Riesel base 3 reservations/statuses/primes - Page 78

pepi37 · 2017-01-01, 20:07

Quote:

Originally Posted by KEP

Thanks!

It was about 40M tasks and it resulted in ~34800 primes (number may be wrong, is taken on top of my head)

So yes, it was a monumental task, however now you have plenty to sieve for the next many years. Really wish that Riesel base 3 could become a focus project for more users and maybe for BOINC

Unfortunantly I have no ressources available to help you sieve.

Happy new year everyone. Take care!

I believe that we , all, here have enough resources for sieve R3 base. It is just question will we join our resources or not :)

MisterBitcoin · 2017-01-02, 14:04

Quote:

Originally Posted by KEP

Thanks!

It was about 40M tasks and it resulted in ~34800 primes (number may be wrong, is taken on top of my head)

So yes, it was a monumental task, however now you have plenty to sieve for the next many years. Really wish that Riesel base 3 could become a focus project for more users and maybe for BOINC

Unfortunantly I have no ressources available to help you sieve.

Happy new year everyone. Take care!

Happy new year, KEP.
Great job you have done there!

The ck from R3 is ~63G, but take a look at S3.
125050976086 is a "bit" more. ;)
I will focus more on S3, S7, S15.

The range 25G-63G is still at n=25K. Yep, a large amount of work. ;)

About the sieving: S66 is needs some time, but after this (and pepi´s work) is done, I´ll take an Range for sieving. Let me know which pmax you want. ;)

KEP · 2017-01-02, 15:23

Quote:

Originally Posted by MisterBitcoin

Happy new year, KEP.
Great job you have done there!

The ck from R3 is ~63G, but take a look at S3.
125050976086 is a "bit" more. ;)
I will focus more on S3, S7, S15.

The range 25G-63G is still at n=25K. Yep, a large amount of work. ;)

About the sieving: S66 is needs some time, but after this (and pepi´s work) is done, I´ll take an Range for sieving. Let me know which pmax you want. ;)

Thank you and happy new year, MisterBitcoin

Yes, there is still about 56 billion k's untested for S3. If you want to start up a range for k>13G, I'm willing to teach you how to use srbsieve. If you know how to use NewPGen, then everything will work out just fine. Of course Gary has to agree with you on how to do the starting up of untested ranges. I however am willing to help you get it right.

Regarding pMax and sieving, try and ask here in the forum, before starting. Maybe VBCurtis who has done the most of the n>100K sieving, would have a very great suggestion for you on how to use your ressources the best

rebirther · 2017-02-04, 10:23

R3 tested to n=125k (100-125k) (6-10G)

1116 primes found, 5687 remain in this range

Results emailed, Base released

VBCurtis · 2017-02-07, 18:36

Quote:

Originally Posted by rebirther

R3 tested to n=125k (100-125k) (6-10G)

1116 primes found, 5687 remain in this range

Results emailed, Base released

Can you email me the primes found, so that I may continue the sieve?

rebirther · 2017-02-07, 18:59

Quote:

Originally Posted by VBCurtis

Can you email me the primes found, so that I may continue the sieve?

done.

rogue · 2017-02-27, 18:46

Quote:

Originally Posted by VBCurtis

How about tackling a 1G range from 25k to 100k? You don't need your entire 12-thread machine for it, just one core to sieve and then a core or two for LLR. Maybe a month per 1G range on a single core?

Is this one month per core per 1G? I would think that it would take longer. I'm considering taking a range of R3 from n=25K to 100K. If I could get a good estimate for how long this would take on a single core, then I might reserve a few G.

VBCurtis · 2017-02-27, 20:28

Correct- my hand-wavy guess is a (haswell or better) core-month per 1G from 25k to 100k. It has been a very long time since I've actually done such a run, so I might be off by a factor of two; you could always run for a core-week and extrapolate to decide whether you feel like doing 1G or 5G?

Note that testing a range of exponents scales for R3 close to linearly with the exponent, rather than quadratically (more accurately something like exponent ^ 1.4 scaling); so many k's are removed that testing 50-55k is just a bit over twice the time of 25-30k. I think 25-100k testing will remove around half the k's. Disclaimer: this is all recollected from circa 2013 when I studied such things.

gd_barnes · 2017-02-27, 23:02

Testing R3 from n=25K-100K actually removes a little over 2/3rds of all k's. The scaling seems accurate to me.

Siemelink · 2017-03-01, 08:56

Quote:

Originally Posted by Siemelink

As a tribute to this massive effort, I'll take a reservation too.

Reserving k=22G to 23G, n=25K to 100K.

One core does not make for much speed. I've reached 50k.

total k in range 6102
Remaining k 3265
primes 2837 46%

Continuing, Willem.

rogue · 2017-03-01, 13:23

Quote:

Originally Posted by Siemelink

One core does not make for much speed. I've reached 50k.

total k in range 6102
Remaining k 3265
primes 2837 46%

Continuing, Willem.

How long did this take on a single core and what CPU are you running on?

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

2017-02-04, 10:23	#851
rebirther Sep 2011 Germany 2⁶×3²×5 Posts	R3 tested to n=125k (100-125k) (6-10G) 1116 primes found, 5687 remain in this range Results emailed, Base released

2017-02-27, 20:28	#855
VBCurtis "Curtis" Feb 2005 Riverside, CA 2×2,437 Posts	Correct- my hand-wavy guess is a (haswell or better) core-month per 1G from 25k to 100k. It has been a very long time since I've actually done such a run, so I might be off by a factor of two; you could always run for a core-week and extrapolate to decide whether you feel like doing 1G or 5G? Note that testing a range of exponents scales for R3 close to linearly with the exponent, rather than quadratically (more accurately something like exponent ^ 1.4 scaling); so many k's are removed that testing 50-55k is just a bit over twice the time of 25-30k. I think 25-100k testing will remove around half the k's. Disclaimer: this is all recollected from circa 2013 when I studied such things.

2017-02-27, 23:02	#856
gd_barnes May 2007 Kansas; USA 3²×13×89 Posts	Testing R3 from n=25K-100K actually removes a little over 2/3rds of all k's. The scaling seems accurate to me.