Generating invalid result using linux client
12 Mar 2005 13:03:48 UTC
Topic 188482
(moderation:
Hi there,
i thought its a good idea to spent my computer resources for scientific computations. But it seems that my client generates only invalid results.
From a scientific standpoint can u say that the clients for different operating systems generate identical results for identical wu's?
If this is not the case, there seems no use for the hole thing.
For now i stopped all boinc computations...
Generating invalid result using linux client
)
> Hi there,
>
> i thought its a good idea to spent my computer resources for scientific
> computations. But it seems that my client generates only invalid results.
>
> From a scientific standpoint can u say that the clients for different
> operating systems generate identical results for identical wu's?
There are certainly differences in the way different OSs do the math, but they should be generating results that are close enough to be considered a match. The validator has a problem with this though. See this thread for more info on that.
However, in this thread, Bruce seems to be saying that as of March 7 the problem was fixed. That certainly does not appear to be the case, as I have several since that time that are still identical to the other linux result, but the results are still being marked "invalid" for whichever OS returned its results last.
Darren
>However, in this thread,
)
>However, in this thread, Bruce seems to be saying that as of March 7 the
>problem was fixed. That certainly does not appear to be the case, as I have
>several since that time that are still identical to the other linux result,
>but the results are still being marked "invalid" for whichever OS returned
>its results last.
>
>Darren
Sometimes even when a problem is fixed it only applies to work generated after the fix. However in this case since it is a fix to the validator that seems unlikely.
BOINC WIKI
BOINCing since 2002/12/8