Sieving for CRUS - Page 67

rogue · 2022-10-07, 10:10

I am sieving all unsieved conjectures at n=10k to 1e11. I might take them to 1e12.

rogue · 2022-10-12, 17:04

Quote:

Originally Posted by rogue

I am sieving all unsieved conjectures at n=10k to 1e11. I might take them to 1e12.

I should be able to finish all 39 conjectures at n=10k that are sieved < 1e12 to 1e12 by January. I have 3 GPUs happily sieving away at them.

wombatman · 2022-10-19, 15:04

Sieving S63 from n=25k-50k for all remaining k-values. Needs to get to ~30 secs/factor.

rogue · 2022-10-19, 16:13

Quote:

Originally Posted by wombatman

Sieving S63 from n=25k-50k for all remaining k-values. Needs to get to ~30 secs/factor.

With which program? This is a big task.

wombatman · 2022-10-19, 17:09

Quote:

Originally Posted by rogue

With which program? This is a big task.

srsieve2cl. Currently approaching P=1e10 in the next day or two.

rogue · 2022-10-19, 19:28

Quote:

Originally Posted by wombatman

srsieve2cl. Currently approaching P=1e10 in the next day or two.

Nice. What command line oparameters are you using?

wombatman · 2022-10-19, 20:18

Quote:

Originally Posted by rogue

Nice. What command line oparameters are you using?

It took a little trial-and-error (and is likely not optimized), but I'm currently running with -g16 -K8 -M500. I think the -M option is only necessary because I'm starting at such a low p-value. I may try a smaller value with -K as well.

Currently, those options give me throughput of ~2.5k-2.8k p/sec (depending on if something else is using the GPU) at p=500e6 with a found factor rate of ~180/sec. CPU was giving around 200 p/sec for comparison.

rogue · 2022-10-19, 21:06

Quote:

Originally Posted by wombatman

It took a little trial-and-error (and is likely not optimized), but I'm currently running with -g16 -K8 -M500. I think the -M option is only necessary because I'm starting at such a low p-value. I may try a smaller value with -K as well.

Currently, those options give me throughput of ~2.5k-2.8k p/sec (depending on if something else is using the GPU) at p=500e6 with a found factor rate of ~180/sec. CPU was giving around 200 p/sec for comparison.

I posted srsieve2cl at sourceforge. Take a look at the -b parameter, which is new. By using -b you might be able to reduce -K, which can in turn increase throughput for the GPU. Removing terms will not allow you to adjust -K, only removing sequences.

wombatman · 2022-10-20, 01:46

Quote:

Originally Posted by rogue

I posted srsieve2cl at sourceforge. Take a look at the -b parameter, which is new. By using -b you might be able to reduce -K, which can in turn increase throughput for the GPU. Removing terms will not allow you to adjust -K, only removing sequences.

Will do. Thanks!

gd_barnes · 2022-11-03, 09:02

All remaining bases with < 20 k's remaining at a search depth of n=100K have now been fully sieved by Yoyo to P=1e15.

We are currently working on sieving the remaining bases with 3 k's remaining at a search depth of n<500K. Most already have sieve files but there are a few without them.

sweety439 · 2022-11-07, 23:15

Can you link this page in the 1st post in this thread?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Sieving Drive all base 2/4 k's worked by CRUS	gd_barnes	Conjectures 'R Us	143	2014-10-21 23:55
Some CRUS stats	vmod	Conjectures 'R Us	213	2014-02-28 21:23
What are your CRUS plans?	rogue	Conjectures 'R Us	35	2013-11-09 09:03
CRUS-like sieving challenge	CRGreathouse	Puzzles	24	2011-10-28 18:30
how high will CRUS go	TimSorbet	Conjectures 'R Us	1	2010-11-08 20:50

2022-10-07, 10:10	#727
rogue "Mark" Apr 2003 Between here and the 5·7·199 Posts	I am sieving all unsieved conjectures at n=10k to 1e11. I might take them to 1e12.

2022-10-19, 15:04	#729
wombatman I moo ablest echo power! May 2013 1852₁₀ Posts	Sieving S63 from n=25k-50k for all remaining k-values. Needs to get to ~30 secs/factor.

2022-11-03, 09:02	#736
gd_barnes "Gary" May 2007 Overland Park, KS 5²·11·43 Posts	All remaining bases with < 20 k's remaining at a search depth of n=100K have now been fully sieved by Yoyo to P=1e15. We are currently working on sieving the remaining bases with 3 k's remaining at a search depth of n<500K. Most already have sieve files but there are a few without them.

2022-11-07, 23:15	#737
sweety439 "99(4^34019)99 palind" Nov 2016 (P^81993)SZ base 36 2·7·263 Posts	Can you link this page in the 1st post in this thread?