PFGW 4.0.4 (with gwnum v30.10) Released

[Happy5214]

Originally Posted by storm5510

This is my major peeve with PFGW, and one of the reasons I do not use it. The other, members working the Riesel prime search strongly urged using LLR. PRP vs LL. I understand their reasons.

PFGW can do a deterministic (i.e. proven prime rather than probably prime) P+1 test on Riesel candidates by passing the -tp argument to it. However, said algorithm is not as efficient as LLR, and I generally only use PFGW on Riesel candidates when the numbers are small (n < 10,000), where the cost of writing the LLR log to disk slows it down considerably versus PFGW (which only writes when a prime is found).

[japelprime]

I also found a big speed difference llr in favor for thouse old cpu I am using. I stick to llr at the moment doing Riesel numbers. Thanks for all the help here.

[paulunderwood]

Originally Posted by japelprime

I also found a big speed difference llr in favor for thouse old cpu I am using. I stick to llr at the moment doing Riesel numbers. Thanks for all the help here.

If there is a "big speed difference" then it is most likely to be due to the different FFT sizes used.

[rogue]

llr will be faster for most, if not all, numbers of the form k*b^n+/-c because it uses a different algorithm. I have purposefully not changed pfgw to use a faster algorithm because it allows for one program to verify the results of the other.

[matzetoni]

Is it possible to automatically stop pfgw once a prp is found?

I read about setting number_primes in the input file, but it doesn't seem to work for me :/
Does number_primes only work with confirmed primes (ignoring prps)?

[rogue]

Originally Posted by matzetoni

Is it possible to automatically stop pfgw once a prp is found?

I read about setting number_primes in the input file, but it doesn't seem to work for me :/
Does number_primes only work with confirmed primes (ignoring prps)?

Let's assume that you have {number_primes,$a,1}. This means that for each distinct value for $a, it will stop PRP testing that value when a prime is found.

So if you have multiple values for $a, it will continue searching.

Post your input file and explain what you are trying to accomplish.

[pepi37]

Originally Posted by matzetoni

Is it possible to automatically stop pfgw once a prp is found?

I read about setting number_primes in the input file, but it doesn't seem to work for me :/
Does number_primes only work with confirmed primes (ignoring prps)?

LLR can do that

[matzetoni]

Originally Posted by rogue

Let's assume that you have {number_primes,$a,1}. This means that for each distinct value for $a, it will stop PRP testing that value when a prime is found.

So if you have multiple values for $a, it will continue searching.

Post your input file and explain what you are trying to accomplish.

I may misinterpreted the function of {number_primes,$a,1}.

I'd like to search for the smallest prp larger than some 10^n and I wish to stop the program once it found it, so it doesn't search the entire sieve file then.

I attached a sieve file for n=4k with candidates of the form 10^4000+c, made with fkbnsieve. So when pfgw finds 10^4000+"something" is prp, I'd like it to stop and search no further.

[rogue]

Originally Posted by matzetoni

I may misinterpreted the function of {number_primes,$a,1}.

I'd like to search for the smallest prp larger than some 10^n and I wish to stop the program once it found it, so it doesn't search the entire sieve file then.

I attached a sieve file for n=4k with candidates of the form 10^4000+c, made with fkbnsieve. So when pfgw finds 10^4000+"something" is prp, I'd like it to stop and search no further.

I have only used number_primes with ABC format, but fkbnsieve doesn't output ABC format.

I think that you will need to use an edited to change to this format:

ABCD $a^4000+$b [10 61] // {number_primes,$a,1}
0 +50

but you might need to experiment. The 0 on the each line means "add 0 to the $a value". This means that $a will also be 10.

[matzetoni]

Originally Posted by rogue

I have only used number_primes with ABC format, but fkbnsieve doesn't output ABC format.

I think that you will need to use an edited to change to this format:

ABCD $a^4000+$b [10 61] // {number_primes,$a,1}
0 +50

but you might need to experiment. The 0 on the each line means "add 0 to the $a value". This means that $a will also be 10.

That works! Thanks a lot!

[bchaffin]

I found another repunit which fails repeatedly when run with special modular reduction. This is using pfgw4.0.1, which uses square_carefully for the last 50 iterations. (See post https://www.mersenneforum.org/showpo...&postcount=415 for a previous case.)

In this case the offender is (10^4568899-1)/9:

Detected in MAXERR>0.45 (round off check) in prp_using_gwnum
Iteration: 15177501/15177550 ERROR: ROUND OFF 0.5>0.45
PFGW will automatically rerun the test with -a1
...
Detected in MAXERR>0.45 (round off check) in prp_using_gwnum
Iteration: 15177501/15177550 ERROR: ROUND OFF 0.5>0.45
PFGW will automatically rerun the test with -a6
(10^4568899-1)/9 ERROR DURING PROCESSING! (96541.0526s+0.0304s)