Primenet and GMP-ECM

[ET_]

Is there a way to use the manual reservation form to add GMP-ECM results to one's stats?

[Prime95]

Yes. You have to reformat the results to look like it came from prime95. The manual results web page will then complain that the checksum isn't right. However, I can add you to the trusted users list and have that web page accept your results anyway.

[heliosh]

I would like to be added too if possible. I was going to send you 28000+ curves from S2 runs in about 2 weeks, but if I could submit them via the manual result form, that would be great too.


Some questions arised regarding the result format. An example in an old results.txt file I found was:

UID: heliosh/crou, M1493 completed 150 ECM curves, B1=110000000, B2=11000000000, Wg8: xxxxxx, AID:xxxxxx

• What would I have to put in for Wg8 and AID?
• Do I only need a single result line for thousands of curves on the same exponent?
• Do I have to put in the actual B2 size (chosen by gmp-ecm) or the mprime default B2 size (B1*100)?

Thanks

[ET_]

Originally Posted by Prime95

Yes. You have to reformat the results to look like it came from prime95. The manual results web page will then complain that the checksum isn't right. However, I can add you to the trusted users list and have that web page accept your results anyway.

I would love it, thanks. That would allow my 6 Raspberry PI minions to help GIMPS

[Prime95]

Originally Posted by heliosh

Some questions arised regarding the result format.

UID: heliosh/crou, M1493 completed 150 ECM curves, B1=110000000, B2=11000000000, Wg8: xxxxxx, AID:xxxxxx

• What would I have to put in for Wg8 and AID?
• Do I only need a single result line for thousands of curves on the same exponent?
• Do I have to put in the actual B2 size (chosen by gmp-ecm) or the mprime default B2 size (B1*100)?

For Wg8 put 00000000
Leave out the ", AID: xxx..."
A single line for the thousands of curves
Use the B2 size chosen by GMP-ECM

[heliosh]

I've tried it with a small batch of 500 curves and it seems to work, thanks a lot.


However, the results are now twice in the database, once for S1 (mprime) and once for the manually submitted S2 (gmp-ecm):
(2^1259-1)

2018-02-18    Hexaflexagon    NF-ECM    250 curves, B1=110000000, B2=110000000
2018-02-18    Hexaflexagon    NF-ECM    250 curves, B1=110000000, B2=110000000
2018-03-09    Hexaflexagon    NF-ECM    500 curves, B1=110000000, B2=794612869066

Also I got an unrealistic amount of GHz-days credited (452.3077) while in reality it was only about 11.2 GHz-days for S2.

[ET_]

My test submission did not appear. Neither on my account nor in the recent list.

[Prime95]

In the future, don't submit the stage 1 only results produced by mprime.

GMP-ECM stage 2 bounds will produce outlandish GHz-days. You are credited as if prime95 did that huge stage 2 bound.

[heliosh]

Oh ok. I was using manual assignments from the webpage in order to keep track of the progress. I guess they can't be prevented from being submitted.

In future I will let mprime generate the jobs.

[kriesel]

Originally Posted by Prime95

Yes. You have to reformat the results to look like it came from prime95. The manual results web page will then complain that the checksum isn't right. However, I can add you to the trusted users list and have that web page accept your results anyway.

Can something like this be done for gmp-ecm P-1 results also? I have some gpu CUDAPm1 partial results I'm contemplating finishing in gmp-ecm, out of necessity, since CUDAPm1 can do stage 1 P-1 to higher levels than it can perform stage 2 for. Limits depend on stage and gpu model and are not documented.