Riesel base 3 reservations/statuses/primes - Page 75

VBCurtis · 2016-06-09, 22:26

Reserving 6G to 10G for sieving from 100k to 500k.
These will be tested by BOINC, eventually.

rebirther · 2016-06-28, 17:41

R3 tested to n=500k (300-500k) (1-2.147G)

241 primes found, 519 remain in this range

Results emailed, Base released

henryzz · 2016-06-28, 18:23

What is the largest first unproven k for a base? Is it the one for R3?

gd_barnes · 2016-06-28, 18:45

Quote:

Originally Posted by henryzz

What is the largest first unproven k for a base? Is it the one for R3?

No. S3 is k=6363484.

VBCurtis · 2016-06-30, 06:04

Quote:

Originally Posted by rebirther

R3 tested to n=500k (300-500k) (1-2.147G)

241 primes found, 519 remain in this range

Results emailed, Base released

Yay! Thanks to this result, I'll soon be starting a sieve for all k's under 2.147G from 500k to 3M with sr2sieve. This will take me a while to reach depth sufficient to send BOINC another batch to test, but by 2017 we should reach top-5000 range for R3!

rebirther · 2016-07-01, 16:40

Reserving R3 to n=200k (150-200k) (2.147-4G) for BOINC

rogue · 2016-07-11, 20:41

What is the plan for getting the rest of R3 to n=100000? Are people going to pick out ranges of 1G and test from n=25000 to n=10000? Do we have an estimate for how deeply to sieve the range? How long would it take to sieve 1,000 k in that range? Do we have an estimate for how long it would take a single "middle-of-the-road" CPU to test 1000 sieved k from n=25000 to n=100000?

VBCurtis · 2016-07-11, 21:28

I haven't done a 1G range for 25-100k yet, but some guesswork based on the sieve depths for 1G ranges from 100k to 500k:
Sieve about a thread-week, to something like 6G, then test 25-40k.
Sieve another week, to 15G or so, test 40-60k.
Sieve another week, to ~25G, test 60-100k.

If you're more interested in testing than sieving, I am interested in doing the sieving for a 1G range to refine these estimates, but I don't wish to test. I'll report sieve speed and sec/factor at each phase, my coworker should report test time for the first and last candidate in each file; once complete we can publicly do some calculations.

VBCurtis · 2016-07-11, 21:35

Quote:

Originally Posted by VBCurtis

Yay! Thanks to this result, I'll soon be starting a sieve for all k's under 2.147G from 500k to 3M with sr2sieve. This will take me a while to reach depth sufficient to send BOINC another batch to test, but by 2017 we should reach top-5000 range for R3!

The sieve has reached 200G. I was just kidding about top-5000 level by 2017; on 5 threads I'm getting ~260kp/sec, so it'll take me 4-5 calendar months (~2 thread-years!) to reach 3T for the first (non top-5000) piece to go to BOINC. Top-5000 level matches roughly with 10T on the sieve, 5+ thread years.

There are ~40M candidates for 941 k's.

KEP · 2016-07-12, 12:42

Quote:

Originally Posted by VBCurtis

I haven't done a 1G range for 25-100k yet, but some guesswork based on the sieve depths for 1G ranges from 100k to 500k:
Sieve about a thread-week, to something like 6G, then test 25-40k.
Sieve another week, to 15G or so, test 40-60k.
Sieve another week, to ~25G, test 60-100k.

If you're more interested in testing than sieving, I am interested in doing the sieving for a 1G range to refine these estimates, but I don't wish to test. I'll report sieve speed and sec/factor at each phase, my coworker should report test time for the first and last candidate in each file; once complete we can publicly do some calculations.

... that pretty much looks like what my Sandy Bridge required. I did however sieve to p=34G for a 1G range and also for the 2G ranges, before reaching optimal sievedepth for my Sandy Bridge for n=100K.

I believe that due to the high removal rate (approximately 46% k's removed every double of n) that p=25G is also very well sieved. At least it looked optimal on my Hasswell, compared to the Sandy Bridge. So if the computer is something between an i5-2300 and an i5-4670 (Haswell), optimal sievedepth would be approximately 30G. But for 25K-40K and 40K-60K your sievedepths is a full match compared to where I broke off these n ranges

Ps. Even though I stated earlier that I wanted to do more R3 work, once R3 13G-21G completes on my brothers Xeon (around september) I will not be doing any more R3 work (except maybe a little Top5000 sieving if I can get the sievefile from Curtis) untill I meet my goals for SR16

Puzzle-Peter · 2016-07-15, 05:43

Quote:

Originally Posted by Puzzle-Peter

Reserving k=21G to 22G, n=25K to 100K.

Done. Uploading the results file to sendspace right now. Gary will receive a message as to where to download the data.

Reserving k=23G to 25G

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Bases 33-100 reservations/statuses/primes	Siemelink	Conjectures 'R Us	1694	2021-08-06 20:41
Bases 6-32 reservations/statuses/primes	gd_barnes	Conjectures 'R Us	1398	2021-08-06 12:49
Bases 251-500 reservations/statuses/primes	gd_barnes	Conjectures 'R Us	2305	2021-08-04 15:09
Bases 101-250 reservations/statuses/primes	gd_barnes	Conjectures 'R Us	908	2021-08-01 07:48
Sierp base 3 reservations/statuses/primes	gd_barnes	Conjectures 'R Us	388	2020-10-21 19:42

2016-06-09, 22:26	#815
VBCurtis "Curtis" Feb 2005 Riverside, CA 2×2,437 Posts	Reserving 6G to 10G for sieving from 100k to 500k. These will be tested by BOINC, eventually.

2016-06-28, 17:41	#816
rebirther Sep 2011 Germany 2⁶·3²·5 Posts	R3 tested to n=500k (300-500k) (1-2.147G) 241 primes found, 519 remain in this range Results emailed, Base released

2016-06-28, 18:23	#817
henryzz Just call me Henry "David" Sep 2007 Cambridge (GMT/BST) 5887₁₀ Posts	What is the largest first unproven k for a base? Is it the one for R3?

2016-07-01, 16:40	#820
rebirther Sep 2011 Germany 2880₁₀ Posts	Reserving R3 to n=200k (150-200k) (2.147-4G) for BOINC

2016-07-11, 20:41	#821
rogue "Mark" Apr 2003 Between here and the 2²·7·227 Posts	What is the plan for getting the rest of R3 to n=100000? Are people going to pick out ranges of 1G and test from n=25000 to n=10000? Do we have an estimate for how deeply to sieve the range? How long would it take to sieve 1,000 k in that range? Do we have an estimate for how long it would take a single "middle-of-the-road" CPU to test 1000 sieved k from n=25000 to n=100000?

2016-07-11, 21:28	#822
VBCurtis "Curtis" Feb 2005 Riverside, CA 2×2,437 Posts	I haven't done a 1G range for 25-100k yet, but some guesswork based on the sieve depths for 1G ranges from 100k to 500k: Sieve about a thread-week, to something like 6G, then test 25-40k. Sieve another week, to 15G or so, test 40-60k. Sieve another week, to ~25G, test 60-100k. If you're more interested in testing than sieving, I am interested in doing the sieving for a 1G range to refine these estimates, but I don't wish to test. I'll report sieve speed and sec/factor at each phase, my coworker should report test time for the first and last candidate in each file; once complete we can publicly do some calculations.