and you also have 80 tasks queued per graphics card. P.S. wow you have a lot of graphics cards!

melk
Send message
Joined: 10 Dec 17
Posts: 47
Credit: 567,990,234
RAC: 435,453

Message 67406 - Posted: 25 Apr 2018, 15:09:11 UTC - in response to Message 67405.

Maybe it is a hard limit so people do not hog all the WU :) I don't know.

Magiceye04
Send message
Joined: 14 Feb 10
Posts: 2
Credit: 5,201,835
RAC: 400

Message 67585 - Posted: 9 Jun 2018, 18:29:10 UTC

I would be happy about 80 WUs.
Also about 4 WUs.
But i only get 1. This means, the GPU is more or less in idle state.
How do i get more than 1 ?

Profile mikey
Avatar
Send message
Joined: 8 May 09
Posts: 2201
Credit: 250,014,349
RAC: 92,239

Message 67588 - Posted: 10 Jun 2018, 11:55:02 UTC - in response to Message 67406.

Maybe it is a hard limit so people do not hog all the WU :) I don't know.


Yes that's true in both cases, 80 per person until you start returning them then you get one as you return one AND it's so one pc doesn't hog them all then crash and the rest of us are stuck with nothing until they expire. My AMD 5870 does a workunit in about 100 seconds each, so a 10 day cache would be around four THOUSAND work units. They do have the Server setup to make new workunits as needed though and it does seem to be doing that fairly well.

Profile mikey
Avatar
Send message
Joined: 8 May 09
Posts: 2201
Credit: 250,014,349
RAC: 92,239

Message 67589 - Posted: 10 Jun 2018, 12:05:58 UTC - in response to Message 67585.

I would be happy about 80 WUs.
Also about 4 WUs.
But i only get 1. This means, the GPU is more or less in idle state.
How do i get more than 1 ?


Check the venue..default, work, home etc and see if it's set at a zero resource share. Also if you are sharing your gpu between several projects that's another reason as Boinc is trying hard to respect your resource share between projects. ie if Seti and MW are both set to 50% then Boinc will try hard to get work from both projects but since Seti gives out less credits per workunit it will get more work from Seti and less from MW. The percentage between projects is a per day thing and is based on credits not number of workunits.

That's one huge reason why alot of us ended up with multiple machines, so we could do work for several projects at once and not be stymied by Boincs quirks.

Jim1348
Send message
Joined: 9 Jul 17
Posts: 26
Credit: 1,758,060
RAC: 25,604

Message 67590 - Posted: 10 Jun 2018, 18:19:40 UTC - in response to Message 67589.

The percentage between projects is a per day thing and is based on credits not number of workunits.

That's one huge reason why alot of us ended up with multiple machines, so we could do work for several projects at once and not be stymied by Boincs quirks.

If you say so. That is so idiosyncratic that only BOINC would think of it. Thanks.
(I do have multiple machines too.)

Henk Haneveld
Send message
Joined: 15 Aug 14
Posts: 5
Credit: 1,404,019
RAC: 3

Message 67591 - Posted: 11 Jun 2018, 8:24:44 UTC - in response to Message 67589.

The percentage between projects is a per day thing and is based on credits not number of workunits.


No. The internal value "rec" in the client_state.xml file is used and this is not based on credits.

If you give 2 projects with different credits the same resource share and let them run for a couple of weeks to stabilize you will see that they will each get half the cpu time.
____________


Post to thread

Message boards : Number crunching : 80 WU limit


Main page · Your account · Message boards


Copyright © 2018 AstroInformatics Group