Can't get manual page to parse a result

[Siegmund]

I've double-checked a batch of Mersenne candidates with the LLR program. (Yes, LLR gives a warning that prime95 gives better performance - and I wanted to find out how much better. It turns out they are very close: I can get 27 GHz-d/d running P95 with 4 threads, vs. about 24 GHz-d/d running three instances of LLR and 26 running four.)

Anyway, I now have results from LLR which are ready to turn in. If I just paste

1*2^nnnnnnnn-1 is not prime. LLR Res64: nnnnnnnnnnnnB2BB Time : nnnnnn.061 sec.

into the Manual Results window, I get a "line not understood error."
The obvious fix, change "1*2^nnnnnnnn-1" to "Mnnnnnnnn", did not work either.

What does Primenet want to see, besides "Mnnnnnnnn is not prime" and a "Res64: xxxx" residue?

I assume this is a rare enough circumstance that it doesn't make sense to ask someone to teach the parsing script how to read LLR output. I just need a tip how to format it so that primenet will take it.

[Prime95]

I doubt LLR residues are compatible with prime95 residues.

[Siegmund]

I doubt LLR residues are compatible with prime95 residues.

I can confirm that the unmasked bits from the first-time test shown on the exponent status page do match the Res64 I got from LLR.

I assume that was by design since he reused good chunk of p95 source when creating his program.

[Prime95]

Interesting. I did not know that LLR used the Lucas-Lehmer primality test.

[Madpoo]

Originally Posted by Siegmund

I've double-checked a batch of Mersenne candidates with the LLR program. (Yes, LLR gives a warning that prime95 gives better performance - and I wanted to find out how much better. It turns out they are very close: I can get 27 GHz-d/d running P95 with 4 threads, vs. about 24 GHz-d/d running three instances of LLR and 26 running four.)

Anyway, I now have results from LLR which are ready to turn in. If I just paste

1*2^nnnnnnnn-1 is not prime. LLR Res64: nnnnnnnnnnnnB2BB Time : nnnnnn.061 sec.

into the Manual Results window, I get a "line not understood error."
The obvious fix, change "1*2^nnnnnnnn-1" to "Mnnnnnnnn", did not work either.

What does Primenet want to see, besides "Mnnnnnnnn is not prime" and a "Res64: xxxx" residue?

I assume this is a rare enough circumstance that it doesn't make sense to ask someone to teach the parsing script how to read LLR output. I just need a tip how to format it so that primenet will take it.

See what George said...

In that one case, if you want to PM myself or George with the exponent and residue, we can check and see if it's actually a match of the first time check done by Prime95. Best case, it's a match. Worst case, it doesn't match and we'll be left to wonder if that first test was bad or if the residues are simply calculated differently.

[Siegmund]

I did not know that LLR used the Lucas-Lehmer primality test

As I understand it, Lucas-Lehmer-Riesel testing is the same iteration as Lucas-Lehmer, except that to test k*2^n-1, you start from somewhere other than 4 (each k has its own starting-place).
But as a statistician, my understanding may only be approximately correct :D

[Siegmund]

George has supplied me a format via PM that enables these to be submitted through the mersenne.org manual results form. Thank you!