I have been waiting for some weeks for a workunit to finish so when it failed at 100% my reaction veered between disappointment and anger.

I see the same workunit 8398079 has also failed for someone else who is using a different operating system (Darwin v Xubuntu) and processor (Intel v AMD).

I don't mind wasting a few hours on experiments but wasting WEEKS of processor time is very annoying. I reckon I've put about 20 kWh of electricity into this project to produce failed results.

Should I give up on this project?

According to the trickle data, that model reached the end and sent back all of the data.
It's just that for unknown reasons, this type of model sometimes "just doesn't know when to quit" when they're finished. And some, like yours, fail shortly AFTER they finish.

The "Darwin" person's model failed long before it got to the end, apparently because of something wrong with the computer.

As for "failed results", I suppose that you know that the purpose of this project is NOT to complete them? Rather, it's to see how far they run, with the values used for the forcing variables used.
The last trickle number received gives the researchers a good idea where this is for those models that become climatically unstable. Which informs them a little more about how well the models mimic the real weather/climate.

If you want to know more about the project, start here at the front page, and click on Climate Science in the menu to the right. There's basic info, plus links to some of the research papers that have been published.

As giving up, that's up to you.

---

Backups: Here

As for "failed results", I suppose that you know that the purpose of this project is NOT to complete them?

Ah, no, I hadn't picked up on that. I understood it was validating models, but did not realise 'Error while computing' equated to 'had enough of this model thank-you-very-much' for this project. I had just assumed it would terminate the work package tidily.

Thank you.