mersenneforum.org Sierp base 3 reservations/statuses/primes
 Register FAQ Search Today's Posts Mark Forums Read

 2022-07-03, 14:46 #408 KEP Quasi Admin Thing     May 2005 17218 Posts It's not always bad to be crazy Just curious, what p value constituted optimal sievedepth on GPU for a 1G range for S3 for n>25K to n=50K? (For those interested, sieving on a CPU had an optimal sievedepth of p=10G for that kind of n-range) Good luck with your effort, once I get back to this project, hopefully late next year or early 2024, I will have some new monsters, of wich some of those ressources will also be dedicated to clearing some k's from S3 among other conjectures If the removal rate ends around 46%, per doubling of n, you will remove about 350,000 k values Looking forward to follow your progress - take care
2022-07-03, 17:42   #409
rogue

"Mark"
Apr 2003
Between here and the

2×13×257 Posts

Quote:
 Originally Posted by KEP It's not always bad to be crazy Just curious, what p value constituted optimal sievedepth on GPU for a 1G range for S3 for n>25K to n=50K? (For those interested, sieving on a CPU had an optimal sievedepth of p=10G for that kind of n-range) Good luck with your effort, once I get back to this project, hopefully late next year or early 2024, I will have some new monsters, of wich some of those ressources will also be dedicated to clearing some k's from S3 among other conjectures If the removal rate ends around 46%, per doubling of n, you will remove about 350,000 k values Looking forward to follow your progress - take care
5e10 seemed to be appropriate, but this is probably a little less than optimal. Took a less than 2 days per range of 1G to sieve to that depth. This leaves about 3.6M tests per range of 1G. Some of my cores will complete PRP testing range of 1G in a little over 30 days. Some will need closer to 50 days.

2022-08-09, 16:25   #410
rogue

"Mark"
Apr 2003
Between here and the

2·13·257 Posts

Attached are the primes for n < 50000 for 20e9 < k < 30e9.
Attached Files
 c02x_primes.7z (158.8 KB, 7 views)

2022-08-10, 05:52   #411
gd_barnes

May 2007
Kansas; USA

2×5,569 Posts

Quote:
 Originally Posted by rogue Attached are the primes for n < 50000 for 20e9 < k < 30e9.
Thanks!

Confirming that you have the range of 6G-123G reserved for S3 n=25K-50K ?

Prefer if they are done in k-range order so there are not so many temporary gaps. :-)

Last fiddled with by gd_barnes on 2022-08-10 at 05:52

2022-08-10, 12:20   #412
rogue

"Mark"
Apr 2003
Between here and the

668210 Posts

Quote:
 Originally Posted by gd_barnes Thanks! Confirming that you have the range of 6G-123G reserved for S3 n=25K-50K ? Prefer if they are done in k-range order so there are not so many temporary gaps. :-)
Yes. How I submit the results are dependent upon the computers working on the 10G ranges. Two are fast enough to complete a 10G range in about 5 weeks. The others take about twice that time. My "back of the envelope" estimate is to complete in December based upon the progress I have made.

 Similar Threads Thread Thread Starter Forum Replies Last Post gd_barnes Conjectures 'R Us 2462 2022-08-12 15:32 Siemelink Conjectures 'R Us 1729 2022-08-05 15:35 gd_barnes Conjectures 'R Us 989 2022-06-26 08:35 KEP Conjectures 'R Us 1134 2022-06-24 20:28 gd_barnes Conjectures 'R Us 1417 2022-05-25 19:33

All times are UTC. The time now is 14:44.

Thu Aug 18 14:44:51 UTC 2022 up 12:13, 0 users, load averages: 1.14, 1.35, 1.46