mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Proth Prime Search (https://www.mersenneforum.org/forumdisplay.php?f=109)
-   -   Question about sieving (https://www.mersenneforum.org/showthread.php?t=25083)

bur 2020-09-04 07:12

Thanks a lot for all the help so far!

I read about srsieve2, but wasn't sure if it's ready yet. Maybe rogue can say if for fixed k and n in the 3M-4M range which of sr1sieve, sr2sieve or srsieve2 is fastest?

[QUOTE]sr2sieve expects a file in format usable by LLR (-g flag from srfile), not abcd[/QUOTE]sr2sieve accepts abcd as input, I'm currently using it.

rogue 2020-09-04 12:12

[QUOTE=bur;555988]Thanks a lot for all the help so far!

I read about srsieve2, but wasn't sure if it's ready yet. Maybe rogue can say if for fixed k and n in the 3M-4M range which of sr1sieve, sr2sieve or srsieve2 is fastest?

sr2sieve accepts abcd as input, I'm currently using it.[/QUOTE]

If k is so large that you cannot build Legendre tables, then srsieve2 is probably faster. The Legendre logic fully coded and tested in srsieve2.

VBCurtis 2020-09-04 15:22

[QUOTE=bur;555988]sr2sieve accepts abcd as input, I'm currently using it.[/QUOTE]

Sorry for the typo, that was intended to be sr1sieve not sr2sieve in the part you quoted.

You should test the 3 programs yourself to see which is faster!

bur 2020-09-06 17:44

I will do a comparison later, first I want to finish sieving... I'm at 5e12 now and factors are coming in considerably slower, but still at about 3 min per factor. The range 3320000 <= n <= 4100000 is down to 14550 candidates.



Somewhere I read proth20 did one test on a gtx 1660 in 6-7 minutes (forgot the specifics of the candidate). So I will continue sieving for a while.

bur 2020-10-18 15:42

At PG forum it was said that sr1sieve is supposed to be faster than sr2sieve for fixed k, so I wanted to give it a try, but cannot find binary for windows. Deoes someone have a link? Thanks.


BTW, I tried srsieve2 on fixed k with magnitude 1E6 and n < 4E6 and only got 36,000 P/s with 70E12 < P < 71E13! With sr2sieve I get 10,000,000 and more. Is it really that slow for these parameters or is there an error somewhere?

rogue 2020-10-18 16:01

The latest Windows build of sr1sieve can be found in [URL="https://www.mersenneforum.org/showthread.php?t=15833"]this thread[/URL].

sr2sieve supports Legendre tables and srsieve2 does not have that feature yet. That is likely why it is faster. For k that are too large for Legendre tables, srsieve2 might be faster.


All times are UTC. The time now is 02:11.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.