any way to tell why I came up invalid?
17 Oct 2014 13:47:31 UTC
Topic 197754
(moderation:
It doesn't happen to me very often, but I'm still curious.
http://einsteinathome.org/workunit/200750830
David
Miserable old git
Patiently waiting for the asteroid with my name on it.
any way to tell why I came up invalid?
)
My guess is that the two AMD CPU's produced a slightly different result than your Intel CPU. Maybe your result would verify against another Intel CPU.
Stuff like this makes me worry about the accuracy of the science done here :)
Do you have your virus
)
Do you have your virus software scanning the ProgramData\BOINC folder? This can sometimes interrupt the writing of important files as the files could be locked by the virus software during the scan. As you state it is rare and usually virus software is set to scan at different intervals (daily, weekly, etc). So on the rare occasion the virus software has a file locked and the Einstein program needs to do a write it may cause just enough of a blip that the signatures do not match when the final counts are sent back to the servers.
It's always been a recommendation to have the virus software skip that directory for this type of reason and others. I would try that first and see if your issue goes away.
Otherwise it could be you have a memory chip with one bad register. This register may not always be utilized, depending on what else may be running through memory. That randomization is very hard to detect also. Even RAM testing software may not catch it.
So, just start with the virus software if you have not yet.
73s
At the time, I was only
)
At the time, I was only running MSE (I have since installed a different AV). I'm not sure if it scans that directory or not, but I'd think if that was the problem it would be more frequent.
David
Miserable old git
Patiently waiting for the asteroid with my name on it.
I looked over the task and
)
I looked over the task and this may or may not be relevant
Shortly after
% checkpoint 219
% Sky point 220/275
it looks to my untrained eye, that the task was restarted from that checkpoint.
I see no other invalids so the sample size is small, i guess it is the same host with the invalids.