Is everyone running dual G4s having problems?

I'd have to say yes for me for CC 5.2.4 on a Dual 1.0 GHz G4.

https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=436

However, either I just got lucky and got some good WUs, or installing CC 5.2.5 fixed the issue. If I see more errors, I'll just "no new work" for a while.

---

-----
derekm
Team MacAddict

update on my little problem. since updating boinc to version 5.2.5 on my athlonXP i had no further errors with the 1n0u units, my duron is still on 5.2.2 and doesnt have any problems with them.

---

update on my little problem. since updating boinc to version 5.2.5 on my athlonXP i had no further errors with the 1n0u units, my duron is still on 5.2.2 and doesnt have any problems with them.

I wish it was that simple for all of us. I am running CP@H on my problem machine and I am almost 25% of the way through the current model. The last time I tried to upgrade at version 5.1, I lost two models that were 50% complete. It will be at least two months before the current model completes. If I have to upgrade BOINC to fix this problem it won't happen till then. It seems that there should be some level of backwards compatability for the science apps. Not everyone can just turn off BOINC and upgrade for a single project.

Regards
Phil

---

We Must look for intelligent life on other planets as,
it is becoming increasingly apparent we will not find any on our own.

i am running climateprediction too on that machine, current model over 80% complete. i upgraded the boincversion more than one time within running that model and never had a problem, i do make backups just in case something goes wrong though.

---

i am running climateprediction too on that machine, current model over 80% complete. i upgraded the boincversion more than one time within running that model and never had a problem, i do make backups just in case something goes wrong though.

Doc,

Based on your post i have tried the upgrade. As you indicated this time it seems to have worked without trashing my CP@H model (so far). Now we will have to see if it solves the R@H problem. Of course you know if it doesn't work I will hunt you down (LOL).

I stil think that at this stage the project should have greater backwards compatability with BOINC versions. But if I can lighten David's load by finding a fix in this well that's ok too.

Thanks and Regards
Phil

---

We Must look for intelligent life on other planets as,
it is becoming increasingly apparent we will not find any on our own.

Based on your post i have tried the upgrade. As you indicated this time it seems to have worked without trashing my CP@H model (so far). Now we will have to see if it solves the R@H problem. Of course you know if it doesn't work I will hunt you down (LOL).

I still think that at this stage the project should have greater backwards compatability with BOINC versions. But if I can lighten David's load by finding a fix in this well that's ok too.

Thanks and Regards
Phil

Well, It seems to be running R@H again. But BOINC 5.2.5 is a real reasource hog compared to earlier versions, and it takes longer to process each WU for all the apps than it did before. But at least the first one through did not error out.

Regards
Phil

---

We Must look for intelligent life on other planets as,
it is becoming increasingly apparent we will not find any on our own.

5.2.5 is not using more resources than earlier versions for me (basically 0), glad to hear you finally got one without error, hope it stays this way.

---

Back on the topic of computing errors... I have been watching my laptop (a 1.7Ghz Pentium M "Dothan") crunch the latest work units. It seems that all of the random "test" units are ok. However it is failing on the "random" units, without fail.
I upgraded from 4.19 to 4.45 and now am on 5.2.5. All 3 applications have consistently failed on the random results.

At home I have an AMD Athlon, running on one of TMR's optimized Boinc versions, and it has completed all of the random units without any problems.

-edit- Forgot to mention that both machines are 100% on Rosetta.

---

While I have seen some sucess on the dual G4 after upgrading to 5.3.5 BOINC, the system is still not really stable. It still spits out the occasional error and just tonight I saw the system dump about 30 R@H WUs along with a running CP@H model, a bunch of P@H WUs and a few SETIs all at once. It was running 2 R@H WUs at the time it went south. I was able to reset the projects and it went back to work, but there is clearly something very wrong with the G4 dual version and BOINC 5.2.5 is not a complete fix.

I have performed a lot of Maint checks on the system and it commes up clean. Everything was ok until the app upgrade and server upgrade a little over a week ago. So while a number of folks have suggested tweeks on the client side, it seems to me that if it was ok, then someone changes the app and the server software, and it starts crashing that it is not the client that needs fixing.

Regards
Phil

---

We Must look for intelligent life on other planets as,
it is becoming increasingly apparent we will not find any on our own.