Mersenne DigestMonday, October 30 2000Volume 01 : Number 788
--
Date: Sat, 28 Oct 2000 09:45:20 -
From: "Brian J. Beesley" [EMAIL PROTECTED]
Subject: Re: RE: Mersenne: Volunteers wanted
Hi guys,
I now have complete results of this project.
Altogether there were 17 segments of one million iterations each. All
but one of these matched the original run.
There was a mismatch between iterations 8.3 million and 8.4 million.
Rerunning this segment on the original system resulted in an interim
residual which matched the value reported by the user who ran this
segment.
Conclusions:
(a) My system "glitched" once during the original run. Quite honestly
this surprised me as the system is not significantly overclocked, is
well cooled, is fitted with ECC memory and has no history of single-
bit memory errors corrected by ECC. Nor has it ever crashed, hung up
or spontaneously rebooted. Therefore I feel it's more than likely
that the glitch was memory corruption caused by a software problem.
This system is running Windows 2000 Professional ...
(b) There is no evidence of any systematic difference between Athlon
and Intel processor families which might affect the GIMPS/PrimeNet
project.
Thanks again to the participants!
Regards
Brian Beesley
_
Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt
--
Date: Sun, 29 Oct 2000 13:59:37 -0800
From: "Ethan Hansen" [EMAIL PROTECTED]
Subject: RE: RE: Mersenne: Volunteers wanted
A possible explanation is a soft error in one of the processors internal
caches. These can be caused either by cosmic rays, or alpha decay
within the lead solder bumps. As processors incorporate more internal
cache, not only on die L1, but the 32 - 128K of other instruction
caches, the possibility of these errors increases. A study using 2MB
cache Xeon processors detected no errors at sea level equvalent cosmic
ray flux, some problems at 1 mile altitude, and serious ( 1 fault per
month) at 4 miles. Moral of the story: don't bring your laptop along
while climbing Mt Everest.
Regards,
Ethan
- -Original Message-
[SNIP]
Conclusions:
(a) My system "glitched" once during the original run. Quite honestly
this surprised me as the system is not significantly overclocked, is
well cooled, is fitted with ECC memory and has no history of single-
bit memory errors corrected by ECC. Nor has it ever crashed, hung up
or spontaneously rebooted. Therefore I feel it's more than likely
that the glitch was memory corruption caused by a software problem.
This system is running Windows 2000 Professional ...
(b) There is no evidence of any systematic difference between Athlon
and Intel processor families which might affect the GIMPS/PrimeNet
project.
Thanks again to the participants!
Regards
Brian Beesley
_
Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt
--
Date: Mon, 30 Oct 2000 10:26:27 +0100
From: Henk Stokhorst [EMAIL PROTECTED]
Subject: Mersenne: bug in Wordpad?
L.S.,
Whenever I have a large file (about 3.000 lines) with exponents to test
(extracted from the nofactor.cmp file) and replace all the 'Test=' with
'Factor=' using wordpad (part of windows accesoires) one line gets to
read 'Factorst='.
Can other people reproduce this result? What OS would you recommend for
such operations ;-)
YotN,
Henk Stokhorst
_
Unsubscribe list info -- http://www.scruz.net/~luke/signup.htm
Mersenne Prime FAQ -- http://www.exu.ilstu.edu/mersenne/faq-mers.txt
--
Date: Mon, 30 Oct 2000 11:13:05 -
From: "Andy Hedges" [EMAIL PROTECTED]
Subject: FW: Mersenne: bug in Wordpad?
Henk Stokhorst,
It is not really the OS (although UNIX does have a lot of nice tools for
this type of thing) it is wordpad that is not really the most capable text
editor.
If you are using Windows I would suggest you try TextPad which is available
for 30 day evaluation from www.textpad.com.
Hope this helps,
Andy
~~
~~
Andy Hedges
Consultant
Proteus the Internet Consultancy
Tel: + 44 (0)207 689
Fax: + 44 (0)207 689 6667
Direct: + 44 (0)207 689 6669
Web: http://www.proteus.co.uk
Email: [EMAIL PROTECTED]
26 Britton St, London EC1M 5UB
A Division of Proteus Ventures Ltd
Registered in England: 3169362
VAT: 675007732
~~
~~
- -Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]]On