I just reinstalled a few days ago from a new DL of Boinc due to a system crash. Running a P4 2.8 with WinXP Pro.
Other than occasional software freezes seemingly fixed by disabling the boinc SS My most pressing problem is that the time to completion on 2 sulphur models is rising at a rate about 1/3 of the cpu time.
I can\'t tell if I\'m getting credit from this machine as I\'m running it on the same user account from 2 other machines.

---

Brian

To find out which computers are getting credits, go to \'your account\' (on the left) and select \'view\' on the computers line. That will bring up a list. Some of those will represent the same machine, but select one and you will find a merge option. You will also find a list of trickles.

A good source of information on this and other things is the BOINC-Wiki.

On the subject of the rising completion time, the system crash probably corrupted a data file and the WU has probably rewound to an earlier point. You won\'t get credits until you pass the point you reached before.

---

Brian

To find out which computers are getting credits, go to \'your account\' (on the left) and select \'view\' on the computers line. That will bring up a list. Some of those will represent the same machine, but select one and you will find a merge option. You will also find a list of trickles.

A good source of information on this and other things is the BOINC-Wiki.

On the subject of the rising completion time, the system crash probably corrupted a data file and the WU has probably rewound to an earlier point. You won\'t get credits until you pass the point you reached before.

I forgot to mention, I tried saving my original WU\'s but ended up detatching & dumping then reinstalling from scratch with new WU\'s. It was frustrating as I was 2/3 through a sulfur model. I am letting it run anyway to see what happens but I have 23+ hours of cpu time each on 2 models and the completion time has risen about 3 hours if I remember the starting point correctly. It fluctuates up and down with an upward trend.

---

I have 23+ hours of cpu time each on 2 models and the completion time has risen about 3 hours if I remember the starting point correctly. It fluctuates up and down with an upward trend.

BOINC isn\'t very good at calculating times for climateprediction. It\'s method of calculation is really designed to suit other projects, though the latest version is said to be better than some. The seconds per timestep figure and the progress made are your best guide to how things are going.

---

BOINC isn\'t very good at calculating times for climateprediction. It\'s method of calculation is really designed to suit other projects, though the latest version is said to be better than some. The seconds per timestep figure and the progress made are your best guide to how things are going.

And it is even worse at remembering how far along it is. I am running 5.2.12 on a PIV Hyperthreader at 3GHz and every time BOINC shuts down (for computer reboots mostly), it thinks it will take about 2500 hours to finish a Sulphur model. Since this is one of the short turnaround models (due in February), it thinks the computer is overcommitted, goes into EDF mode and finishes all other project WUs first. When those are finished, it restarts CPDN, figures out fairly quickly that it will only need about 950 more hours (in fact, it will need about 600), and goes into normal mode. Of course, the debt now keep CPDN running on one (logical) processor pretty much full time.

In the scheme of things, it really doesn\'t matter much, I think those short turn arounds are all gone now and it shouldn\'t go into EDF if it thinks it has almost a year to go. But it is really weird to watch (one of the earlier versions was even weirder, but that\'s history).

Bill - Did you try an upgrade to the latest BOINC version? Did it help? I ask, because it has been claimed that it should be better than early 5.2 versions in this respect. I know that you are not the only one to have had this problem.

---

Bill - Did you try an upgrade to the latest BOINC version? Did it help? I ask, because it has been claimed that it should be better than early 5.2 versions in this respect. I know that you are not the only one to have had this problem.

Yes... I\'m running 5.2.12. My problem is that when it shuts down, it forgets that it has done anything and thinks it will take 2000+ hours or so to finish. On restart, the scheduler goes into EDF mode. The CPDN WU isn\'t due until February, so it is last to run and the cache is emptied of other project WUs. Once CPDN runs a few minutes, it figures out the correct (more or less) time to complete and all is well, the time remaining comes down.

I don\'t shut the machine down all that often, but it is Windows, so it does get restarted every few days for one reason or another.

As far as the correction factor, before I went to 5.2.12 (from 5.2.6 or 7), the error had come down to about 18%. On two other computers, it is about the same. The version change seems to mess the correction factor up for a while until it readjusts (which seems to take several days). That part seems to be working as advertised for CPDN (as well as SETI, Rosetta, and Predictor as well).