How to make GMP-ECM skip stage 1 when feeding the prime95 results

[Ensigm]

I am basically following the method in this post, but it seems that GMP-ECM still runs stage 1 for a long time, which is already done by prime95, before doing stage 2 which it is good at. Does anyone know where the problem could be?


An example of the output:

GMP-ECM 7.0.4 [configured with GMP 6.2.0, --enable-asm-redc, --enable-openmp] [ECM]
Resume warning, skipping line with no '=' after: [Sun Aug 23 21:
Resuming ECM residue saved with Prime95 
Input number is 0x249C4CFAABA959C8705B51C84281F03253A8524BE39CFD8FDAA914BBF71E457C8C9F3417ABCF9C23E7F708B91D373F46A5EC530FB55C95FF69C98E03678FCAD94860EC429B1FBEB76854487C4BB6C655EDD9B94B38021E11A3BBDEC6F4EA30CA54DF94CD321F61 (248 digits)
Using B1=44000000, B2=4400000000, polynomial Dickson(6), sigma=0:2856607019773923
Step 1 took 170265ms
Step 2 took 3675ms

Also, how do I report the number of curves done using this hybrid method, when no factor has been found?

[VBCurtis]

Try setting B1 of 1, and let B2 be chosen by GMP-ECM. It's pretty pointless to use GMP-ECM for stage 2 while restricting B2 to 100* B1.

[Ensigm]

Originally Posted by VBCurtis

Try setting B1 of 1, and let B2 be chosen by GMP-ECM. It's pretty pointless to use GMP-ECM for stage 2 while restricting B2 to 100* B1.

How to set B1=1 when resuming from the Stage 1 save file? I tried

ecm -resume ecm_input.txt 1 > ecm_output.txt

but it just set B2=12. And I don't think stage 1 is skipped even if I let gmp-ecm automatically determine B2.

[Ensigm]

It seems that my version of gmp-ecm can make use of the savefile produced by itself, but not that produced by prime95. When it reads the prime95 savefile, it is able to recognize N and SIGMA, but seems to ignore the fact that stage 1 was already done.

The file produced by prime95 has 3 fields: N, QX, and SIGMA;
The file produced by gmp-ecm has: METHOD=ECM, PARAM=0, SIGMA, B1, N, X, CHECKSUM, PROGRAM, Y=0x0, X0=0x0, Y0=0x0, WHO, and TIME.

[axn]

 ecm -resume ecm_input.txt 44e6-44e6 > ecm_output.txt

Ask gmp-ecm to do stage 1 from 44e6 to 44e6 (or whatever the B1 limit was for P95 stage1 run), and it will correctly figure out the B2 and do it.

[Ensigm]

Originally Posted by axn

 ecm -resume ecm_input.txt 44e6-44e6 > ecm_output.txt

Ask gmp-ecm to do stage 1 from 44e6 to 44e6 (or whatever the B1 limit was for P95 stage1 run), and it will correctly figure out the B2 and do it.

Thank you, this works.

[Ensigm]

Now the second question: how do I report number of curves done using this method, if no factor has been found?

[Prime95]

Originally Posted by Ensigm

Now the second question: how do I report number of curves done using this method, if no factor has been found?

Email me with the user id for CPU credit, exponent, curve counts and bounds

[Ensigm]

Originally Posted by Prime95

Email me with the user id for CPU credit, exponent, curve counts and bounds

Thank you, but if there's not an automatic way then I'd rather not bother and waste the time of both of us. I don't have that much computing power to be worth doing it manually. Unless if I submit it like once half a year.

[Prime95]

Yes, the few that do this kind of work, run 1000 curves or more and then email me.