Not quite the same thing, but my personal high, in 4001 exponents tested, is
FATAL ERROR: SUM(INPUTS) != SUM(OUTPUTS), 4.637189956e+013 !=
1.744531645e+022
Roundoff is normally bounded by 0.5 even when the
calculation of residue goes hugely awry.
Ken
At 04:47 PM 1999/09/30 +0200, Steinar Gunderson wrote:
>After overclocking my poor 400 MHz PII to 496 MHz (and upping
>the voltage from 2.00V to 2.30V), I figured out I should perhaps
>do a torture test for a few days, before throwing it at my real
>exponents :-) After 10 seconds, I got:
>
>Test 1 of 15, 400 Lucas-Lehmer iterations of M19922945 using 1024K FFT
length.
>ERROR: ROUND OFF (5.283699122e+13) > 0.40
>Possible hardware failure, consult the readme file.
>
>Has anybody experienced any _bigger_ roundoff errors than this?
_________________________________________________________________
Unsubscribe & list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
