mersenneforum.org > YAFU YAFU 2.0
 Register FAQ Search Today's Posts Mark Forums Read

2021-05-03, 03:16   #78
bsquared

"Ben"
Feb 2007

3·1,193 Posts

Quote:
 Originally Posted by bsquared No, those symbols should be in libmsieve, the linker just can't find the library to link them in. I don't know anything else to try. You just need to make sure that a path to the location where libmsieve.a exists is in the LIBS variable of the makefile. You could try an absolute path.
Oh, and you also need a -lmsieve in there somewhere too.

LIBS += -L../msieve -lmsieve

 2021-05-03, 17:23 #79 bur     Aug 2020 79*6581e-4;3*2539e-3 2×11×19 Posts Thank you.... !!! :D That was it. Why does it seem to work for almost everyone without that but not for me? Or did I delete it at some point?
2021-05-03, 17:37   #80
bsquared

"Ben"
Feb 2007

3·1,193 Posts

Quote:
 Originally Posted by bur Thank you.... !!! :D That was it. Why does it seem to work for almost everyone without that but not for me? Or did I delete it at some point?
It was deleted back when I told you that you could simplify the ifeq (\$(NFS),1) block back in post 54. you can see it there at the end of the original NFS block in that post. Just been overlooking it since then.

 2021-05-03, 17:55 #81 EdH     "Ed Hall" Dec 2009 Adirondack Mtns 3·372 Posts I have gotten around to starting a new "How I. . ." thread specifically for YAFU2. It's based on Makefile modification rather than command line massaging, and it's meant only as a starting point to getting it up and running. Sorry, it's a little late. Any comments are quite welcome.
 2021-05-03, 18:13 #82 bur     Aug 2020 79*6581e-4;3*2539e-3 2·11·19 Posts As a general question, how often do you guys build a new version of yafu or cado? Should I just check the repository for the changelog?
 2021-05-04, 23:10 #83 BudgieJane     "Jane Sullivan" Jan 2011 Beckenham, UK 1000101112 Posts Well, I've just got it to work, which is quite something for me! I persuaded it to do Code: nfs(29^165 - 23^165) which has 240 decimal digits, and it only took a few hours to come up with Code: ***factors found*** P1 = 2 P1 = 3 P1 = 3 P1 = 7 P2 = 31 P2 = 31 P2 = 67 P2 = 97 P3 = 199 P3 = 331 P4 = 2441 P4 = 3301 P37 = 3126480058316999071399277955199677901 C125 = 12724988150963378912700523668030562454652746526621625001448059255431251645943099394682519998040029913891319413189363203274486 P25 = 5021659044902186258192281 P18 = 102397541093560351 P39 = 965650242131801601223534344035881987891 ans = 0 which I am quite satisfied with as I already had the factors for the C125 cofactor.
 2021-05-05, 12:46 #84 BudgieJane     "Jane Sullivan" Jan 2011 Beckenham, UK 4278 Posts Things missing in output from Version 2.0 Following on from yesterday's post, I have now started using yafu2.0 in my application. What gets run is the following (example): Code: ..\..\Maths\yafu\Versions\yafu-2.0\yafu-x64 factor(1093398016684859039437357097655002878473578346491832162155479546307786424669365857546034634142930096044072838434811201) -forceDLP -v -v -v -noecm -silent -R -logfile io\log12.txt -of io\out12.txt The first thing I noticed is the version number and the host name do not appear in the log file, whereas they did in previous versions. These may be important for figuring out which version was used when comparing runs between future versions. The second thing I noticed is that whereas the -logfile was created, the -of (output file) was not. This is the way I pass the output from yafu to my application, as it is much easier to parse this than to try to parse the logfile. Last fiddled with by BudgieJane on 2021-05-05 at 13:18 Reason: Adding host name to the first thing
2021-05-05, 12:51   #85
LaurV
Romulan Interpreter

"name field"
Jun 2011
Thailand

2·4,903 Posts

Quote:
 Originally Posted by BudgieJane I noticed is the version number does not appear in the log file, whereas it did in previous versions. This may be important for figuring out which version was used when comparing runs between future versions.
+1. Needed. We did such comparisons many times in the past.

2021-05-05, 13:28   #86
bsquared

"Ben"
Feb 2007

1101111110112 Posts

Quote:
 Originally Posted by LaurV +1. Needed. We did such comparisons many times in the past.
I guess I didn't realize that was useful to folks, sorry about that. I removed it, well, actually because the code reorg made it harder to implement, but I told myself it was because it wasn't useful. I will look into putting it back.

As for the -of output, can confirm that it isn't working. That was not tested at all and apparently I broke it somehow. I'll get it fixed.

Also, Jane, this one had a small-ish factor (found by ecm):

Code:
found factor 24616664227663950740856773434561 in stage 2 in thread 9, vec position 4, with sigma = 3667020861

found factor 24616664227663950740856773434561 in stage 2 in thread 35, vec position 7, with sigma = 544625459

found factor 24616664227663950740856773434561 in stage 2 in thread 40, vec position 5, with sigma = 3261427245
ecm: 768/912 curves on C118 @ B1=1000000, B2=100*B1
ecm: process took 21.6734 seconds.

ecm: found prp32 factor = 24616664227663950740856773434561

ecm: found prp86 (co)factor = 44416985444197989346245849525036029769673491288448206702392870361399028860459766362241
Total factoring time = 40.2258 seconds

***factors found***

PRP32 = 24616664227663950740856773434561
PRP86 = 44416985444197989346245849525036029769673491288448206702392870361399028860459766362241

Last fiddled with by bsquared on 2021-05-05 at 13:29 Reason: ecm details

2021-05-05, 14:39   #87
BudgieJane

"Jane Sullivan"
Jan 2011
Beckenham, UK

32×31 Posts

I can't remember why I added the -noecm keyword in there, although I may have thought the small amount of ecm processing slowed the average factoring time down a bit. Obviously it didn't. I shall remove it.

BTW the file docfile.txt claims to be for Version 1.35-beta. Should this be changed to 2.0?

Quote:
 Originally Posted by bsquared I guess I didn't realize that was useful to folks, sorry about that. I removed it, well, actually because the code reorg made it harder to implement, but I told myself it was because it wasn't useful. I will look into putting it back.
It was appearing on every line of the logfile in the old versions. In my opinion it is only needed in there once, if that makes things any easier, although you might want to get LaurV's opinion on that, too.

Quote:
 Originally Posted by bsquared As for the -of output, can confirm that it isn't working. That was not tested at all and apparently I broke it somehow. I'll get it fixed. Also, Jane, this one had a small-ish factor (found by ecm): Code: found factor 24616664227663950740856773434561 in stage 2 in thread 9, vec position 4, with sigma = 3667020861 found factor 24616664227663950740856773434561 in stage 2 in thread 35, vec position 7, with sigma = 544625459 found factor 24616664227663950740856773434561 in stage 2 in thread 40, vec position 5, with sigma = 3261427245 ecm: 768/912 curves on C118 @ B1=1000000, B2=100*B1 ecm: process took 21.6734 seconds. ecm: found prp32 factor = 24616664227663950740856773434561 ecm: found prp86 (co)factor = 44416985444197989346245849525036029769673491288448206702392870361399028860459766362241 Total factoring time = 40.2258 seconds ***factors found*** PRP32 = 24616664227663950740856773434561 PRP86 = 44416985444197989346245849525036029769673491288448206702392870361399028860459766362241
I got the same answer (eventually!):
Code:
05/05/21 13:18:03, prp86 = 44416985444197989346245849525036029769673491288448206702392870361399028860459766362241
05/05/21 13:18:03, prp32 = 24616664227663950740856773434561
05/05/21 13:18:03, NFS elapsed time = 8640.5802 seconds.
There is one thing I have forgotten to say on all these emails, and that is THANK YOU for all the work you have put into updating yafu so that retired old codgers like me can keep my brain active in my dotage.

2021-05-05, 15:06   #88
bsquared

"Ben"
Feb 2007

3×1,193 Posts

Quote:
 Originally Posted by BudgieJane I can't remember why I added the -noecm keyword in there, although I may have thought the small amount of ecm processing slowed the average factoring time down a bit. Obviously it didn't. I shall remove it.
It depends on the inputs, of course. If you are attempting these inputs for the first time then some ecm will help a lot. If they have already been ecm'ed, by you or someone else, then more ecm might not help much. -plan custom and -pretest_ratio X can help fine tune how much time you spend on ecm.

Quote:
 Originally Posted by BudgieJane BTW the file docfile.txt claims to be for Version 1.35-beta. Should this be changed to 2.0?
Yes. Every time I contemplate updating that file I suddenly find myself doing something else. Maybe this isn't a coincidence (documentation is a chore).

Quote:
 Originally Posted by BudgieJane It was appearing on every line of the logfile in the old versions. In my opinion it is only needed in there once, if that makes things any easier, although you might want to get LaurV's opinion on that, too.
If you use session.log (active by default), then the version and build info appears in there. Date stamps can be correlated to factor.log output, should that be needed. Is that enough? Feedback welcome.

Code:
05/05/21 09:31:52, =====================================
05/05/21 09:31:52, System/Build Info:
05/05/21 09:31:52, YAFU Version 2.0
05/05/21 09:31:52, Built with Intel Compiler 1910
05/05/21 09:31:52, detected Intel(R) Xeon(R) Gold 6254 CPU @ 3.10GHz
detected L1 = 32768 bytes, L2 = 25952256 bytes, CL = 64 bytes
05/05/21 09:31:52, using 1 random witness for Rabin-Miller PRP checks
05/05/21 09:31:52, Cached 664579 primes: max prime is 9999991

05/05/21 09:31:52, Random seed: 11617470479886605489
05/05/21 09:31:52, Processing: factor(rsa(300))
05/05/21 09:32:19, Result    : 1
Quote:
 Originally Posted by BudgieJane I got the same answer (eventually!): Code: 05/05/21 13:18:03, prp86 = 44416985444197989346245849525036029769673491288448206702392870361399028860459766362241 05/05/21 13:18:03, prp32 = 24616664227663950740856773434561 05/05/21 13:18:03, NFS elapsed time = 8640.5802 seconds.
I can occasionally run on some really nice hardware so don't feel bad... 40 sec is not typical, even if some ecm had been done (I was running 768 curves in parallel)

Quote:
 Originally Posted by BudgieJane There is one thing I have forgotten to say on all these emails, and that is THANK YOU for all the work you have put into updating yafu so that retired old codgers like me can keep my brain active in my dotage.
You're welcome, glad it is finding some use and thank you for all the feedback.

 Similar Threads Thread Thread Starter Forum Replies Last Post chris2be8 YAFU 6 2019-10-17 16:22 EdH YAFU 8 2018-03-14 17:22 bsquared YAFU 119 2015-11-05 16:24 storflyt32 YAFU 2 2015-06-29 05:19 bsquared YAFU 28 2012-07-20 16:17

All times are UTC. The time now is 22:40.

Sat Nov 27 22:40:09 UTC 2021 up 127 days, 17:09, 0 users, load averages: 1.38, 1.69, 1.65