Could we have the ability to re-reserve exponents for P-1 factoring?

[ixfd64]

Some of us, myself included, have been redoing P-1 on exponents with only stage 1 done. However, PrimeNet doesn't seem to allow those assignments to be registered. This isn't a huge issue but can result in duplicated work.

I know factoring known composite exponents won't help us find a Mersenne prime. It's also debatable whether redoing P-1 on double-check candidates is worthwhile. But for the sake of convenience, could we please at least have the option to register P-1 assignments for unverified exponents?

Thanks!

[chalsall]

Originally Posted by ixfd64

But for the sake of convenience, could we please at least have the option to register P-1 assignments for unverified exponents?

You can, so long as they haven't been double checked, nor are already assigned to someone else.

This brings up a point... I've got a few of my machines working in the 40M to 49M ranges (inclusive), redoing P-1 "well" where stage two hasn't been run. I try to reserve them so people see this activity.

Separately, Wayne (Sid and Andy) has some of his machines working in the 50M to 59M ranges doing similar work. However, he doesn't appear to reserve them, so please try to avoid "stepping on toes".

[petrw1]

Originally Posted by chalsall

Separately, Wayne (Sid and Andy) has some of his machines working in the 50M to 59M ranges doing similar work. However, he doesn't appear to reserve them, so please try to avoid "stepping on toes".

He is correct. I'm using Pminus1 work type which won't register even though I'd like to.

[petrw1]

Originally Posted by petrw1

He is correct. I'm using Pminus1 work type which won't register even though I'd like to.

I want to choose my own bounds...

[kruoli]

Example for some user defined P-1:

[Main thread Feb 28 20:34] Mersenne number primality test program version 29.4
[Main thread Feb 28 20:34] Optimizing for CPU architecture: Core i3/i5/i7, L2 cache size: 256 KB, L3 cache size: 8 MB
[Comm thread Feb 28 20:34] Registering assignment: P-1 M11962133
[Comm thread Feb 28 20:34] URL: http://v5.mersenne.org/v5server/?v=0.95&px=GIMPS&t=ra&g=XXXXXXXX&c=0&w=3&A=1&b=2&n=11962133&C=-1&B1=8924365&B2=330201505&ss=6823&sh=9318B8D1AD8
[Main thread Feb 28 20:34] Starting worker.
[Comm thread Feb 28 20:34] == Info: About to connect() to v5.mersenne.org port 80
[Comm thread Feb 28 20:34] == Info:   Trying 162.212.57.131... 
[Comm thread Feb 28 20:34] == Info: connected
[Comm thread Feb 28 20:34] == Info: Connected to v5.mersenne.org (162.212.57.131) port 80
[Comm thread Feb 28 20:34] => Send header: GET /v5server/?v=0.95&px=GIMPS&t=ra&g=XXXXXXXX&c=0&w=3&A=1&b=2&n=11962133&C=-1&B1=8924365&B2=330201505&ss=6823&sh=9318B8D1AD84E36F67
[Comm thread Feb 28 20:34] Host: v5.mersenne.org
[Comm thread Feb 28 20:34] Accept: */*
[Comm thread Feb 28 20:34] 
[Comm thread Feb 28 20:34] <= Recv header: HTTP/1.1 200 OK
[Comm thread Feb 28 20:34] <= Recv header: Content-Type: text/html; charset=UTF-8
[Comm thread Feb 28 20:34] <= Recv header: Server: Microsoft-IIS/8.5
[Comm thread Feb 28 20:34] <= Recv header: X-Powered-By: PHP/7.1.10
[Comm thread Feb 28 20:34] <= Recv header: Date: Wed, 28 Feb 2018 19:34:52 GMT
[Comm thread Feb 28 20:34] <= Recv header: Content-Length: 80
[Comm thread Feb 28 20:34] <= Recv data: pnErrorResult=44
[Comm thread Feb 28 20:34] pnErrorDetail=ra: unsupported assignment work type: 3
[Comm thread Feb 28 20:34] ==END==
[Comm thread Feb 28 20:34] 
[Comm thread Feb 28 20:34] == Info: Connection #0 to host v5.mersenne.org left intact
[Comm thread Feb 28 20:34] == Info: Closing connection #0
[Comm thread Feb 28 20:34] RESPONSE:
[Comm thread Feb 28 20:34] pnErrorResult=44
[Comm thread Feb 28 20:34] pnErrorDetail=ra: unsupported assignment work type: 3
[Comm thread Feb 28 20:34] ==END==
[Comm thread Feb 28 20:34] 
[Comm thread Feb 28 20:34] PrimeNet error 44: Invalid assignment type
[Comm thread Feb 28 20:34] ra: unsupported assignment work type: 3

Especially interesting is the last line, Pminus1 is not supported at all by the server.

[ixfd64]

Yeah, I couldn't reserve any of the older exponents for P-1 factoring even after multiple tries. But at least we could still coordinate the effort on the forums to avoid duplicate work to some degree.

[Brian-E]

I believe (am I correct?) that it's more efficient to do an LL test (double check) on exponents which have only had stage 1 of the P-1 factoring, from the point of view of determining whether the number is prime or composite.

So while there's nothing to stop people from re-doing the P-1, it might not be a good idea to encourage it by making it an official assignment possibility in PrimeNet?

[ixfd64]

One workaround I've found is to manually reserve the exponents as double-checks. However, this doesn't work for Category 0 and 1 exponents.

[S485122]

Originally Posted by ixfd64

One workaround I've found is to manually reserve the exponents as double-checks. However, this doesn't work for Category 0 and 1 exponents.

Please do not meddle with category 0 or 1 exponents. The queues of the individual workers in those categories are small (less than 5 days of work.) and in category 0 the number of assignments is 200 by design. Once exponents are in one of these categories one should leave them alone or assign them following the rules, which means that if the factoring attempt does not work you must do the LL check otherwise your core will lose its category 0 or 1 status. I understand you do not want to duplicate extra P-1 work, I would ask you not to make people doing normal PrimeNet work.

Jacob

[petrw1]

Originally Posted by Brian-E

I believe (am I correct?) that it's more efficient to do an LL test (double check) on exponents which have only had stage 1 of the P-1 factoring, from the point of view of determining whether the number is prime or composite.?

I don't understand how this could be the case

[petrw1]

Toe avoidance can be achieved if we all note our P1 efforts here:

For me:
- 50-59M range
- For any .1M range that has more than 1999 unfactored
- For exponents that the current P1 has B1=B2 (and not excessively large)
I am running PMinus1 with B1=1000000 B2=20000000 (1M,20M)