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

2010-09-13, 16:14   #419
rekcahx

Oct 2009
Oulu, Finland

111102 Posts

Bsquared. My gcc is 4.4.4. Comprehensive gcc -v output:
Code:
rekcahx@oah:~\$ gcc -v
Using built-in specs.
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Debian 4.4.4-4' --with-bugurl=file:///usr/share/doc/gcc-4.4/README.Bugs --enable-languages=c,c++,fortran,objc,obj-c++ --prefix=/usr --enable-shared --enable-multiarch --enable-linker-build-id --with-system-zlib --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --with-gxx-include-dir=/usr/include/c++/4.4 --program-suffix=-4.4 --enable-nls --enable-clocale=gnu --enable-libstdcxx-debug --enable-objc-gc --with-arch-32=i486 --with-tune=generic --enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu --target=x86_64-linux-gnu
gcc version 4.4.4 (Debian 4.4.4-4)
Relevant parts of session.log and factors.log are in compressed attachment.
Attached Files
 foo.tar.gz (10.6 KB, 46 views)

2010-09-13, 16:37   #420
bsquared

"Ben"
Feb 2007

64418 Posts

Quote:
 Originally Posted by rekcahx Relevant parts of session.log and factors.log are in compressed attachment.
Thanks, there are a few clues in there:

The lines containing "0 sieve locations out of 0", "using 0 polys (-1 A polys)", "sieving found inf rels/poly", etc, mean that this was a restart of a factorization. The significant number of errors when reading relations and the fact that there are about twice as many relations in the savefile as there should be suggest that relations were perhaps being dumped into the savefile from a different job.

The crash is apparently a symptom of the linear algebra not completing correctly (something is tried to be freed when it hasn't actually been allocated, is most likely), so I'll have to look into that, but the root cause of the error seems like save file corruption. Did you have more than one factorization going at once, somehow?

 2010-09-13, 17:05 #421 rekcahx   Oct 2009 Oulu, Finland 1E16 Posts I think that there is only one factorization process at once, but I'm not sure. I've a trivial looping script which starts yafu, wait until yafu ends and then do something with results. I've a quad core processor and I used 3 or 4 threads with yafu, but I think that this case is not thread related.
 2010-10-22, 16:58 #422 10metreh     Nov 2008 91216 Posts On a 64-bit i5, when using factor(), yafu uses QS time estimates for a 32-bit P3 to estimate the amount of ECM that needs to be done, so much more ECM is done than is necessary. Does this occur for anyone else?
2010-10-26, 17:26   #423
bsquared

"Ben"
Feb 2007

3,361 Posts

Quote:
 Originally Posted by 10metreh On a 64-bit i5, when using factor(), yafu uses QS time estimates for a 32-bit P3 to estimate the amount of ECM that needs to be done, so much more ECM is done than is necessary. Does this occur for anyone else?
This is similar to the problem over here, in that the i3/5/7 processor and its features are not properly detected.

Its been awhile since I've worked on any updates - I'll try to get one out soon.

 2010-11-04, 13:25 #424 bsquared     "Ben" Feb 2007 D2116 Posts YAFU 1.20 available ... in the usual place. New in version 1.20 + faster SIQS for windows 64bit machines - up to 20% speedup observed for a c82 + better cache and processor detection + new flag (vproc) prints TONS of cpu/cache info + tweaks to SIQS code for 1-3% speedup in core2, nehalem, and amd64 machines + enhancements to the general purpose factor routine and fermat factorization This version should address the issue 10metreh was seeing. I also just now realized that I never fixed the batch file processing code. Sorry - that will have to wait till the next release I guess. Happy factoring!
 2010-11-06, 17:21 #425 kar_bon     Mar 2006 Germany 2·1,433 Posts small factor output Using the 'factor' command with 'bigger' numbers will not print small factors. Example: Calling yafu with Code: factor(13998160709839581048153733514208588350923069415793585600447722863427118587832492939474407917725430759878249822) a 100-digit number (last index from aliquot seq. 25986) will print on screen: Code: factoring 13998160709839581048153733514208588350923069415793585600447722863427118587832492939474407917725430759878249822 div: primes less than 10000 fmt: 1000000 iterations rho: x^2 + 1, starting 1000 iterations on C104 rho: x^2 + 3, starting 1000 iterations on C104 rho: x^2 + 2, starting 1000 iterations on C104 pp1: starting B1 = 20K, B2 = 1M on C104 pp1: starting B1 = 20K, B2 = 1M on C104 pp1: starting B1 = 20K, B2 = 1M on C104 pm1: starting B1 = 100K, B2 = 5M on C104 ecm: 25 curves on C104 input, at B1 = 2K, B2 = 200K ecm: 90 curves on C104 input, at B1 = 11K, B2 = 1100K ecm: 168 curves on C104 input, at B1 = 50K, B2 = 5M and factor.log contains this: Code: 11/06/10 18:07:00 v1.20 PC, 11/06/10 18:07:00 v1.20 PC, **************************** 11/06/10 18:07:00 v1.20 PC, Starting factorization of 13998160709839581048153733514208588350923069415793585600447722863427118587832492939474407917725430759878249822 11/06/10 18:07:00 v1.20 PC, **************************** 11/06/10 18:07:00 v1.20 PC, rho: x^2 + 1, starting 1000 iterations on C104 11/06/10 18:07:01 v1.20 PC, rho: x^2 + 3, starting 1000 iterations on C104 11/06/10 18:07:01 v1.20 PC, rho: x^2 + 2, starting 1000 iterations on C104 11/06/10 18:07:01 v1.20 PC, pp1: starting B1 = 20K, B2 = 1M on C104 11/06/10 18:07:01 v1.20 PC, pp1: starting B1 = 20K, B2 = 1M on C104 11/06/10 18:07:01 v1.20 PC, pp1: starting B1 = 20K, B2 = 1M on C104 11/06/10 18:07:01 v1.20 PC, pm1: starting B1 = 100K, B2 = 5M on C104 11/06/10 18:07:03 v1.20 PC, Finished 25 curves using Lenstra ECM method on C104 input, B1 = 2000, B2 = 200000 11/06/10 18:07:19 v1.20 PC, Finished 90 curves using Lenstra ECM method on C104 input, B1 = 11000, B2 = 1100000 So for the original C110 there were small factors found and the remaining C104 is testing now. Breaking this job won't print the small factors (not on screen nor file). Better: Receiving the CTRL-C command will print the small factors found so far and the remaining Cxx.
 2010-11-23, 18:30 #426 richs     "Rich" Aug 2002 Benicia, California 4A616 Posts I was using YAFU siqs to factor a C94 and I get the following application error while reading the relations: The instruction at "0x00424ad6" referenced memory at "0x00000008". The memory could not be "written". Click OK to terminate. Click CANCEL to debug the program. I've restarted YAFI twice after saving the siqs.dat file and received the same error each time. Any advice? Thanks.... Last fiddled with by richs on 2010-11-23 at 18:31
 2010-11-23, 18:39 #427 bsquared     "Ben" Feb 2007 3,361 Posts A couple questions: Is this the latest version, 1.20? What OS are you using, linux or windows? 32 or 64 bit? If you post the relations on sendspace i'll take a look.
 2010-11-23, 19:57 #428 richs     "Rich" Aug 2002 Benicia, California 2×5×7×17 Posts I'm using the latest version 1.20 on Windows XP 32 bit. Here is the link to the siqs.dat file: http://www.sendspace.com/file/tkwgr1 The C94 is 7068706539813365378146651705387378799401977279120346480088927073809694130066011392618070305369 Thanks! Last fiddled with by richs on 2010-11-23 at 19:57
 2010-11-23, 21:34 #429 bsquared     "Ben" Feb 2007 D2116 Posts I've verified the problem, and I have a fix. This fix drastically reduces the amount of memory consumed during post-processing, which was a problem noted in a different thread. 32bit system should now be usable for larger problems and not limited by memory during post-processing. I'll try to get the fixed version posted in the next day or so.

 Similar Threads Thread Thread Starter Forum Replies Last Post 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 12 2012-11-08 04:12 bsquared YAFU 21 2012-09-04 19:44

All times are UTC. The time now is 20:27.

Thu Jan 21 20:27:17 UTC 2021 up 49 days, 16:38, 1 user, load averages: 1.31, 1.76, 2.01