On Sunday 10 December 2006 13:48, david eddy wrote: > > rH = 17 . You are right! How did you do that?
Long long ago the save file name for exponent nnnnnnn was pnnnnnnn (exponent left padded with zeros if necessary). This scheme failed at exponent 10,000,000. So the second character became A for 10 million, B for 11 million etc. Something else happens after Z for 36 million.... This is all bound up with some constraints which have become partially or completely obsolete, like 8 char length limits, case insensitivity etc. > > > What should happen is that, when rXnnnnnn has been written properly, > > qXnnnnnn is deleted, pXnnnnnn is renamed to qXnnnnnn and rXnnnnnn is > > renamed to pXnnnnnn. Then when the program is restarted it will not have > > wasted any time by recovering from the last save file. If writing > > rXnnnnnn fails then some time will have been wasted but there is no > > reason to suspect the integrity of the run. > > If an error occured writng to rHXXXXXX, why didn't it go back to an older > pH/qH ? The active save file is the "p" save file, "q" is the backup. "r" is used only temporarily whilst the file is being written (and is therefore bound to be unusable for restarting after suspension, as writing it failed). > > > If this is happening often, I would suggest exiting the program manually > > before initiating reboot. > > I think I did this. You'll still get the "error writing savefile" if the target disk/partition becomes full, your quota gets exceeded (if quotas are enabled), probably if there is a media fault somewhere in the range selected for writing the savefile (though this should probably show up in other ways not connected with prime95/mprime) ... Regards Brian Beesley _______________________________________________ Prime mailing list [email protected] http://hogranch.com/mailman/listinfo/prime
