mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Aliquot Sequences

Reply
 
Thread Tools
Old 2019-12-24, 22:34   #320
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

2×3×733 Posts
Default

Yes, please do! Your work plus mine means half a t55 will be done by tomorrow, so only a t60 (more like 80% of one, if we use 0.31 * digits for ECM depth) needs to be sent to yoyo.

Swellman will let us know when he sends that in, and we'll stop running our t55 once he does.

I'm running on half an i7-5820 (6 cores) and half an HP Z620 (20 cores Ivy Bridge Xeon). I'll keep running t55-sized curves (B1=150M, to be specific) after my 0.2*t55 is done, since you're helping, so that we don't have to send any to yoyo.
VBCurtis is offline   Reply With Quote
Old 2019-12-24, 23:53   #321
swellman
 
swellman's Avatar
 
Jun 2012

23·359 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
Yes, please do! Your work plus mine means half a t55 will be done by tomorrow, so only a t60 (more like 80% of one, if we use 0.31 * digits for ECM depth) needs to be sent to yoyo.

Swellman will let us know when he sends that in, and we'll stop running our t55 once he does.

I'm running on half an i7-5820 (6 cores) and half an HP Z620 (20 cores Ivy Bridge Xeon). I'll keep running t55-sized curves (B1=150M, to be specific) after my 0.2*t55 is done, since you're helping, so that we don't have to send any to yoyo.
Second the notion that you both run ECM to achieve a complete t55 if possible. I can enqueue this aliquot in Yoyo anytime but there will likely be a 1-2 week delay before Yoyo even starts to work it. See the queue here.

If no factors are found by either of you then I can then start Yoyo on ECM towards a t60.
swellman is offline   Reply With Quote
Old 2019-12-25, 20:39   #322
jagotu
 
Oct 2019

3×7 Posts
Default C189 i2206

Just hit a total of 4500@11e7. If we want to hit full t55, that's something like 20000 curves?
So if we split it in half, it'll take me about two-three more days to hit the 10000.

Btw. is there an advantage for 15e7 over 11e7?
jagotu is offline   Reply With Quote
Old 2019-12-25, 20:59   #323
swellman
 
swellman's Avatar
 
Jun 2012

1011001110002 Posts
Default

Quote:
Originally Posted by jagotu View Post
Just hit a total of 4500@11e7. If we want to hit full t55, that's something like 20000 curves?
So if we split it in half, it'll take me about two-three more days to hit the 10000.

Btw. is there an advantage for 15e7 over 11e7?
The t55 level is typically defined as 18,000 curves @B1=11e7.
swellman is offline   Reply With Quote
Old 2019-12-25, 21:04   #324
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

2·3·733 Posts
Default

Quote:
Originally Posted by jagotu View Post
Just hit a total of 4500@11e7. If we want to hit full t55, that's something like 20000 curves?
So if we split it in half, it'll take me about two-three more days to hit the 10000.

Btw. is there an advantage for 15e7 over 11e7?
I'm a little faster than you are; if you hit 7000 total at 11e7, I'll take care of the rest of the t55.

As for 15e7 vs 11e7: If I run GMP-ECM -v for one curve at each size, the estimated time to complete a t55 is a bit shorter for 15e7, while the chances of finding a larger factor are better. It's a win-win, but a very small one. It has to do with the default B2 choices GMP-ECM makes for particular B1 values + memory use for those B2s (in a sense, more memory use for a small speedup).
I find a similar result for 6e7 vs 43e6 for a t50, and for 2e7 vs 11e6 for a t45. Basically, I looked for the smallest B1 values above the standard choice for which GMP-ECM chooses k=2 (and a higher memory-use + larger B2 choice), and in each case I find the larger B1 choice yields a faster t-whatever.

Last fiddled with by VBCurtis on 2019-12-25 at 21:11
VBCurtis is offline   Reply With Quote
Old 2019-12-25, 21:04   #325
jagotu
 
Oct 2019

3×7 Posts
Default

Quote:
Originally Posted by swellman View Post
The t55 level is typically defined as 18,000 curves @B1=11e7.
That's the page I was coming from, although I was looking at the GMP-ECM 7 column...

Quote:
Originally Posted by VBCurtis View Post
As for 15e7 vs 11e7: If I run GMP-ECM -v for one curve at each size, the estimated time to complete a t55 is a bit shorter for 15e7, while the chances of finding a larger factor are better.
A very interesting experiment! Will definitely try to test this on my HW once we're done with this.

Last fiddled with by jagotu on 2019-12-25 at 21:06
jagotu is offline   Reply With Quote
Old 2019-12-26, 09:07   #326
jagotu
 
Oct 2019

3×7 Posts
Default Default C189 i2206

Hit a total of 7180@11e7 (managed to free a few more cores so it went a bit faster).

Will try a few curves at 26e7.

Last fiddled with by jagotu on 2019-12-26 at 09:34
jagotu is offline   Reply With Quote
Old 2019-12-27, 18:23   #327
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

112E16 Posts
Default

I ran:
2200@6e7
2000@8e7
3200@15e7
930@30e7
200@42e7

Combined with jagotu's ~7200@11e7, that's a t55 and 14% of a t60.

That leaves 75% of a t60 for yoyo, please.
VBCurtis is offline   Reply With Quote
Old 2019-12-27, 18:34   #328
jagotu
 
Oct 2019

258 Posts
Default

In addition to the 7180@11e7 I ran 2535@26e7, so that should maybe help with the t60 some more (another 6% of t60?).

Last fiddled with by jagotu on 2019-12-27 at 18:39
jagotu is offline   Reply With Quote
Old 2019-12-27, 19:58   #329
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

2×3×733 Posts
Default

Nice! So 20% of t60 is done, leaving about 55% for yoyo. I'm using 0.31* digits to get ~59 digits of ECM, which is something like 75% of t60. (my previous post asked for too much of a t60)

Say, 24,000 curves at 26e7 from yoyo?
VBCurtis is offline   Reply With Quote
Old 2019-12-27, 20:01   #330
swellman
 
swellman's Avatar
 
Jun 2012

23×359 Posts
Default

I’ll queue the job with Yoyo. Two staggered separate rounds of 12,000 curves @B1=260e6.

Eta: First job enqueued.

Last fiddled with by swellman on 2019-12-28 at 02:45 Reason: Typo
swellman is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Primes in n-fibonacci sequence and n-step fibonacci sequence sweety439 And now for something completely different 17 2017-06-13 03:49
Team sieve #41: C165 from 3366:i2098 RichD Aliquot Sequences 36 2013-11-29 07:03
80M to 64 bits ... but not really reserved petrw1 Lone Mersenne Hunters 82 2010-01-11 01:57
What's the next in the sequence? roger Puzzles 16 2006-10-18 19:52
Sequence Citrix Puzzles 5 2005-09-14 23:33

All times are UTC. The time now is 20:34.

Wed Oct 28 20:34:43 UTC 2020 up 48 days, 17:45, 1 user, load averages: 1.86, 1.86, 1.90

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, 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.