On 03-Nov-2011 12:53 PM, Israel Figueroa wrote:
> 1) When the client it's a multicore (core2duo, QuadCore or GPU/CPU), how > the p_iops and p_fops information should be interpreted? as "per core" > (asuming that the total host it's ncpus * p_iops ) or the number of cores > it's integrated in that infomation. This is for statistical porpouses, I > know the benchmarks just gives you a raw estimation. They're both per CPU. > > 2) Who decides when to regenerate the internal CPID. Does the client > somehow decides to do so by itself? or it's an instrucction sent to the > client? The client generates a new CPID when there's evidence that its state file was copied from another host. > > and, in an unrelated topic: > > It seems that the client has some issue with the BAM<signing_key> field. > Doing some test, the clients failed to recognise the key because it has > some extra tabs, or "\n" at the end (because usual way to send it it's to > wrap the key into several lines). It seems that it does a strcmp before > stripping the "whitespaces" chars, and if they mismatch then rejects the > key even if it's essentially "valid". This is probably fixed in the client currently being tested (6.13.10). -- David _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
