mersenneforum.org  

Go Back   mersenneforum.org > Search Forums

Showing results 1 to 25 of 44
Search took 0.01 seconds.
Search: Posts Made By: Till
Forum: Factoring 2021-07-24, 15:19
Replies: 74
Views: 13,361
Posted By Till
I finished the next run on that 400 bit number...

I finished the next run on that 400 bit number and this time logged the missing quantity (touched sieve locations), too. As expected, using only as many threads as physical cores was significantly...
Forum: Factoring 2021-07-21, 19:18
Replies: 74
Views: 13,361
Posted By Till
Ok, so the number of collected partials is...

Ok, so the number of collected partials is similar again. I had
* run 1: factorBaseSize=304k, relations={65k perfect, 44k from 1LP, 195k involving 2LP}, collected partials={1075k 1LP, 7046k 2LP}
*...
Forum: Factoring 2021-07-21, 18:02
Replies: 74
Views: 13,361
Posted By Till
So far I haven't been collecting exactly the same...

So far I haven't been collecting exactly the same data, but lets see.

We might compare your C120 to my 400bit/C121. I had two runs on that number, the first (slower) one with Cmult=0.32 and the...
Forum: Factoring 2021-07-21, 14:13
Replies: 74
Views: 13,361
Posted By Till
Thanks. My opinion is that the missing...

Thanks. My opinion is that the missing "post-filter" of sieve hits was the biggest hole in my code so far, and I was aware of it for a few years, just didn't manage to do right. I am happy that it...
Forum: Factoring 2021-07-20, 16:18
Replies: 74
Views: 13,361
Posted By Till
My factoring package experienced some notable...

My factoring package experienced some notable improvements lately.

1. The most important one resulted from following this line:


I had the small primes variant implemented before, but passed...
Forum: Factoring 2021-05-28, 17:35
Replies: 74
Views: 13,361
Posted By Till
Of course I wouldn't say that 151 is always...

Of course I wouldn't say that 151 is always faster than 1. But the numbers are a bit more than specific examples. Those where all the cases I found in my test where my SIQS used a multiplier > 73. So...
Forum: Factoring 2021-05-27, 18:15
Replies: 74
Views: 13,361
Posted By Till
Hi Ben, today I did several hundred runs on 240...

Hi Ben,
today I did several hundred runs on 240 bit numbers of my PSIQS checking Knuth-Schroeppel multipliers k up to 480 vs. your k<=73 cutoff and compared the factor logs manually whenever a...
Forum: Factoring 2021-05-25, 21:19
Replies: 74
Views: 13,361
Posted By Till
Very interesting data, thanks. So when it...

Very interesting data, thanks.
So when it comes to even k, you are only checking (some) k with k==2 (mod 4), and apart of k=2, only k=6, 10 and 14 have been successful.

...
Forum: Factoring 2021-05-25, 20:31
Replies: 74
Views: 13,361
Posted By Till
I'll have to investigate this... Is...

I'll have to investigate this...



Is k=2 a special case, or did you also find other k with kN == 2 (mod 8) helpful, or even with kN == 4 or 6 (mod 8) ?
Forum: Factoring 2021-05-25, 18:01
Replies: 74
Views: 13,361
Posted By Till
Thank you. Actually my last serious...

Thank you.


Actually my last serious adjustment experiments were some time ago, don't know if it's 2 or 3 years. I just got my new computer and fed the numbers to it. First some 360 bit (avg. 2...
Forum: Factoring 2021-05-25, 15:23
Replies: 74
Views: 13,361
Posted By Till
400 bit

With some better hardware at hand, I factored my first 400 bit number (a C121) using PSIQS :smile:


It took about 2 days and 8 hours on a Ryzen 3900X, running 20 sieve threads and using Dario...
Forum: Factoring 2018-10-06, 10:50
Replies: 74
Views: 13,361
Posted By Till
New project scope, name and location

In the last time I couldn't find any notable improvements anymore. Thus I decided to broaden the scope of the project and moved it to github. You find it here:...
Forum: Factoring 2018-03-05, 17:36
Replies: 74
Views: 13,361
Posted By Till
Thanks Ben. I'ld estimate that the speedup by...

Thanks Ben. I'ld estimate that the speedup by Q2(x) is 5-10%. You'ld have to adjust KS, too, because kN==1 (mod 8) becomes more favorable then.
Forum: Factoring 2018-03-05, 17:25
Replies: 74
Views: 13,361
Posted By Till
Thanks! I'ld like to attack RDS' lattice QS idea,...

Thanks! I'ld like to attack RDS' lattice QS idea, http://www.mersenneforum.org/showthread.php?p=481628#post481628. Just posted a question there, maybe you can help?
Forum: Factoring 2018-03-04, 20:29
Replies: 74
Views: 13,361
Posted By Till
Performance comparison

All methods were tested with 6 threads.
Sorry I forgot to repeat that point.
Forum: Factoring 2018-03-04, 18:20
Replies: 74
Views: 13,361
Posted By Till
PSIQS 4.0 release

Hi all,
today I released PSIQS 4.0, which can be found here: http://www.tilman-neumann.de/psiqs.html

It is now licensed under GPL3, and my dearest thanks go once again to Dario alpern to permit...
Forum: Factoring 2017-09-24, 13:20
Replies: 74
Views: 13,361
Posted By Till
New PSIQS release

Today I released PSIQS 3.0, which you find here:
http://www.tilman-neumann.de/psiqs.html

In terms of asymptotic SIQS speed there was no big improvement, just 5% or so. But I added a few new...
Forum: Factoring 2017-02-06, 15:28
Replies: 74
Views: 13,361
Posted By Till
I didn't try a bucket sieve, because I believe...

I didn't try a bucket sieve, because I believe that it is inefficient in Java. The reason is that the bucket sieve is also designed to optimize cache access, in combination with a segmented sieve for...
Forum: Factoring 2017-02-06, 11:58
Replies: 74
Views: 13,361
Posted By Till
Performance comparison

Hi all,
for those who like to see some numbers, here is a new comparison between PSIQS 2.1, Dario Alperns Siqs, and Yafu 1.34.5, 64bit Windows version.
All programs were run with 6 threads.

The...
Forum: Factoring 2017-02-05, 17:07
Replies: 74
Views: 13,361
Posted By Till
New minor release 2.1

Shortly after the 2.0 release I found another (P)SIQS performance improvement:
The ainvp computation is much faster now, exploiting that (1/a) mod p = (1/(a%p) mod p

Other improvements:
* The...
Forum: Factoring 2017-01-31, 13:17
Replies: 74
Views: 13,361
Posted By Till
Version 02

Hi all,
I am happy to be able to inform you that version 02 of the PSIQS package has been released.

It is about factor 3 faster than version 01 for larger numbers (300 bit and more). The memory...
Forum: Factoring 2016-12-09, 14:36
Replies: 74
Views: 13,361
Posted By Till
Thanks for the explanation! I agree that...

Thanks for the explanation!

I agree that English is very flexible, and I feel that it is very hard for a non-native to reach a level where you don't expose your non-nativeness too quickly. One...
Forum: Factoring 2016-12-08, 19:39
Replies: 74
Views: 13,361
Posted By Till
My english... What I wanted to say is that I'll...

My english... What I wanted to say is that I'll be looking in more often than before. I suppose the right expression would've been "on alert" ?
Forum: Factoring 2016-12-08, 17:23
Replies: 74
Views: 13,361
Posted By Till
Hi, I was used to get an email on replies, but...

Hi,
I was used to get an email on replies, but this time I missed it. I'm sorry too ;-)
Nevertheless I'm happy you like my work. And soon, there'll be an update.
If you have some particular...
Forum: Factoring 2016-01-20, 06:34
Replies: 74
Views: 13,361
Posted By Till
Part II: Smooth even multipliers

G&W start their study of multipliers on p.28 from the spot with multipliers being products of distinct small odd primes. I suppose that other multipliers are not considered because they escaped their...
Showing results 1 to 25 of 44

 
All times are UTC. The time now is 10:01.


Tue Dec 7 10:01:08 UTC 2021 up 137 days, 4:30, 0 users, load averages: 1.13, 1.28, 1.36

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.