Please can someone give me a pointer?

I have been running CPDN for years on several machines. I am not IT illiterate, but equally not expert. One by one, the machines on which I have been running it have worked through the tasks that they were assigned and run out of work. This is not unusual if there are no jobs available to download, and so I have been monitoring the situation, but not doing anything about it. Yesterday, after several months of inactivity on CPDN (but BOINC running perfectly happily with Rosetta) I decided I should just check whether or not there were any jobs available. I looked and there were 84 jobs available, so I clicked on 'Update'. I got a 'Communication deferred' message (for an hour).

Nothing has changed in my setup which has served me faithfully for many months/years. What steps should I take to double-check that my preferences aren't precluding me from downloading CPDN jobs please?

Hi Seb

You need to do 2 things:
Look at the application name in Project status
and then
Look at the Applications page to see which computer OS that type is for.

At the moment, there's only LInux work, with no hint of when Windows work may show up.

so I clicked on 'Update'. I got a 'Communication deferred' message (for an hour).

Another thing worth knowing Seb,

Even once windows work is available, if you have failed to get work and you press the update button again before the hour is up, it just starts the hour again without asking the project for work even if tasks for your OS have appeared on the servers in the meantime.

Thanks both. It never occurred to me that jobs might be OS-specific. It's not logical to me that anyone wanting to utilise distributed computing would want to restrict themselves by OS-type, but I guess it's just a question of the limits of their own resources in terms of setting up the jobs in the first place.

I only really joined BOINC to contribute to CPDN so it's frustrating not to be able to do so. Absolutely not your fault and thank you for explaining. Amazed at the extent of my own iggorense! I shall apply my CPU time where it's at least useful - I had the machine sitting idle awaiting jobs for a while before I decided that was wasteful. Let's hope that CPDN get some Windows work up soon.

Thanks again.

Hi Les, appreciate the mentioning of the fact that there is "currently" no Windows work. Suggest the following:

The application page is "nice" for me as a user (to explain what is computed where), but not particularily useful from the viewpoint of a user like Seb who would like to contribute. Maybe I have not found the appropriate page which contains relevant information. So what I would like to see on this(?) page per App is:

* If the app is "activ"e that is crunching work or not (could be a RAG-radio button)

• If green radio button, then: TOTAL, ETC for work including projected end date (at the current rate of crunching - I am sure the respective project leads are asking for this information as well
  * If yellow (no crunching currently) WHEN is work scheduled to come up for grabs

Brgds, Florian[/list]

---

Thanks both. It never occurred to me that jobs might be OS-specific. It's not logical to me that anyone wanting to utilise distributed computing would want to restrict themselves by OS-type, but I guess it's just a question of the limits of their own resources in terms of setting up the jobs in the first place.

It isn't just about the resources of CPDN either. The OpenIFS tasks which are the most recent type come from a European collaboration an that particular one has only been developed for Linux and Mac which like Linux is a Unix based system. Some other model types that run on Windows have been attempted on Linux but some bug causes just about all tasks to crash. It is much simpler to develop tasks for single platforms. The alternative is to go down the virtual machine route. there are projects that have done this but that in itself adds overheads.

The application page is "nice" for me as a user (to explain what is computed where), but not particularily useful from the viewpoint of a user like Seb who would like to contribute. Maybe I have not found the appropriate page which contains relevant information.

Information about what is current is available on the server status page
This shows what model types are current. The users in last 24 hours column tells how many users have completed tasks of that type.

Sadly most of the time there is very little information in advance of work being released and on a number of occasions I have put stuff out believing the work is on its way and something has happened to delay it by days/weeks/months so unless fairly sure I have stopped doing it.

Well, I have spare Windows cycles, too.

Will just await additional work.

I know the idea is to get a lot of things done quickly, hence the desire to develop at low levels. However, while machine and OS independence isn't entirely possible, the whole idea of using HOLs is to be able to run the same thing on multiple architectures. So, if something like R or Python 3 were used, that would be easier. Yes, it would be slower than a compiled, but, then, you could access a bigger footprint of systems.

Just my two cents.

- Jan

There are a limited number of research groups around the world that are part of this project, and several of them haven't been heard of for over a year.
And even then, only small numbers of tasks are needed to provide the data that's needed.

So, with only a couple of thousand tasks at a time, and over 20,000 computers, lots of them with a large number of processors, work doesn't last long.

It's just the way this project is.