Maximum B1 for P+1 = 4294967295?

[Andi47]

What is the reason why max. B1 for p+1 = 4294967295 ?

[xilman]

Originally Posted by Andi47

What is the reason why max. B1 for p+1 = 4294967295 ?

32-bit machine. Go figure.


Paul

[Andi47]

Originally Posted by xilman

32-bit machine. Go figure.


Paul

2^32. Sorry for asking, I was just surprised to see this limit because with the same version of GMP-ECM (6.1.2) it is possible to do p minus 1 with B1 far bigger than 2^32 on the same machine. (I already did 150e9)

[akruppa]

The B1 value is stored in a variable of type double, so that would allow for values bigger than 2^32-1. However, the PRAC algorithm we use in P+1 and ECM uses integers.

Alex

[Andi47]

Originally Posted by akruppa

The B1 value is stored in a variable of type double, so that would allow for values bigger than 2^32-1. However, the PRAC algorithm we use in P+1 and ECM uses integers.

Alex

So it is also not possible to run ECM with B1 > M32?

Currently running P+1 on M1061 with B1 = M32, B2 = 1e14 (I guess it is also possible with p+1 to run stage 2 in steps of B2 = 1e13, B2=1e13-2e13, B2=2e13-3e13 etc. - am I right?)

[akruppa]

Yes, it's possible. I'd recommend against running P-1/P+1/ECM on M1061, though. On one hand, it is almost certainly wasted effort, on the other hand, if it isn't, someone somewhere might scream like a banshee.

Alex

[Andi47]

Originally Posted by akruppa

Yes, it's possible. I'd recommend against running P-1/P+1/ECM on M1061, though. On one hand, it is almost certainly wasted effort, on the other hand, if it isn't, someone somewhere might scream like a banshee.

Alex

OK. What is the next Mersenne number which no known factor and how much ECM/P-1/P+1 has been done on it?

[Prime95]

See http://mersenne.org/ecm1.htm for ECM info on small Mersennes. I think they've been P-1'ed to 4 billion.

[Andi47]

Originally Posted by Prime95

See http://mersenne.org/ecm1.htm for ECM info on small Mersennes. I think they've been P-1'ed to 4 billion.

To whom should I report finished ecm curves on these numbers? The linked page does not tell this.

[Prime95]

Originally Posted by Andi47

To whom should I report finished ecm curves on these numbers? The linked page does not tell this.

[Joe O]

Originally Posted by akruppa

The B1 value is stored in a variable of type double, so that would allow for values bigger than 2^32-1. However, the PRAC algorithm we use in P+1 and ECM uses integers.

Alex

So how about using a 64 bit integer variable to store the B1 value?