mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Operation Kibibit

Reply
 
Thread Tools
Old 2011-04-22, 10:43   #12
xilman
Bamboozled!
 
xilman's Avatar
 
"π’‰Ίπ’ŒŒπ’‡·π’†·π’€­"
May 2003
Down not across

34·53 Posts
Default

Quote:
Originally Posted by jasonp View Post
The advantage to RSA768 over all the other similarly large numbers is that we have a known high-quality polynomial to compare against, the result of many machine-decades of computation that nobody has been able to beat (yet) even with more powerful algorithms. It doesn't matter to me at all that RSA768 is finished, since I wouldn't try to actually proceed with the sieving even if the target was unfactored.
Okey-doke. Now running on a C1060

Range is 100k at around 19M

Paul

Last fiddled with by xilman on 2011-04-22 at 11:13 Reason: Add range info.
xilman is offline   Reply With Quote
Old 2011-04-22, 11:39   #13
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

25·149 Posts
Default

Quote:
Originally Posted by jasonp View Post
Yes, the resource use of the GPU code is pretty noticeable, and I'll understand if you want to switch to the CPU version instead. I'll take anything between 10^4 and 10^9.
No, no, I will proceed with CUDA on my range (845000-900000), just a bit slower.

Luigi
ET_ is offline   Reply With Quote
Old 2011-04-22, 13:49   #14
debrouxl
 
debrouxl's Avatar
 
Sep 2009

11110100012 Posts
Default

I'll try 30K-40K.
debrouxl is offline   Reply With Quote
Old 2011-04-22, 17:59   #15
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

23×7×163 Posts
Default

Quote:
Originally Posted by jasonp View Post
Yes, the resource use of the GPU code is pretty noticeable, and I'll understand if you want to switch to the CPU version instead. I'll take anything between 10^4 and 10^9.
J- What attrition rate in the 2nd stage would you expect for these?
I've tried the currently accumulated ~1900 candidates (in c6~=24000 vicinity) and -np2 very quickly returned nothing. -S
Batalov is offline   Reply With Quote
Old 2011-04-22, 18:16   #16
xilman
Bamboozled!
 
xilman's Avatar
 
"π’‰Ίπ’ŒŒπ’‡·π’†·π’€­"
May 2003
Down not across

34·53 Posts
Default

Quote:
Originally Posted by jasonp View Post
The advantage to RSA768 over all the other similarly large numbers is that we have a known high-quality polynomial to compare against, the result of many machine-decades of computation that nobody has been able to beat (yet) even with more powerful algorithms. It doesn't matter to me at all that RSA768 is finished, since I wouldn't try to actually proceed with the sieving even if the target was unfactored.
I accept your arguments and I'm contributing some computrons to the effort. However, even if you've no intention of sieving an alternative, the result of this computation may be of use to others.

Paul
xilman is offline   Reply With Quote
Old 2011-04-22, 19:56   #17
bdodson
 
bdodson's Avatar
 
Jun 2005
lehigh.edu

100000000002 Posts
Default

Quote:
Originally Posted by xilman View Post
I accept your arguments and I'm contributing some computrons to the effort. However, even if you've no intention of sieving an alternative, the result of this computation may be of use to others.

Paul
Perhaps if/when we have nontrivial improvements over the known RSA768
poly, we might switch to RSA232, or some other likely candidate. I expect
to switch our tesla's over as soon as my Primegrid tasks clear. -Bruce
bdodson is offline   Reply With Quote
Old 2011-04-22, 20:31   #18
jrk
 
jrk's Avatar
 
May 2008

21058 Posts
Default

Quote:
Originally Posted by Batalov View Post
J- What attrition rate in the 2nd stage would you expect for these?
I've tried the currently accumulated ~1900 candidates (in c6~=24000 vicinity) and -np2 very quickly returned nothing. -S
If you're going to run -np2, you should use the latest SVN revision. If you use 1.48, you'll eventually get stuck in an infinite loop in stage2 which was fixed. I think this is the area that Jason is aiming to improve by doing these tests.

Last fiddled with by jrk on 2011-04-22 at 20:31
jrk is offline   Reply With Quote
Old 2011-04-23, 01:59   #19
bdodson
 
bdodson's Avatar
 
Jun 2005
lehigh.edu

210 Posts
Default

Quote:
Originally Posted by bdodson View Post
Perhaps if/when we have nontrivial improvements over the known RSA768
poly, we might switch to RSA232, or some other likely candidate. I expect
to switch our tesla's over as soon as my Primegrid tasks clear. -Bruce
Doesn't appear to be running like I expected. I tried
Code:
 1039  (./msieve_gpu_1.48 -g 0 -l msieveg0o.log -v -np1 2000001,3000000 ) > se0a.err &
 1040  (./msieve_gpu_1.48 -g 1 -l msieveg1o.log -v -np1 10000001,11000000 ) > se1a.err &
 1047  (./msieve_cpu_1.48 -v -l msieve.dat1 -nf msieve.fb1 -s msieve.dat1 -np1 10000001,11000000 ) > sc1a.err &
 1051  (./msieve_cpu_1.48 -v -l msieve.dat0 -nf msieve.fb0 -s msieve.dat0 -np1 2000001,3000000 ) > sc0a.err &
 1054  (./msieve_cpu_1.48 -v -l msieve.dat2 -nf msieve.fb2 -s msieve.dat2 -np1 11000001,12000000 ) > sc2a.err &
 1055  (./msieve_cpu_1.48 -v -l msieve.dat3 -nf msieve.fb3 -s msieve.dat3 -np1 3000001,4000000 ) > sc3a.err &
 1063  (./msieve_cpu_1.48 -v -l msieve.dat4 -nf msieve.fb4 -s msieve.dat4 -np1 12000001,13000000 ) > sc4a.err &
 1064  (./msieve_cpu_1.48 -v -l msieve.dat5 -nf msieve.fb5 -s msieve.dat5 -np1 4000001,5000000 ) > sc5a.err &
to run 1 task each on the two tesla cards, and one task each on the six
cpus. I was expecting to see hardly any cputime on the two gpu tasks,
and c. 100% on the cpu tasks. But instead, I'm seeing ... Nooo!

Geez! Seven tasks on card 0, One on card 1; no cpu tasks ... as in
Code:
 more msieve.dat3
Fri Apr 22 21:48:01 2011  
Fri Apr 22 21:48:01 2011  
Fri Apr 22 21:48:01 2011  Msieve v. 1.48
Fri Apr 22 21:48:01 2011  random seeds: bc36a0ad 34b8141e
Fri Apr 22 21:48:01 2011  factoring 1230186684530117755130494958384962720772853569595334792197322452151726
4005072636575187452021997864693899564749427740638459251925573263034537315482685079170261221429134616704292
14311602221240479274737794080665351419597459856902143413 (232 digits)
Fri Apr 22 21:48:02 2011  no P-1/P+1/ECM available, skipping
Fri Apr 22 21:48:02 2011  commencing number field sieve (232-digit input)
Fri Apr 22 21:48:02 2011  commencing number field sieve polynomial selection
Fri Apr 22 21:48:02 2011  searching leading coefficients from 3000001 to 4000000
Fri Apr 22 21:48:02 2011  using GPU 0 (Tesla C2050)
for each of dat0, dat1, ..., dat5 --- which were supposed to be running
on cpus. Killed 'um, just two ranges running --- what was I supposed to do
to get tasks on the cpus?
Regards, Bruce

PS --- Glah. Silly me, this binary was compiled to submit gpu jobs; I need
to recompile (instead of just copying the gpu binary onto one _named_ cpu_1.48).
I'll fix in the morning.

Last fiddled with by bdodson on 2011-04-23 at 02:29 Reason: wishful thinking, where recompiling was needed, sigh
bdodson is offline   Reply With Quote
Old 2011-04-23, 02:16   #20
Batalov
 
Batalov's Avatar
 
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2

23·7·163 Posts
Default

I've changed my oneliners to
Code:
set X=1
msieve -v -l msieve.log$X -nf msieve.fb$X -s msieve.dat$X -np1 $START,$END
-l is for the log. It was a copy-paste error evidently...
Batalov is offline   Reply With Quote
Old 2011-04-23, 02:29   #21
jrk
 
jrk's Avatar
 
May 2008

1,093 Posts
Default

Quote:
Originally Posted by bdodson View Post
for each of dat0, dat1, ..., dat5 --- which were supposed to be running
on cpus. Killed 'um, just two ranges running --- what was I supposed to do
to get tasks on the cpus?
Regards, Bruce
Make two msieve binaries, one with CUDA=1 and one without.
jrk is offline   Reply With Quote
Old 2011-04-23, 13:34   #22
bdodson
 
bdodson's Avatar
 
Jun 2005
lehigh.edu

210 Posts
Default

Quote:
Originally Posted by jrk View Post
Make two msieve binaries, one with CUDA=1 and one without.
Better, thanks. My trunk_file from january seemed to insist on looking
for gpu files (or something with "gpu"); so I ended up picking up a new
copy. So the GPU's are running 1.48, while the CPUs are running 1.49.
Here's how the current outfiles look
Code:
  1050 Apr 23 09:26 msieve.dat5b.m
  3220 Apr 23 09:26 msieve.dat0b.m
  3409 Apr 23 09:26 msieve.dat4b.m
  3767 Apr 23 09:26 msieve.dat1b.m
 19162 Apr 23 09:26 msieve.dat.m
  4830 Apr 23 09:26 msieve.dat3b.m
  5619 Apr 23 09:26 msieve.dat2b.m
with both cards writing to msieve.dat.m and
Code:
wc -l msieve.dat.m

272 msieve.dat.m
from 12 hours, and
Code:
    02:24:53 ./msieve_gpu_1.48 -g 0 -l msieveg0o.log -v -np1 2000001,3000000
    03:06:34 ./msieve_gpu_1.48 -g 1 -l msieveg1o.log -v -np1 10000001,11000000
for the cputime used by the cards/ 12hrs. -Bruce
bdodson is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Polynomial selection Max0526 NFS@Home 9 2017-05-20 08:57
Improved NFS polynomial selection jasonp Operation Kibibit 5 2014-09-07 11:02
Call for volunteers: RSA896 jasonp Operation Kibibit 134 2013-09-03 22:08
2^877-1 polynomial selection fivemack Factoring 47 2009-06-16 00:24
Polynomial selection CRGreathouse Factoring 2 2009-05-25 07:55

All times are UTC. The time now is 04:16.

Tue Oct 27 04:16:21 UTC 2020 up 47 days, 1:27, 0 users, load averages: 1.87, 1.82, 1.81

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.