Computation error
13 Feb 2008 2:20:47 UTC
Topic 193510
(moderation:
Every Einstein file that runs on this computer is now producing a "computation error", and each one happens very early in the processing - around 3 to 6 percent complete.
Reading some of the posts led me to reset the project, and have done that twice, however the problem persists. Einstein is running on BOINC in conjunction with SETI@HOME, and has been running fine for the past few months. SETI still runs perfectly. This is on a Core 2 Duo Intel processor with 2G of RAM.
What am I missing that might be causing this?
Thanks in advance
Computation error
)
Not fully true. Your results list shows a few tasks which have completed successfully and a few that have gone quite a way towards completion. By looking at the stderr.out for each failed task, there appear to be a number of 'different' errors being reported, unhandled exceptions, signal 8s, etc.
Bernd will perhaps be able to glean something from these so thanks for the report. My guess is that you may have some sort of intermittent hardware issue - hot CPU, flakey RAM, unstable power, that sort of thing. Another possibility is overclocking. Is your machine overclocked at all? If it is you might try backing it off just a little.
Cheers,
Gary.
The CPU is not overclocked at
)
The CPU is not overclocked at all. The system is completely stock. It's odd, though, that Einstein is the only program that shows any problem. SETI is running at the same time with no errors, and I use some pretty sophisticated amateur radio programs that work fine. (I turn BOINC off when I run those).
Could I be asking too much of the CPU? I have it set for using both processors, and using 95% of the CPU time.
Thanks!
Doc
RE: By looking at the
)
I'm looking into the - exit code -1073741678 (0xc0000092) which is a BOINC error. I know it's EXCEPTION_FLT_STACK_CHECK, but that's all I got. [url=http://msdn2.microsoft.com/en-us/library/ms679356(VS.85).aspx]MSDN2[/url] says "The stack has overflowed or underflowed, because of a floating point operation." which is as much help as if it were in Chinese.
It could signal a problem with the FPU, but I'll report back on this.
RE: The CPU is not
)
OK, that rules overclocking out.
Not really that odd! The different projects do stress the hardware in different ways and I've seen examples before where only one application was failing. Also, program bugs tend to affect a number of different computers with the same type of error. Hardware problems tend to have different types of errors on the same hardware system. This latter description seems to fit better with what your results list is showing.
I wouldn't have thought so. I have many computers running at 100% CPU utilisation 24/7. a large number are overclocked and all run at quite high ambient temperatures eg 35 degrees C room temperature. I don't have a single example where a CPU has failed, even after 3-4 years of continuous running. I have plenty of examples of other failures that cause tasks to crash - things like faulty RAM, flakey power, fan failures leading to overheating, swelling caps on motherboards, etc. Every time when a machine has crashed due to hardware problems like these, the CPU itself has been fine.
Try setting your CPU to 100% and see what happens.
Cheers,
Gary.
RE: I'll try that when I
)
Which OS are you
)
Which OS are you running?
If Vista then if you are shutting down the comp at anytime (i.e. overnight) then
then Vista shuts down too fast for Boinc and especially Einstein.
Warning! This post contains atrocious spelling, and terrible grammar. Approach with extreme edginess.
It's not the speed, but the quality - Until I get a faster computer
I'm running XP Pro here, and
)
I'm running XP Pro here, and the computer rarely gets shut down, although it usually gets reset once a day or so. Normal maintenance is run once a week (virus scan, spyware scan, defrag).
I put the preference up to 100% cpu time a few minutes ago - I had set it at 75% this morning. the SETI file it was running immediately started showing the progress in reverse. The Einstein file continued to run, although quite slowly.
From the statistics tab, it looks like I haven't completed an Einstein unit for 4 days in a row, while SETI is smoking along, although I have 3 other computers running SETI, and not Einstein.
Very interested to see what you might see in my error reports!
Thanks
Doc
RE: Before I left, I
)
Well, this points clearly to a hardware problem: cooling, power supply , mother board ....
Bikeman
RE: - exit code -1073741678
)
This error points to a problem with the CPU or motherboard.