It is interesting one year ago CPDN had over 47K tasks actively being crunched. Today it is less than 100 Tasks being worked.

Bill F

In reply to Bill Freauff's message of 1 Jun 2026:

It is interesting one year ago CPDN had over 47K tasks actively being crunched. Today it is less than 100 Tasks being worked.

Bill F

That 47K will have included many tasks that had been abandoned from model types that have not been used for years. Since then they have had a clear out. only the tasks from the two current batches are now shown as active. The first of 2048 tasks has 95% of the tasks completed and validated. The second smaller batch is just 200 tasks with a code change to allow up to 8 cores per task to be used. Assuming the smaller test batch is successful, all work using the code from the European Centre for Medium range Weather Forecasting will in future allow up to 8 cores per task.

Another reason for the change is that CPDN has switched to deadlines of a couple of months rather than a year which was a hangover from the days when tasks could take over 6 months to complete on slower machines. (I successfully completed tasks on an Intel Atom netbook to prove it could be done! The downside is that as climate models become more complex taking more variables into account, they do become more demanding and require more recent hardware to run. (The current multi-core tasks require a smidgeon over 26GB of RAM per task at peak demand. At least now the figure on the server status page of 180 tasks in progress is fairly accurate.

It would be nice if there were a more constant supply of work from CPDN but for some years now, the work has not in the main been generated by Oxford Uni but comes from research projects in universities around the world ranging from masters degrees to post doc work.

News from Glenn, Honorary Researcher

CPDN are pleased to announce we will be adding the Weather Research and Forecasting Model (WRF) to CPDN. WRF is a state-of-the-art atmospheric modeling system from NCAR in the USA. It is designed for both meteorological research and numerical weather prediction. It offers a host of options for atmospheric processes and can run on a variety of computing platforms. It has a long history of development and extensive use in the meteorological community.

Development work to add WRF into CPDN is well advanced and we plan to deploy a test version of the application in the next couple of weeks. More technical details will be available then, however, it will be linux-only to begin with and multi-core.

It will be used for a project looking at urban heat islands led by scientists at Oxford University.

The addition of a new model type to add to the OIFS and Met Office models increases the number of scientists who can use CPDN for their work. I look forward to seeing them!

Uploads to CPDN are not working following a power outage in the Oxford area. A statement on the page for CEDA and JASMIN services posted yesterday said today at the earliest for resumption of normal access. Some services running slowly, some need rebuilding.

Still PFS (Parallel File System) stopping uploads. Also a warning on the status page about services being throttled due to the Red temperature warning.

We had a warning from a couple of (UK based) data centres that they were going to be reducing power to the racks because the cooling systems are not intended to run with ambient temperatures above about 30C. What does this mean? Well it all depends on how the racks are configured, some will be able to reduce clock speeds on the processors, some will shed very low lad servers, and others will just wave a white flag and shut down.....

In reply to robsmith's message of 24 Jun 2026:

... because the cooling systems are not intended to run with ambient temperatures above about 30C.

Which bit of AI hasn't been trained on climate change yet?

In reply to Richard Haselgrove's message of 24 Jun 2026:

In reply to robsmith's message of 24 Jun 2026:

... because the cooling systems are not intended to run with ambient temperatures above about 30C.

Which bit of AI hasn't been trained on climate change yet?

Isn't AI by its power hungry nature designed to accelerate climate change?

In reply to Dave's message of 24 Jun 2026:

Isn't AI by its power hungry nature designed to accelerate climate change?

If it goes on enhancing itself without joining that particular set of dots, that's suicide for AI and mass extinction for us.

And nooooobooddyyyyyy has seen that comiiinnggggg....sarcasm off

---

---------- 24/7 Crunching since 2011 -----------
DSKAG Austria: http://www.dskag.at