Questions and Answers :
Windows :
None of my WU\'s seems to finish properly on one of my hosts
Message board moderation
Author | Message |
---|---|
Send message Joined: 29 Aug 04 Posts: 11 Credit: 1,281,270 RAC: 0 |
Today I just aborted calculation on tasks for the zillionth time. It took over a week to complete the first % of calculation. No trickles are send. When I open the Grafical display I see it takes ages to calculate a timestep. This happens every time I receive a new task. What is happening here? I haven\'t finish a trickle for about 2 month now ... Simmel |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
one possibility is that you have this project set at too low a priority, thus not allowing a model to run for long enough to reach a savepoint. But, as it seems that none of your other projects are running at present, it may be that the computer itself isn\'t running for long enough to allow a model to reach a savepoint. In either case, the model will restart from the beginning each time that you turn on the computer. |
Send message Joined: 9 Jan 07 Posts: 467 Credit: 14,549,176 RAC: 317 |
I think Les is right. You have a number of machines, but I assume your post relates to 485517, which looks like a laptop. That machine has picked up a series of HADAM3 models. These models have the longest interval between savepoints of the three model types currently on offer at CPDN (45 minutes on my P4). If the model is stopped during that period it will simply revert to the last savepoint and never make any progress. If that laptop is used for short periods, then you could de-select HADAM3 in your project preferences, so that a model type with a shorter savepoint interval will then be downloaded. There is another laptop in your computer list, which has a lot of crashes and user aborts (i.e. 869782). Some of those aborts appear to be ice-world slabs. The crashes may be due to the laptop hibernating: it is a good idea to stop the model manually on any machine, if possible - and models should never be hibernated. Best of luck. |
Send message Joined: 29 Aug 04 Posts: 11 Credit: 1,281,270 RAC: 0 |
Well guys, true about running on laptop, but I anticipated on that to run my laptop overnight (for more than 10 hrs continue). Same result though. But what I also see is that calculating a simgle timestep is taking forever. I checked that with the graphical display on to see the progress (well, actually the lack of that) Simmel |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
Sometimes a computer shows up that\'s just not capable of running climate models. And it looks like your laptop is one of these. Best not to push things; laptops don\'t have the \"stamina\" required for these long models at the best of times. And you\'re just wasting data sets trying. |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
I have had one thought on a possibility: Your computer may be spending a lot of time running an indexing program. And BOINC (and the science app) make a LOT of file writes. |
Send message Joined: 29 Aug 04 Posts: 11 Credit: 1,281,270 RAC: 0 |
I checked the indexing service, which was disabled by me previously. I also excluded the 4-us scanner from the BOINC install dir a couple of weeks ago. So, sad to say, I might have to quit CPDN for this host ... Simmel |
Send message Joined: 6 Oct 06 Posts: 204 Credit: 7,608,986 RAC: 0 |
The same thing happened to me once. Instead of 4 s/Ts it jumped to 4000 s/Ts but that was due to faulty RAM. I changed it, the time came back to normal. The RAM for some reason was misbehaving only with climate? Just a thought. Regards Masud. |
©2024 climateprediction.net