On the 6th March last Daren posted an article describing how to use one
machine to do all the p-1 factoring for other machines. The article can be
read here:
http://www.mail-archive.com/[EMAIL PROTECTED]/msg07966.html
although it is in the middle of a long thread starting here:
http://www.mail-archive.com/[EMAIL PROTECTED]/msg07962.html
I'm not planning to do that myself at the moment but I was wondering if I
could use the same technique to cause the p-1 factoring to be done earlier.
I'm thinking about machines that will only do p-1 for a few hours each
night. Suppose I had this work lined up on a machine:
Test=181xxxxx,66,1
Test=182xxxxx,66,1
Test=193xxxxx,66,1
Test=204xxxxx,66,0
That is 3 exponents that require no p-1 factoring followed by one that does.
As things stand the p-1 factoring for this exponent will not start until the
other 3 exponents have finished. However by this point this list of work may
look like:
Test=204xxxxx,66,0
Test=195xxxxx,66,0
Test=206xxxxx,66,0
ie. all the exponents need p-1 factoring. In this situation it will do the
p-1 factoring 1st stage for all 3 exponents but then it will start the LL on
the first one during the day and do 2nd stage p-1 factoring on the first
exponent during the night. However, this LL may be pointless since the p-1
factoring may end up factoring the exponent. There are several things that
you can do to avoid this potentially useless LL testing. One is to cause it
to do the p-1 factoring on the 20M exponent sooner. This can be done by
changing the original work to do list to (assuming I understand Daren's
posting correctly):
Test=181xxxxx,66,1
Pfactor=204xxxxx,66,0
Test=182xxxxx,66,1
Test=193xxxxx,66,1
Test=204xxxxx,66,1
This tells it to start the P-1 factoring as soon as the current exponent is
finished (you could put it event above the current exponent - but I would
advise stopping Prime95 while you save the work to do file and restarting it
again afterwards if you want to do this). And the bottom line has been
changed to say that the P-1 has been finished - which it should have been by
the time you get that far.
I've not tired this myself yet, but I'm going to do so soon. Thus I can't
be totally sure if there aren't any pitfalls.
Any and all comments welcome.
Doug.
_________________________________________________________________________
Unsubscribe & list info -- http://www.ndatech.com/mersenne/signup.htm
Mersenne Prime FAQ -- http://www.tasam.com/~lrwiman/FAQ-mers
