I've tried to wait for the new Boinc (it's great and goes without any problem with all the other projects - Seti, Protein and Einstein), to see if the problems with climate predictions disappeared, but here I am again, seeing that climate predictions ends before time running single packets of data (very small, as if it didn't unzip them), and uploads them reporting a computation error. This occurred both in Panther, with a G5 and permissions repaired (and the library is in order), and in Tiger. Now I've decided to suspend the work, 'til someone finds a solution.

---

Lana

I have the same problem it seems. I'm on a G4 iMac 1GHz. I'm told that the data is 100% complete (very fast!!) but under the status column, is written Computation error. My messages show computational times in only seconds!

Can anybody out there help us please?


Andrew

Lana

I found another message on this forum with instructions that work! I seem to be up and running after following them carefully. They appear in a message with solution written by Clay (THANKS! Clay) at:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=2635

I copy the solution here to save time going to that message and to increase the chance of others finding the solution:

Clay
Joined: Mar 1, 2005
Posts: 2
ID: 59201
Credit: 37,523
RAC: 1,179
Message 12976 - Posted 31 May 2005 15:58:55 UTC
Last modified: 31 May 2005 15:59:45 UTC
As to Climate not running: the current setup is looking for a few library files with an erroneous path. To make climate run, I simply put these files where the model expects them.

- Create a folder name \"opt\" on the top level of your boot drive. In it create a folder named \"ibmcmp\" and in that folder create one named \"lib\".

- Copy the following three files: \"libxlsmp.A.dylib\" \"libxlfmath.A.dylib\" and \"libxlf90.A.dylib\" into the \"lib\" folder. You will find these files in /Library/Application Support/Boinc Data/projects/climateprediction.net/. (If running Menubar, look for a similar location in your hime area.)

- Restart climate and all should be well. (Though I had one model crash out of the box. BOINC promptly threw it away and got a new one which ran, so no problem with that.)

As to scheduling, I experience the problem as well, but only on my dual CPU macs. It seems to be most troublesome with Einstein, so I\'ve discontinued Einstein on my duals. However, at times I do see the duals (with Climate, Protein & SETI) trying to run 3 threads simultaneously. Though this reduces overall productivity of BOINC (due to decreased memory cache effectiveness) it, IMO, isn\'t that big a deal. Another poster mentioned that there is a 4.44 in the works to address scheduling problems.

ID: 12976 / Score: 5



Hope this works for you and whoever else is battling to get going.


All the best


Andrew

---

I see I\'m using Boinc Manager 4.44 -- but there\'s a v5.2.13 available tonight at the UCBerkeley download page. I\'m seven percent into a \'Sulfur Cycle\' that will finish up, around, let me see, yeah, about when hell freezes over, at this rate (grin).

I realize partial work is useful so I\'m keeping it running (OSX 10.3.9, 900mnz G3), learned that tonight in the forums.

What about the Boinc version -- is it possible to halt, or pause, a work cycle in order to update the Boinc software? IF so, will it work on this old G3, and do I need to download the \'extra applications first\' as the site warns?

I\'ve just left it running and not paid attention for several months, now catching up on current events and finding I\'m puzzled.

---

Hank
You need to be out of BOINC to upgrade.
I\'d suggest that you Suspend, wait until the model says Suspended, then Exit.
Then make a backup of the entire BOINC folder.

And yes, you do need to download the extra apps. first before starting a model.