PFGW 3.8.3 (with gwnum v28.7) Released

[pepi37]

Originally Posted by rogue

PFGW has supported AVX since version 3.6.2. It is almost always recommended to use the latest release.

Yes, I always download latest version , because there was number of candidates where LLR stops working. In that case PFGW is great choice and it never fail :)

[Antonio]

Flag -f0 doesn't work in version 3.7.7, it recognises the command and then ignores it! See attached screenshot.

[paulunderwood]

You are asking PFGW to do a full primality test, which is slower than a 3-PRP test. Drop the "-t" flag until you find a PRP. If the input file has not been sieved you ought to use the "-f" flag.

[henryzz]

A n+1 or n-1 primality test works by factoring n+1 or n-1. You should be using n+1(-tp) not n-1(-t) for those numbers. Factoring can't be turned off for those tests for a reason. It shouldn't be trying to factor n with the command you used.

[paulunderwood]

Originally Posted by henryzz

A n+1 or n-1 primality test works by factoring n+1 or n-1. You should be using n+1(-tp) not n-1(-t) for those numbers. Factoring can't be turned off for those tests for a reason. It shouldn't be trying to factor n with the command you used.

He should use neither -t nor -tp. A PRP test is enough to discover a likely prime and is quicker. He can then use: -tp -q"candidate". Proving requires factoring of N-1 or N+1 to 33.33%. That is what the screenshot shows

pfgwdoc.txt

-t Deterministic test.
This switch does not require any arguments. It will default to
-tm if not specified, but if specified, must be p, c, or m.
This option defaults to a N-1 test.
This is NOT a probable test.
You will want to use this mode whenever your number is easily
factorable when you subtract 1. (for example n!+1)
If the factorisation will be less then 33.33%, an F-strong test
will be performed.

-tp N+1 test.
uses the N+1 test to check whether the number is prime.
This is NOT a probable test.
You will want to use this mode whenever your number is easily
factorable when you add 1. (for example n!-1)
If the factorisation is less then 33.33%, an F-strong test
will be performed.

-tc Combined N+1 and N-1 test.
When you are short of factoring N-1, or N+1, and the other
has some factors, you can try this mode to achieve a prove.
This too is NOT a probable test.
If the factored portions are F1 and F2, with F1>F2, and 3*F1+F2 is
100% or more, pfgw will be able to complete the proof. If this total
is slightly below 100%, it should still be able to force a proof
with some square tests using the -x flag.

-tm Use optimal choices.
This option will make pfgw use a minimal of factors for the proof.

[Antonio]

Originally Posted by henryzz

A n+1 or n-1 primality test works by factoring n+1 or n-1. You should be using n+1(-tp) not n-1(-t) for those numbers. Factoring can't be turned off for those tests for a reason. It shouldn't be trying to factor n with the command you used.

Thank's - I spotted that mistake just after posting and corrected it.

He should use neither -t nor -tp. A PRP test is enough to discover a likely prime and is quicker He can then use: -tp -q"candidate". Proving requires factoring of N-1 or N+1 to 33%. That is what the screenshot shows

Thank you for the PRP hint Mr. Underwood, it does speed things up somewhat.

[Citrix]

One feature request for PFGW.

For numbers of the form k*b^n+-1 -- when is k is large or b is too large, srsieve cannot sieve them.

I am using the -f flag currently to sieve them.
Is it possible to use GPU to speed up the trial factoring?

Thanks!

[rogue]

Originally Posted by Citrix

Is it possible to use GPU to speed up the trial factoring?!

Yes, but this is really something better suited to an external application.

[ATH]

I tried to run pfgw64.exe -V -tp on the new big wagstaff prps, but it doesn't create any intermediate savefiles and it is a very long job. I tried with Win64PFGW as well.

Is there anyway to force the save files? According to the pfgwdoc.txt there is not, and in that case it is a bug it is not creating them:

6. Save/Resume is automatic. There is no user interaction.
For numbers large enough, the save file is written upon
early shutdown (^C or in WinPFGW, the stop button). Also
the save files are written every 20 minutes or so.

[paulunderwood]

I am running "-tp" too. My o/c 4770k reset itself this morning. I lost 3 days of calculation on the new Wagstaffs. It is going to take 5-6 days to complete them.

[rogue]

pfgw does not create temp files during primality tests. Unfortunately such an enhancement would be very difficult because of how pfgw as written.