I am on MacBook Pro1,2 Intel Core Duo running OSX 10.5.5
All hadcm3ivolc units fail.
Why should this be?
I am using BOINC Manager 6.3.23

Keith Scott

Not a MAC user -- but the usual problem has been reserved memory allocation (imprecise terminology). See this Topic:
http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=6353

---

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.

I\'ve checked with Tolu.

HADCM3 had to be compiled with the latest Fortran compiler, all other applications were compiled with the previous version. A Fortran dynamic library has to be re-distributed with Mac applications, with a different version being required for the 2 compilers. But the files have the same name, which creates an unavoidable conflict:

• if the first CPDN task you download is not HADCM3 you\'ll be able to run any CPDN application except for HADCM3.
  
  
• if the first CPDN task you download is HADCM3 you won\'t be able to run any other CPDN application.
  

Tolu is working on resolving this, but in the meantime your best option is to modify your project preferences to exclude HADCM3.

---

"The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

I\'ve checked with Tolu.

[/list]
Tolu is working on resolving this, but in the meantime your best option is to modify your project preferences to exclude HADCM3.

Many thanks for explanation and remedy suggestion.
I will put that into effect straight away.
Will we get a notification when Volc (HADCM3) will be supported for Mac Intel?
I would like to know if and when it is.

Keith Scott

Will we get a notification when Volc (HADCM3) will be supported for Mac Intel?

At the very least, there\'ll be a post in News and Announcements on all of the boards.

---

Backups: Here

I\'ve checked with Tolu.

HADCM3 had to be compiled with the latest Fortran compiler, all other applications were compiled with the previous version. A Fortran dynamic library has to be re-distributed with Mac applications, with a different version being required for the 2 compilers. But the files have the same name, which creates an unavoidable conflict:

• if the first CPDN task you download is not HADCM3 you\'ll be able to run any CPDN application except for HADCM3.
  
  
• if the first CPDN task you download is HADCM3 you won\'t be able to run any other CPDN application.
  

Tolu is working on resolving this, but in the meantime your best option is to modify your project preferences to exclude HADCM3.

What chance is there that this Fortran problem will be resolved?
I would like a chance to run a volc unit as soon as this is done.

As it is some 3 months or more since this problem arose, I am sure that many crunchers have been deterred from using CPDN after repeated \"computation errors\". At least a warning and help message would be helpful to crunchers such as these, instead of waiting until (as I did) for exasperation to take over and a search of the forums made eventually to find the reason and remedy.

Even on the preferences page, a warning would be extremely helpful (and informative where it is likely to be viewed) instead of a bland statement that HADCM3 is available to Mac intel with no qualification.

Is there any likelihood of resolution of this Fortran problem in the near (or distant) future?
And, if so, when, please?

Keith Scott

Hi Keith

There is at least one other serious unresolved bug in certain CPDN models, though it doesn\'t affect Mac users. It\'s been there for a long time. The root of the problem is that CPDN has only had two programmers for a year and eight months since Carl left. Two are not enough for a project the size of CPDN with very complex models. Our chief programmer, Tolu, spent a large part of 2008 making the models BOINC 6 compatible. There have been new model types and new experiments with existing model types. In addition, completely new and very different models and experiments are currently in alpha and beta testing. One planned experiment is way behind schedule. Our two programmers, Tolu and Milo, are very overworked.

That\'s an explanation of the situation, not an excuse, but I hope you\'ll understand that none of us can promise an early resolution of this problem (or indeed any problem apart from server functionality which is always top priority). I can see that it\'s very frustrating not to be able to run a particular model type; many of us like to try all the models, not just some of them.

I will check whether Thyme Lawn\'s explanation of this problem on Macs is in the CPDN READMEs and if it isn\'t I\'ll add a link. I\'ll see whether a warning to Mac users can be added to the CPDN preferences page.

The moderators try to announce all bugs and problems in the forum News threads. Apologies that we didn\'t announce this one. (You\'ll find the News thread at the top of Number Crunching on this forum and can subscribe to that thread to receive an email message every time something new is announced.)

The good news is that Tolu and Milo are now getting some extra programming help and there\'s a possibility of more. I\'m keeping my fingers crossed that the current economic situation, which is already affecting UK universities, will not scupper these plans.

---

Cpdn news

Experiment time. (Or should that be )

Set project to No new work
Run down all current climate models.
Exit FULLY from BOINC.
Make a backup of the BOINC data folders, just in case.

Carefully delete all Application files for the models > only <.
(These should have names starting with the type of model(s) that you CAN run.

Set your server prefs for ONLY the volc models.
Re-start BOINC.
NOW try to download and run a volc model.

Perhaps some other files/file-type will need to be deleted also. I\'ve no idea how the Mac OS works/stores things.

---

Backups: Here

Actually, come to think of it, that\'s what the Reset project button in the projects tab does: kill all the work units and associated files.
Have you tried this?

THANKS to Les Bayliss, mo.v, & Thyme Lawn

Excellent and prompt replies, precise and to the point, regarding the issue reported.

My comments have been taken as constructive criticism and have very helpfully been acted upon in that manner, which is most welcome and to be congratulated.

THANKS also to Carl for all the work he has done, and to Tolu & Milo for keeping up the good work.


I will take your advice on a reset, Les, using preferences forcing my Mac to use other fortran compiler.
But I have several units to be completed before I do that, which will take about 14 days.
(I am shepherding 7 units to complete at around the same time, which you may find somewhat unusual.)

THANKS again, all of you.

Keith