I have started running a climateprediction.net HadCM3 Coupled Model Experiment 5.15, which started out with a (Time) \"To Completion\" of 2952:15:32.
It has been running for several hours, and the (Time) \"To Completion\" remains the same, with the minutes and seconds fluctuating, while the hours remain the same.

Is it due to the computer being unable to actually calculate this [(Time) \"To Completion\"], or is it an error?
If so, is there anything I can do?

Thanks.

It\'s not an error. Due to the length of time it takes for any such model to run, BOINC can\'t figure out how long it\'ll take to run this one. Not until you have run it for a couple of weeks will the time come down.

---

Jord.

It\'s not an error. Due to the length of time it takes for any such model to run, BOINC can\'t figure out how long it\'ll take to run this one. Not until you have run it for a couple of weeks will the time come down.

Is there some formula that I can use to calculate the \"Time to Completion\"?

I find it\'s more useful to have an idea in real-world time rather than computer hours or seconds. You can see how long it takes to process one model year ie from trickle to trickle, Dec to Dec. Multiply by 160 for the number of model years, then add computer downtime for housekeeping (reboots, scans, defrags, cleanups, downloads, backups) and days or weeks you may be on holiday with the computer turned off. If the model crashes and you have to restore a backup, that will probably add a few days.

My 2 models are doing 1.56sec/TS (2 model years per day each) and the shortest time I reckon I can finish them in is 90 days approx for each, ie 3 months. That\'s running 24/7. But I have a model on an old computer that was one year old yesterday and still has a lot of work to do.....

From the point of view of the science, reliability is more important than speed. The way to increase reliability is to check out the READMEs (see my sig), particularly the top tips in the one about running the model, and items #1 by Les and #5 by Mike in the one about problems and crashes.

---

Cpdn news