Sieving for CRUS

[lalera]

hi,
here is a sieve-file for R15
n= 200k to 400k
sieved to 29T

[lalera]

hi,
reserving R6 n=5M to 6M to sieve for BOINC

[VBCurtis]

Lalera-
Look into sieving 5M-10M instead, or an even-larger region. It's pretty inefficient to sieve a small range compared to a large one ("small" measured as either number of expected primes in the candidate pool, or multiple of starting exponent).

If you would have sieved 5-6M to, say, 100T, that same sieve depth might take 60% more time for 5-8M, but you end up with a file that won't need all that much more effort to be ready to test 6-8M. That extra time may not be worth it to you, but you should be aware of how little it is.

[lalera]

Originally Posted by VBCurtis

Lalera-
Look into sieving 5M-10M instead, or an even-larger region. It's pretty inefficient to sieve a small range compared to a large one ("small" measured as either number of expected primes in the candidate pool, or multiple of starting exponent).

If you would have sieved 5-6M to, say, 100T, that same sieve depth might take 60% more time for 5-8M, but you end up with a file that won't need all that much more effort to be ready to test 6-8M. That extra time may not be worth it to you, but you should be aware of how little it is.

hi,
i do not like to take a bigger range for R6 because it would take too long for me
and maybe there is a prime in that range

[gd_barnes]

Originally Posted by lalera

hi,
reserving R6 n=5M to 6M to sieve for BOINC

You do as you please but in my opinion, sieving only n=5M-6M is a waste of long-term resources. There is very little chance of a prime in that range. Sieving anything where nmax/nmin < 2 is inefficient as implied by Curtis. Consider sieving n=5M-10M or getting help with sieving or not sieving it at all.

[lalera]

Originally Posted by gd_barnes

You do as you please but in my opinion, sieving only n=5M-6M is a waste of long-term resources. There is very little chance of a prime in that range. Sieving anything where nmax/nmin < 2 is inefficient as implied by Curtis. Consider sieving n=5M-10M or getting help with sieving or not sieving it at all.

hi,
for the range R6, n=5M-6M
i estimate that after sieving there are about 5000 candidates left for prp-testing
on a sbe3930k one cand. would take 24h-36h - that is about 6250 days on one core
i think that you are both right that it is more efficient to take a bigger range
of exponents - but it would take too long for me to sieve - several months
(and i think on R72)

so please do cancel my reservation for R6

[gd_barnes]

Originally Posted by lalera

hi,
for the range R6, n=5M-6M
i estimate that after sieving there are about 5000 candidates left for prp-testing
on a sbe3930k one cand. would take 24h-36h - that is about 6250 days on one core
i think that you are both right that it is more efficient to take a bigger range
of exponents - but it would take too long for me to sieve - several months
(and i think on R72)

so please do cancel my reservation for R6

If you would like to sieve a small base with one k remaining that is at a high search depth, may I suggest sieving S9 for n=2M-5M? Size-wise, it would be more manageable for you and testing it would bring it up to where R6 is currently.

[lalera]

Originally Posted by gd_barnes

If you would like to sieve a small base with one k remaining that is at a high search depth, may I suggest sieving S9 for n=2M-5M? Size-wise, it would be more manageable for you and testing it would bring it up to where R6 is currently.

hi,
i want to do only riesel bases

[lalera]

hi,
here is a sieve-file for R171
n= 25k to 100k
sieved to 7T

[gd_barnes]

Originally Posted by lalera

hi,
i want to do only riesel bases

The Sierpinski bases will start to feel lonely and sad. They will begin to cry.

[lalera]

Originally Posted by gd_barnes

The Sierpinski bases will start to feel lonely and sad. They will begin to cry.

hi,
at this time i have much to do - but maybe next year