Hi there,

I'm having difficulties to understand what's happening with ClimatePrediction on my Mac

I'm running SETI, Rosetta and ClimatePrediction.
First two work fine, calculating and reporting OK.

But ClimatePrediction always report Error while computing:

Task ID
click for details
Show names Work unit ID
click for details Computer Sent Time reported
or deadline
explain Status Run time
(sec) CPU time
(sec) Claimed credit Granted credit Application
17034162 9129941 1337163 14 Sep 2014 10:08:58 UTC 14 Sep 2014 11:09:41 UTC Error while computing 17.69 0.27 0.00 --- UK Met Office HadCM3 short v7.24
17027271 9133138 1337163 13 Sep 2014 11:00:00 UTC 20 Sep 2014 20:17:17 UTC Error while computing 62,741.34 59,707.57 622.08 622.08 UK Met Office HadCM3 short v7.24
17027188 9132710 1337163 13 Sep 2014 11:00:00 UTC 13 Sep 2014 12:00:39 UTC Error while computing 15.36 0.24 0.00 --- UK Met Office HadCM3 short v7.24
17004457 9132374 1337163 13 Sep 2014 12:00:39 UTC 13 Sep 2014 13:01:19 UTC Error while computing 16.10 0.23 0.00 --- UK Met Office HadCM3 short v7.24
17004412 9132330 1337163 13 Sep 2014 13:01:19 UTC 13 Sep 2014 14:18:52 UTC Error while computing 15.36 0.24 0.00 --- UK Met Office HadCM3 short v7.24
17001571 9129583 1337163 14 Sep 2014 11:09:41 UTC 21 Sep 2014 13:21:20 UTC Error while computing 60,718.20 57,807.55 622.08 622.08 UK Met Office HadCM3 short v7.24
16891986 9077634 1337163 23 Aug 2014 22:53:16 UTC 05 Sep 2014 18:33:57 UTC Error while computing 67,704.87 62,144.99 622.08 622.08 UK Met Office HadCM3 short v7.24

Is there anything I can do to fix this or is this a Bug on the OSX-Version of ClimatePrediction.
I don't really see the point in "wasting" CPU-time if the the effort is not usable anyway, ...

Would be nice to get some information how to fix this.
Thanks and Best Regards

Robin

This model type seems to be very susceptible to problems if stopped and restarted. If you are unable to run them without interruption on mac and windows machines it is probably best to exclude them via your account page. Strangely they all seem to be working fine on my linux box with the exception that they don't clean up after themselves leaving a folder almost 1GB in size after they have reported.

It may be worth going into your projects/CPDN folder to see if the ones that have crashed have done that so as not to waste disk space.

Never having had a Mac, you will have to look through posts in this section if you don't know where this folder is as I don't have a clue!

Hi Robin

You have two problems:

1) A few of the models crashed with: INITTIME: Atmosphere basis time mismatch.
This is a problem with some of the data sets in that series, and was discussed in one of the threads in Number crunching.
So, not your fault.

2) The rest crashed with: Suspended CPDN Monitor - Suspend request from BOINC....
This is probably due to you using the default setting for the Computing preferences option: Suspend work if CPU usage is above .
This setting of 25% is fine for other projects, but, as Dave said, these climate models don't like continually being stopped and started. So, sooner or later, they usually crash.

Am also getting the same message "Suspend request from BOINC" I had a look at preferences but they are set to use the computer anytime and all of it so far as I can tell. The 25% is not checked. I am running Mac 0SX Ver 10.5.8.

Like Robin I wonder if I am wasting computer time on this. Any deas anybody?

Helen (Helen & �live)

To start with it may well be the current batch of work at fault, the only problem free platform at the moment being linux as far as I can see.
To minimise the chances of your settings contributing to the problem,
Ensure under tools computing preferences disk and memory usage that leave tasks in memory while suspended is checked, under processor usage, while processor usage is less than set to 0% which means no restriction.

Thanks Dave, I've done both those things so I'll see what transpires. Helen

I've always had problems with the reliability of ClimatePrediction on my Mac. I recently had a task running since June and failed at well over 80%. Am I wasting my resources or is this useful to the project. I am seriously considering whether I should abandon ClimatePrediction. I don't have these issues with SETI or Rosetta.

---

Task that failed is 18615058.

---

Hello Nigel

Yes, it's possible that you are wasting your time.
The researchers want their results ASAP, and the "deadline" is just to accommodate BOINC, and not a project deadline to finish.

Partial results aren't very useful, as each "model" is just part of the long string of modelling that results when each segment is joined to the others either side of it.
To do this, the final zip file, 13, is needed, as this contains the info to start the next segment.

The model in question has been discussed extensively in other parts of the forum, and it's considered one of the tougher ones to complete, as it needs to be run without interruption. There are a large number of Suspends in the error log of the one that you mentioned, which may be why it eventually failed.

ALL of the other projects have trivial maths compared to this project, which has long been considered to be "not for everyone".

Greetings,

I recently upgraded to El Capitan and the next thing I know my Climate prediction task got stuck on a "completed percentage" of 99.762. This continued for some days all the while clocking up the hours and minutes but the percentage remained the same. Eventually it stopped with an Error while computing.

Could this have been because of El Capitan or because I had shut down the computer while I went out of state? (i.e to New South Wales)
Since then I have not received any tasks from CPDN.

Will CPDN be providing code for the newer Mac operating systems?

At the moment SETI is getting 100% of my computer time.

Any infor gratefully received,
Thanks in advance,

Pramsec

Peter:

If the model is hadcm3n_sc85_1980_40_010021939_2, then these HADCM3N models were prone to problems on Zip file uploads and finishing when interrupted (including a never-ending loop at the end) - and an OS version upgrade would certainly count as an interruption. So I would not conclude that there is an El Capitan angle to that model's problems. The good news is that model probably uploaded all its data as the loop generally occurs after the last Zip had been sent.

The only model available for Mac at the moment is the ANZ regional model, which does appear in batches from time to time. I've no idea about future plans for Mac.

Hi Les! Thanks for the info. Does this apply to all the hadam3p models? Looks like I will have to restrict processing to hadcm3n for now.

Just tried to run another task which aborted with computation error. I'd like to help the project but there doesn't seem to be much point running it on MACs.

---

The only modelling program that runs on Macs is the hadcm3s, and tasks using this program have a high failure rate on Macs.

Also, 32 bit support for Macs has stopped, and Apple is also planning to move away from Intel processors to risc type processors.

So, with the very limited number of tasks available at present, it's not really worth while trying to use a Mac on cpdn any more.

Apple Confirms Mojave is the Last macOS Release to Support 32-Bit Apps

Apple planning to replace Intel CPUs in Macs with custom ARM processors by 2020