On Tuesday 17 January 2006 21:05, Paul Karagianis wrote: > On Tue, 17 Jan 2006 17:12:03 "Alasdair Tompson" said: > >I saw this in the output screen of one of my clients and wondered what > >it meant: > > > >Iteration: 18835326/29642771,ERROR:ROUND OFF (0.40625) > 0.40 > >Continuing from last save file. > >Resuming primality test of M29642771 at iteration 18830465 [62.52%] > >Disregard last error. result is reproducible and thus is not a hardware > >problem. > > Speaking just from my own personal experience, this *can* be a hardware > problem. I've had cases where de-tuning my system and rolling back to > a save file that was several hours old gave correct results, that is, > from my point of view the problem was not reproducible, it appeared to > be the result of overly agressive memory settings and/or overclocking.
There are two variants of the "round off" error message - one "result is reproducible", one which the rerun doesn't match. Paul, please check that you were getting _reproducible_ errors on a failing system (you should only have to trawl your results.txt file) - if you were then maybe we need to re-evaluate the logic we're using. Regards Brian Beesley _______________________________________________ Prime mailing list [email protected] http://hogranch.com/mailman/listinfo/prime
