How to make llr do a prp on +3 or +5

[Joshua2]

I assume the best way to test K*2^n + 3 or + 5 is prp-with llr. What is the format to use? I was able to figure out how to get it to do -1 and +1 by copying the newpgen output for a +1 or -1 sieve. Thanks!

[Joshua2]

It seems I should be able to use ABC format, like
ABC $a*2^34000+3 //+1 and -1 didn't work either.
61570182
430131372
...
But it doesn't seem to work.

EDIT: Found this seems to work. However it was slower than pfgw, about 25% as slow actually. Instead of like 3.2 something seconds each its 4.2 something.
ABC $a*2^$b+3
61570182 34000
430131372 34000

Maybe I'm doing something wrong? LLR and Proth take 1.2-1.5 seconds, with LLR program. Also I discovered that ABC $a*2^$b+5 LLR does in the same fast 1.5 seconds, whereas pfgw again takes 3.2. I suspect it is because I was testing +3 and +5 instead of +5 and +7?

[ET_]

Originally Posted by Joshua2

It seems I should be able to use ABC format, like
ABC $a*2^34000+3 //+1 and -1 didn't work either.
61570182
430131372
...
But it doesn't seem to work.

EDIT: Found this seems to work. However it was slower than pfgw, about 25% as slow actually. Instead of like 3.2 something seconds each its 4.2 something.
ABC $a*2^$b+3
61570182 34000
430131372 34000

Maybe I'm doing something wrong? LLR and Proth take 1.2-1.5 seconds, with LLR program. Also I discovered that ABC $a*2^$b+5 LLR does in the same fast 1.5 seconds, whereas pfgw again takes 3.2. I suspect it is because I was testing +3 and +5 instead of +5 and +7?

IIRC, LLR is best optimized for k small.

Luigi