Jeff Woods wrote:
2^2147483647-1 = Next Mersenne prime ?
Using a single PC it would take me about 600 years to check this. Using the GIMPS project it should be feasible.
What do you think about that?
Unfortunately, it will still takes GIMPS about 600 years (a number I didn't check) to complete this check. You can't divide up a test of an exponent among many computers. The way the test works, it has to be done by a single computer. The results are linear -- each interim result depends on the last.
[cut]
Hence, only one machine can do it.
Well, the quote from <http://www.mersenne.org/>:
'It took more than 50 days of calculations on Dr. Nowak's 2.4 GHz Pentium 4 computer. The new prime was independently verified in 5 days by Tony Reix of Grenoble, France using a 16 Itanium CPU Bull NovaScale 5000 HPC running the Glucas program by Guillermo Ballester Valor of Granada, Spain. A second verification was completed by Jeff Gilchrist of Elytra Enterprises Inc. in Ottawa, Canada using 15 days of time on 12 CPUs of a Compaq Alpha GS160 1.2 GHz CPU server at SHARCNET.'
As we can see, multiprocessor machines seems to be faster than ones with one CPU (at least if they use Glucas program, because AFAIK mprime is single threaded). I don't know how Glucas exactly works (is there huge amount of data transferred between processors in short time or not), but there might be a way to split it somehow with reasonable improvement.
-- No sig today. _______________________________________________ Prime mailing list [email protected] http://hogranch.com/mailman/listinfo/prime
