One more try

F. Prefect
F. Prefect
Joined: 7 Nov 05
Posts: 135
Credit: 1,016,868
RAC: 0
Topic 190161

After spending over 3 hrs digging into the help files of the BOINC manager and finding nothing that accurately describes what's happening, let me start from the beginning.

Several days ago I installed the BOINC manager on 4 machines (3 Athlon T-bird cores)running at approx 1.2, and 1 Athlon 64 3400. On one of the t-birds I attached to the Einstein project, the other 3 I attached to the Einstein project as well as the Rosetta project.

Everything was running smoothly until my 2nd Einstein job had a completion time of over 80 hrs. and I received a computational error upon it's completion. In a attempt to rid myself of this job, one thing lead to another and I eventually uninstalled and reinstalled BOINC on that one machine. I read somewhere in the help files that account information can be "transferred" from one machine to another when connected. It appears something to that effect occurred here. The first thing I noticed was that the Rosetta jobs upon starting would very quickly run up to 10 to 30% completed and then seemed to stop. The %completed would remain unchanged and the time to completion would actually increase with time. I suppose it would be possible that the job ran up to 30% in 2 or 3 minutes because there was very little "work" in that part of the job and BOINC was attempting to "correct" the completion when the job reach a point where in ran more slowly.

Anyway I've been unistalling and reinstalling until I'm about to drop. As of right now I only have Einstein installed on all 4 machines and 2 of the 1.2 Athlons are running at a pace that seems to be about 3 times that of the Athlon 64.

If it's "normal" for some jobs to run very quickly through the first part of a job, I may no have a problem, but none of this began to occur until I had that 80 hr Einstien job and eventually uninstalled and reinstalled the BOINC manager. Done.

F. Prefect

In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.....Douglas Adams

F. Prefect
F. Prefect
Joined: 7 Nov 05
Posts: 135
Credit: 1,016,868
RAC: 0

One more try

Quote:

After spending over 3 hrs digging into the help files of the BOINC manager and finding nothing that accurately describes what's happening, let me start from the beginning.

Several days ago I installed the BOINC manager on 4 machines (3 Athlon T-bird cores)running at approx 1.2, and 1 Athlon 64 3400. On one of the t-birds I attached to the Einstein project, the other 3 I attached to the Einstein project as well as the Rosetta project.

Everything was running smoothly until my 2nd Einstein job had a completion time of over 80 hrs. and I received a computational error upon it's completion. In a attempt to rid myself of this job, one thing lead to another and I eventually uninstalled and reinstalled BOINC on that one machine. I read somewhere in the help files that account information can be "transferred" from one machine to another when connected. It appears something to that effect occurred here. The first thing I noticed was that the Rosetta jobs upon starting would very quickly run up to 10 to 30% completed and then seemed to stop. The %completed would remain unchanged and the time to completion would actually increase with time. I suppose it would be possible that the job ran up to 30% in 2 or 3 minutes because there was very little "work" in that part of the job and BOINC was attempting to "correct" the completion when the job reach a point where in ran more slowly.

Anyway I've been unistalling and reinstalling until I'm about to drop. As of right now I only have Einstein installed on all 4 machines and 2 of the 1.2 Athlons are running at a pace that seems to be about 3 times that of the Athlon 64.

If it's "normal" for some jobs to run very quickly through the first part of a job, I may no have a problem, but none of this began to occur until I had that 80 hr Einstien job and eventually uninstalled and reinstalled the BOINC manager. Done.

F. Prefect

If I could delete the above post I would. I believe I may have solved the mystery. I won't go into the embarrassing details, but it appears to have been a case not seeing the forest because there weren't any trees to be seen. Time will tell.

F. Prefect

In the beginning the Universe was created. This has made a lot of people very angry and been widely regarded as a bad move.....Douglas Adams

Michael Roycraft
Michael Roycraft
Joined: 10 Mar 05
Posts: 846
Credit: 157,718
RAC: 0

F. Prefect, Regarding your

F. Prefect,

Regarding your observation about Rosetta %/completion time, Einstein processes extremely linearly, whereas Rosetta and Seti, among others, do not. What I'm saying is that each % of an Einstein WU requires the same amount of processor time as the next %, Rosetta and Seti go through the first 15-20% fairly quickly and then run into the harder, more intense calculations, and require more processing time per, they "bog down".

As far as feeling embarrassed, LOL, don't worry - it's a rare one who hasn't wished to eat back a few words.

Michael

microcraft
"The arc of history is long, but it bends toward justice" - MLK

John McLeod VII
John McLeod VII
Moderator
Joined: 10 Nov 04
Posts: 547
Credit: 632,255
RAC: 0

I am going to suppose that

I am going to suppose that one of the trees is that an uninstall of BOINC does NOT remove any of the project data. BOINC will normally sort out results that have computation errors all by itself.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.