mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Conjectures 'R Us

Reply
 
Thread Tools
Old 2020-12-29, 06:00   #1057
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

52·11·17 Posts
Default

I'm still working R3 from 60-63G. I won't be the one to hold back the 2021 goal to get R3 to 100k, so I've split 61-63G onto a bunch of cores in n=5K chunks (e.g. 57-65, 65-70, 70-75, etc to 100). It's an old Xeon, not very fast, so it'll still take me 6 months or so to get 60-63G to 100k.
60-61G is just above 70k now, and split into 3 pieces to finish to 100k in the same time frame.
VBCurtis is offline   Reply With Quote
Old 2020-12-29, 14:15   #1058
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

2·32·347 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
I'm still working R3 from 60-63G. I won't be the one to hold back the 2021 goal to get R3 to 100k, so I've split 61-63G onto a bunch of cores in n=5K chunks (e.g. 57-65, 65-70, 70-75, etc to 100). It's an old Xeon, not very fast, so it'll still take me 6 months or so to get 60-63G to 100k.
60-61G is just above 70k now, and split into 3 pieces to finish to 100k in the same time frame.
If possible, PRPNet is your friend for this type of task. It also makes it easier for you to estimate when it will complete.
rogue is offline   Reply With Quote
Old 2020-12-29, 17:11   #1059
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

124316 Posts
Default

I suppose it's time I look into learning how to use prpnet. Thanks for the reminder.
VBCurtis is offline   Reply With Quote
Old 2021-01-07, 17:05   #1060
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

186616 Posts
Default

Taking R3 35G range to n=100000. I needs to be sieved more deeply first. The factor removal rate is about 1/6 of what it needs to be before testing.

Depending upon how this goes, I will go on to 36G, etc.

Last fiddled with by rogue on 2021-01-07 at 17:05
rogue is offline   Reply With Quote
Old 2021-01-07, 17:30   #1061
MisterBitcoin
 
MisterBitcoin's Avatar
 
"Nuri, the dragon :P"
Jul 2016
Good old Germany

80210 Posts
Default

Quote:
Originally Posted by rogue View Post
Taking R3 35G range to n=100000. I needs to be sieved more deeply first. The factor removal rate is about 1/6 of what it needs to be before testing.

Depending upon how this goes, I will go on to 36G, etc.



Let me know if you need someone for deeper sieving >.>
MisterBitcoin is offline   Reply With Quote
Old 2021-01-07, 18:07   #1062
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

2×32×347 Posts
Default

Quote:
Originally Posted by MisterBitcoin View Post
Let me know if you need someone for deeper sieving >.>
Right now I'm using srsieve2cl, which I haven't released yet. Its kicking out slightly more than 1 factor per second at where the sieving was left off. It looks like I can sieve about 48G a day for the 3257 sequences in that range. We'll see where it is tomorrow.

I actually realized that I used the wrong n when estimating the factor removal rate. I need to remove one term every 10 seconds. I will let everyone know how far it has to be sieved with srsieve2cl when it is done.
rogue is offline   Reply With Quote
Old 2021-01-07, 18:15   #1063
MisterBitcoin
 
MisterBitcoin's Avatar
 
"Nuri, the dragon :P"
Jul 2016
Good old Germany

32216 Posts
Default

Quote:
Originally Posted by rogue View Post
Right now I'm using srsieve2cl, which I haven't released yet. Its kicking out slightly more than 1 factor per second at where the sieving was left off. It looks like I can sieve about 48G a day for the 3257 sequences in that range. We'll see where it is tomorrow.

I actually realized that I used the wrong n when estimating the factor removal rate. I need to remove one term every 10 seconds. I will let everyone know how far it has to be sieved with srsieve2cl when it is done.

48e9 in a single day is a word. What processor are you using if i might ask?
MisterBitcoin is offline   Reply With Quote
Old 2021-01-07, 19:01   #1064
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

11000011001102 Posts
Default

Quote:
Originally Posted by MisterBitcoin View Post
48e9 in a single day is a word. What processor are you using if i might ask?
i7-8805H at 2.6 Ghz. I am using only a single worker so most of the time is spent in the GPU, a Quadro P3200.
rogue is offline   Reply With Quote
Old 2021-01-08, 13:10   #1065
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

2·32·347 Posts
Default

I will revise my estimate to about 32G per day as opposed to what I wrote yesterday. It is easier to get accurate numbers after many hours of runtime. Here is the current line:

p=61209583927, 14.10K p/sec, 83052 factors found at 1.43 sec per factor (last 134 min)

so it has been running a little under 20 hours.
rogue is offline   Reply With Quote
Old 2021-01-09, 06:05   #1066
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

1030910 Posts
Default

Are you sieving only k=35G-36G ?

I posted those sieve files for R3 as fully sieved to P=35G using srsieve. What is your test time at n=80K for R3? I suggest that you use n=80K for a test time. This is an extremely prime base.

IMHO sieving to 35G times 6 will be a waste of sieving effort. At most I feel like it could be sieved twice as high and that's only if your new siever is much faster than srsieve.

It took me a couple months on 6-8 cores just to sieve k=25G-55G to P=35G and I was getting a removal rate close to the LLR test time for n=80K.

Last fiddled with by gd_barnes on 2021-01-09 at 06:07
gd_barnes is offline   Reply With Quote
Old 2021-01-09, 13:40   #1067
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

2·32·347 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
Are you sieving only k=35G-36G ?

I posted those sieve files for R3 as fully sieved to P=35G using srsieve. What is your test time at n=80K for R3? I suggest that you use n=80K for a test time. This is an extremely prime base.

IMHO sieving to 35G times 6 will be a waste of sieving effort. At most I feel like it could be sieved twice as high and that's only if your new siever is much faster than srsieve.

It took me a couple months on 6-8 cores just to sieve k=25G-55G to P=35G and I was getting a removal rate close to the LLR test time for n=80K.
My WAG is that I can do a range 1G in 8 months on a single core. I have 24 cores at my disposal.

I did a test at n=70000, but you are correct, n=80000 is probably more accurate. When I test this range I should be able to get a fairly accurate estimate of the factor removal rate I need based upon how many tests are skipped due to found primes. With PRPNet I can easily compute the average time to do a test across millions of completed tests.

I cannot sieve all k for 35G-45G at once with srsieve2cl. I could probably do a range of 2G at a time (with a single worker thread) before running into memory limits of my GPU. When I start sieving the next range, I will take a bigger chunk. With these large k I have not run any tests to determine how much saving there could be by adding more k.

I am getting a factor removal rage of about 1 candidate per second at p=64G with srsieve2cl with this range so clearly it needs to sieved more deeply.
rogue is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Bases 251-500 reservations/statuses/primes gd_barnes Conjectures 'R Us 2257 2021-02-27 09:18
Bases 101-250 reservations/statuses/primes gd_barnes Conjectures 'R Us 890 2021-02-24 18:35
Bases 33-100 reservations/statuses/primes Siemelink Conjectures 'R Us 1677 2021-02-21 21:42
Bases 4-32 reservations/statuses/primes gd_barnes Conjectures 'R Us 1431 2021-02-18 16:05
Sierp base 3 reservations/statuses/primes gd_barnes Conjectures 'R Us 388 2020-10-21 19:42

All times are UTC. The time now is 10:19.

Fri Mar 5 10:19:24 UTC 2021 up 92 days, 6:30, 0 users, load averages: 2.21, 1.93, 2.03

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

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.