Any way to pipe GMP mpz_t into PFGW?

[hansl]

Hi, I've been toying with some C/C++ code on linux that uses GMP library to generate sequences of various large numbers (on the order of ~10kbit each).

I already have written some small utilities which are meant to be piped together, which filter out via trial division for example, and just read and write raw mpz data. The code for a filter utility looks more or less like this:

int main() {
  mpz_t x; mpz_init(x);
  while (mpz_inp_raw(x, stdin)) {
    if (SatisfiesCondition(x)) mpz_out_raw(stdout, x);
  }
  return 0;
}

I just learned about pfgw and it sounds like it is much more efficient than GMP for these large numbers. So I'm wondering if there is some way I can pipe this raw data into or out of PFGW in a similar way for testing PRP?

I would prefer to not have to convert to intermediate decimal ASCII string if possible.

[paulunderwood]

Perhaps you should look into GWNUM programming. It is the library behind PFGW and LLR. You can include some GWNUM libraries to program with. Some things are quicker with GMP like "jacobi". You can find the GWNUM library with the source of Prime95/mprime.

[a1call]

You can write your output numbers to a text file and input that file to PFGW. Look in the PFGW folder for ABC file format specs.

[rogue]

You should also look into using a pfgw script file. It can read inputs, do math with them (or manipulate in other ways), then do factoring and PRP testing.