I have an old G4 (350MHz) that is currently crunching its way through a sulphur WU. I\'m attached to seven projects, so when I went to check on the status of my G4 and saw only two projects listed (one of which, an Einstein, was done and ready to report)--especially because my Powerbook was happily gorging on close to 20 work units (I set the work buffer too high)--I was curious and started investigating. Then I saw the entry in the log: it\'s using earliest-deadline-first scheduling.

Then I looked at the estimated time to finish: 9327 hours. According to Google\'s useful calculator, this is 388 days, or almost two months past the deadline data.

And this is assuming CPDN uses 100% of my processor time and doesn\'t begin working on any other project.

I believe that (based on reading stuff on this forum and elsewhere) that CPDN doesn\'t enforce the project deadline as long as the computer is trickling results back to the servers. Regardless, though, an entire year is a long time to commit to the project: what if the computer dies, the drive crashes, the WU gets corrupted, etc.--that\'s potentially a year\'s worth of calculations lost.

So, my question is thus: what should I do? Should I go ahead and leave it be and have it work on the climate model for the next year? Or should I suspend the project and donate my time to curing AIDS and cancer? (Er, not AIDS yet--Africa@home isn\'t Mac yet, I believe).

Or, would my old machine work fine with shorter CPDN WUs like slab models? If that\'s the case, is there a way to assign a preference to only get the slab models on one particular machine?

I knew CPDN was computationally intensive...

---

There are no more slabs. At the moment there is only sulphur.
Soon there will be an even longer model, the coupled ocean model, (experiment 2).
Sulphur models will probably be unneeded in a year from now.
However, phase 1 of sulphur models contains data that the researchers need, so at least keep going to this point.

BUT: It\'s possible that there will be no more \'Mac\' models.
Now that Apple has moved to Intel processors, it appears that future models for the \'Mac\' will be Linux, so that the programmers can use their Linux compiler for Mac and Linux. But it\'s only at the rumor stage, and you know what they are like.

With the G4 in EDF mode your only interaction with the other projects is going to be to report the Einstein result. The CPDN model will almost certainly be run exclusively until it finishes, when the other projects will have a massive processing debt to make up.

Unfortunately there\'s no way of forcing CPDN to download a less intensive model.

Only you can make the decision on what to do about the sulphur model you\'re running, but as you\'re running other projects you might want to set the project to no work fetch, abort the result (so it\'s sent out again more quickly than if you just detached) and then detach from the project.

---

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

However, phase 1 of sulphur models contains data that the researchers need, so at least keep going to this point.

How far in is Phase 1? I\'m willing to let it go there if it can make it there sooner than a year.

If Phase 1 is more than about 25% of the WU, I\'ll go ahead and detach and let someone with a faster CPU work my model.

---

Five phases, so 20%.

Five phases, so 20%.

80 or so days? I guess I\'ll stick with it...of course, this is assuming the percentage climb is linear.

Thanks for your help.

Chris

---