climateprediction.net home page
Posts by bl

Posts by bl

1) Message boards : Number crunching : Computation error Weather At Home 2 (wah2) 8.24 on WIN7 64-bit (Message 56389)
Posted 15 Jun 2017 by bl
Post:
I've had them too (using linux here), and my suspicion is that when a computation is interrupted (either by BOINC manager cycling tasks to another project, or by system shutdown) the chance for a computation error goes from near zero to very likely.

I wish my boinc manager had a way to mark tasks priority. That way I could update and get more tasks on another project without the likelylyhood that it will suspend and error out one (or mulitple) of my weather model tasks.
2) Message boards : Cafe CPDN : Cloudy planet (Message 52474)
Posted 25 Aug 2015 by bl
Post:
Two-thirds of Earth is covered in clouds

A new global cloud map based on nearly a decade of NASA satellite data have found that about 67 per cent of Earth's surface is covered by clouds at any given time.

NASA's Moderate Resolution Imaging Spectroradiometer (MODIS) map shows an average of cloud observations collected between July 2002 and April 2015.

Cloud cover is especially dominant over the oceans where less than 10 per cent of the sky is completely cloud free at any one time. On average only 30 per cent of land surface is cloud free.




Yes, with cross country flights over N. America this has been really noticeable over the past five years. Does anyone know anymore details about the cloud cover numbers, especially the trend?

From that data, in addition to May-June 2015 assuming they collected it, has there been an increase in cloud cover on this hemisphere from the first 5 years starting July 2002 to the cloud cover of the current past five years (5 years starting July 2011)?

It now seems cloudier than I remember the 90s or 80s to be. But maybe it's poor observation or tricks of the mind, or local changes only.
3) Questions and Answers : Wish list : Enhance scheduling/throttling strategies (Message 52473)
Posted 25 Aug 2015 by bl
Post:
Heat is a problem; I find that loading CPUs at low clock rates is really efficient; no fan spinup, and quite good battery life running unplugged under load.

I'm forced to run 8-thread machines at 50% 24/7 to ensure that they're quiet during working hours


You might try reducing the number of cores you're running on as well as the clockrate. Lower clockrates usually mean higher efficiency.

One thing I wish would be an option with BOINC to vary the # or cores depending on whether a machine has been idle for a while or is under use.

To manage heat and efficiency during non-heating season I'm using a rudimentary downclocking tool to automatically assign nice value 19 tasks to downclocked cores.

https://boinc.berkeley.edu/dev/forum_thread.php?id=10461


For windows user, I see that there is a much fancier application; I haven't tried it yet.

https://boinc.berkeley.edu/dev/forum_thread.php?id=3517


4) Message boards : Number crunching : ANOTHER UPLOAD PROBLEM (Message 50718)
Posted 2 Nov 2014 by bl
Post:
oops, my bad---and no such boinc bug after all.

After trying what you suggested I realized my computing prefs we set to allow only for 7 cpu's to work on boinc.

sorry all, and thanks for the feedback/pointer.
5) Message boards : Number crunching : ANOTHER UPLOAD PROBLEM (Message 50708)
Posted 2 Nov 2014 by bl
Post:
Hello everyone.

Yes, I'm experiencing the same problem and was hoping it would get fixed, yet day after day it's still stuck. The stuck units seem to be preventing a different project that requires all 8 cpu's from executing---because of a bug in BOING client I believe, that makes the BOINC client think that finished tasks queued for upload still require a cpu. (My BOINC client may not be the newest, but not ancient; it's only a version or two behind.)





©2020 climateprediction.net