Message boards :
Number crunching :
Did everyone get work 02 Nov UTC?
Message board moderation
Previous · 1 · 2 · 3
Author | Message |
---|---|
Send message Joined: 13 Jul 05 Posts: 456 Credit: 75,142 RAC: 0 |
my 'backoff' is < 4 hours , lhc is set alike pirates and simap, and i get wu on all boxes ,when they 'burst' Once there has been no work for a few days or a week, the client falls into a pattern of random backoffs where the *max* is 4hrs and the min is 1min. Because the interval is selected at random each time it should be around 2hrs on average, but you should see individual stand offs that are both significantly larger than 2hrs and significantly less than that. Rather imprecisely, we tend to talk of a "4hr backoff" to describe this. To make sure everyone in long term backoff gets work, it does need to be available continuously on the server for four hours, otherwise in a large community like ours, someone is going to suffer the double bad luck of having a nearly four hour random wait that also steps over the time when work was there. That is why it makes a kind of imprecise sense to call it a 4hr backoff. R~~ |
Send message Joined: 18 Sep 04 Posts: 38 Credit: 173,867 RAC: 0 |
Doh. i was thinkin' 'connect every' .I do many projects so I keep my caches real small cuz i don't like it when i have pages of wu to scroll through to see what's running (or if) |
Send message Joined: 13 Jul 05 Posts: 456 Credit: 75,142 RAC: 0 |
Doh. i was thinkin' 'connect every' .I do many projects so I keep my caches real small... Hi Morgan, connect every < 4hrs is cool. The default is 2.4hrs of course. cuz i don't like it when i have pages of wu to scroll through to see what's running (or if) I see you have a whole bunch of computers (sometimes called a farm). Do you use BoincView to keep track? I recommend it. BV allows you to see what is going on across several boxes at once. I use BV to monitor boxes on my LAN and outside my own LAN - these are both easy to do. BV only runs on win, but can view/control clients on win lin and mac in any mixture. By clever sorting of the BV display I group all the running tasks together near the top of the display, and can easily see what is running on each box, and easliy spot if a box is not running anything. No mega scrolling even when Leiden sends me a few hours worth of 7min tasks! In addition the BV icon in the systray is coloured green when all is ok on all boxes, and various other colours if BV thinks there might be a problem. I get an instant overview of all my boxes every time I look at the systray. Good for picking up problems early. R~~ |
Send message Joined: 22 Sep 05 Posts: 21 Credit: 6,350,753 RAC: 1 |
I agree with the BoincView. I run 8 projects over 4 work systems (3 colleague machines) and my home system. We found out recently that it's a good way of confirming that a NIC on a colleagues machine was intermittently dropping connections. Once the on-board NIC was disabled and one put in, then BV showed the machine as solid blue. This isn't meant to start any thread battles and is only my opinion. Since I run multiply projects on each system, when 1 project doesn't have any work, then to me that means other projects get that much more time so I've stuck with the stock caching. Now that I'm on the soapbox, what's the deal with joining all the projects? (Again, I really don't want to know,) it seems to me instead of concentrating on some core ones (3 (lhc, predictor, and simap) of my 8 have recently usually been out of work), the people who do a couple of work units and then move on are only taxing the project / statics systems. William |
Send message Joined: 15 Feb 06 Posts: 67 Credit: 460,896 RAC: 0 |
Got NO LHC work at all. Ah, but you didn't exacerbate global warming by burning unnecessary ressources either. Help yourself to a feeling of smugness and superiority anyway. I got a couple of WUs, as I happened to switch a computer on at around 08:00 UTC and must have got some of the tail end. sQuonk Plague of Mice Intel Core i3-9100 CPU@3.60 GHz, but it's doing its bit just the same. |
Send message Joined: 22 Sep 05 Posts: 21 Credit: 6,350,753 RAC: 1 |
After thinking about your comment, I guess you have a valid point about the smugness. I guess I lashed out because of all the "no work" related comments. Probably something similar to all of the drivers we have in around the Seattle area that have no problems parallel parking doing 70mph on the freeways. As far as the global warming, I'm just trying to use the electricity generated by the hydro power instead of the natural gas to run the furnance at home (but how much heat can I generate with a P4 2.8 anyways). William Got NO LHC work at all. |
Send message Joined: 14 Jul 05 Posts: 275 Credit: 49,291 RAC: 0 |
http://www.climateprediction.net/info/part_faq.php#q3.1 |
Send message Joined: 22 Sep 05 Posts: 21 Credit: 6,350,753 RAC: 1 |
Great!
|
Send message Joined: 13 Jul 05 Posts: 456 Credit: 75,142 RAC: 0 |
As far as the global warming, I'm just trying to use the electricity generated by the hydro power instead of the natural gas to run the furnance at home (but how much heat can I generate with a P4 2.8 anyways). You need more computers then :) Personally I can't afford a dozen P4 2.8 machines, so I got a dozen or so old machines and have them all running in the winter. They keep my lounge nive and warm in all but the coldest weather. R~~ |
Send message Joined: 18 Sep 04 Posts: 38 Credit: 173,867 RAC: 0 |
:) pokeing so many holes in so many machines as i see go through here, and remembering what/where is too much for me, last week i used a combo of putty from winxp and 'places' 'connect to server' from a linux machine to do some BOINC'n with some opterons :) .i bam for the remote and keep a variety on all the part time machines William Timbrook wrote:
of 35 projects 8 have less work than lhc :lol: 3 have deadlines of today, 4 require more than 1/2G ram so 5 machines cant do them. that leaves about 7 'core' and 7 not the not are alpha projects which need a variety of machines os and setup which are the ones i'm doing now and the <lhc any chance i get :D i am aiming for 80+ % at all projects atm. P.S. Thanks River for the bv tip anywayz tho and 1 p.c. had 3 wu from chess on the 2nd ( forgot that last week lol too) i see we need a new system for xml stats eh, seti stats are getting to big :lol: |
Send message Joined: 22 Sep 05 Posts: 21 Credit: 6,350,753 RAC: 1 |
You stated that 4 of the projects require more than 512meg of ram. I know world comm grid / cancer requires 720. What are the other 3 that you found? William |
Send message Joined: 14 Jul 05 Posts: 275 Credit: 49,291 RAC: 0 |
I just got LHC work!! And I have no reason to hide it. Everybody go and get some! *SOME*, not five hundred. I really think a cache limiter on LHC would help a lot. RCN added a quick n dirty hack to the scheduler to limit cache and they really get things done faster. And also helps make sure there are a lot of people getting some work, instead of some people getting a lot of work, while the rest gets nothing. |
Send message Joined: 18 Sep 04 Posts: 38 Credit: 173,867 RAC: 0 |
@ William :embarass: meant 1/4 cpdn sap is more than 1/2 + wcg cancer over 1/4 are rosetta ralph hashclash Project specs and Hardware requirements (by Honza) |
©2024 CERN