Computation error???
12 Dec 2008 15:49:17 UTC
Topic 194081
(moderation:
Since last week i have a strange problem. Last week wendsday at one moment all wu's finished with computation error. The same happens wendsday this week, between every thing is working fine. Made no changes on the system, no hardware no software, also the system is running overclocked to 3 since month and i have no problems at seti.
Yesterday and right now the same, all wu's gone to computation error. But this time i see something i did not understand.
One wu was finished and uploadet ...ok
Second just finished and was trying to upload....upload error, at this moment all other wu's go to computation error, missing outputfile ??? is this normal? And now the funning thing, seconds later the upload was finished. I have no idea where to search the problem.
Thanks for any idea, all are welcome, for now i go to no new work.
System q6600@3
Boinc 6.2.19
Einstein 6.05 or 6.09, trying both versions
Computation error???
)
You're getting several different types of errors the first ones seem to be 119 MD5 check failed, which could be something to do with the app, or you may have to do a project reset becasue it could be that your public key got corrupted.
Could also be the OC and you may want to consider bringing it down to stock, run some more and see if you get the error.
Also one of the other errors I saw could be becuase in your Boinc Manager settings under processor usage you may have the cpu time set to something other than 100%
Thank's so far. I also
)
Thank's so far.
I also thought about a project reset, but i want to wait with it to have the chance to solve a problem with the app IF there is one. I really want to find out what the problem is, because it happens with a different bunch of wu's 618.30 and 618.35. i think.
Why i get all the different error masseges within seconds and not always the same.
Why i have no errors on seti with the OC.
fyi, my cpu time is at 100%
Thank's to everyone
RE: Also one of the other
)
People are always saying this causes errors, but I use it from time to time (I set it to 80%) and I've never even gotten one error because of that =/
RE: RE: Also one of the
)
It doesn't cause errors with all the projects. I don't know how Einstein handles it, but ABC@Home is finicky with it. What the 80% time does is start and stop applications very quickly, I believe every second. So it runs the application for 0.8 seconds, then stops it for 0.2 seconds. If it's not in a second, then it some other unit of time, but that principle. If a project's application doesn't checkpoint often enough or at all, it can cause errors.
RE: People are always
)
It does cause frequent restarts of tasks with some machines (single and multi core, though often stated otherwise), at Einstein too.
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
Returning to crazyrabbits
)
Returning to crazyrabbits problem ....
The majority of errors are related to failing MD5 checksum checks, which suggests a problem with your PC or maybe with your internet connection. If there is a weekly pattern in the occurance of the problem, it might be worth looking at antivirus scan or other periodic stuff that might interfere with BOINC.
CU
Bikeman
Thank's just checked the
)
Thank's
just checked the disk- look's ok
made a defrag
no virus found
no periodic task. only a daily update of the virussoftware, but yesterday it happens not at the same time with the computation error.
just reseted the project and deleated all file in the einsteinfolder.
Downloaded new work and wait what happens.
Is there any tool to check the internetconnection? I realized no problems so far.
RE: ... at one moment all
)
I have six Q6600s, all overclocked to near 3.0GHz and all with the stock heat sink and fan. The combination of overclocking and 100% CPU load creates a lot of heat and it's not surprising that problems could arise, particularly if there is a relatively high ambient temperature - perhaps around 35-40C in my case. I have seen pretty close to exactly the same sort of behaviour that you describe a couple of times previously. I'm pretty confident this was due to an overclock that was too ambitious for the stock cooler to cope with.
I cured this by backing off a little on the overclock. I probably could have also cured it by using a better cooler and/or by upping the Vcore until full stability was achieved. The easiest option for me was to lower the frequency so I chose that. The drop in performance was quite small.
It's a very long story but I was able to recover virtually the full cache of tasks (on several occasions) that had been trashed without having to reset the project or replace any data or program files that had previously been downloaded. Even though there were supposed to be files with bad MD5 checksums, I never found any evidence of file corruption. I believe that problems were possibly being caused by overclock induced failures in zipping/unzipping operations.
Cheers,
Gary.
Thank's again, just
)
Thank's again,
just finished the first new wu, no problems. next one in half an hour, just see.
heat should not be the problem, cpu 40c (107f) peak 43C (109f), cores 60c (140f) peak 65c (148f), seen higher values during summer.
Problems with the OCing after month without problems, hmmm, then something is going down.
If the problem is coming up next time i try to step down the oc a little bit. or raise the voltage on the memory a little bit, working since the beginning with undervoltage, who knows.
RE: Thank's again, just
)
Is this happening on only one pc? Is so try going back to the standard, no overclocking, and see if it still happens. If not then you have your answer, if it does keep happening, then you have an answer too. Oc'ing can be a very touchy thing, one month it works fine, the next everything is down because this or that part was not capable of handling it long term.