Hi guys,

I've been assisting in quality assurance testing of Prime95 for some 
time and have had one of my systems run halfway through a LL test on 
a 10 million digit number. Unfortunately the intermediate result I 
have does not match the only other available value, which means that 
one of us is wrong.

The problem could be a program bug (unlikely), an undiscovered bug in 
either Athlon or Pentium III floating-point arithmetic (again 
unlikely) or a random hardware glitch in either my system or the 
system against which I'm checking (much the most likely).

To eliminate the first two possibilities it is neccessary to re-run 
the first half of the test. As I have savefiles at 1 million 
iteration intervals this could be done in parallel by 17 systems; I 
will be running (at least) one of these myself but would like to 
recruit 16 volunteers to check the remaining intervals.

Volunteers should be running Prime95 or mprime v20.6 on an Intel 
Pentium II or PIII system. Please do not volunteer if you intend to 
run on an Athlon since this would make it impossible to determine if 
a processor bug is to blame. You will need to upload the save file, a 
binary file approximately 5 megabytes in size, from my ftp site, and 
unzip it; your segment of the work will take 13 CPU days (on a 
PII-400, according to George Woltman's benchmarks) with checkpoints 
at 1.3 day intervals. You should be prepared to check each checkpoint 
and report to me immediately should you find a discrepancy. I would 
like to finish the work as soon as reasonably possible, so reasonably 
fast systems are indicated - at least PII-400 running continuously.

Unfortunately I cannot offer any recognition other than my personal 
thanks. (i.e. no CPU time credit, sorry!)

If you're interested, please reply to me and I'll supply further 
instructions (how to download your specific file, precise details of 
how to set up the run and a list of my intermediate residuals at your 
checkpoints to compare with your values).


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

Reply via email to