msieve hangs up when MPQSing a c79?

[Andi47]

msieve (Win32 binary) was started from an UBasic script by creating a worktodo.ini file and starting the doscomand "msieve.exe".

The logfile reads:

Fri Jan 16 16:55:28 2009  
Fri Jan 16 16:55:28 2009  
Fri Jan 16 16:55:28 2009  Msieve v. 1.39
Fri Jan 16 16:55:28 2009  random seeds: ff0cddbc 284d6690
Fri Jan 16 16:55:28 2009  factoring 1984969981693391894358019559704813808103157758211906517455430577734283035099891 (79 digits)
Fri Jan 16 16:55:28 2009  searching for 15-digit factors
Fri Jan 16 16:55:30 2009  commencing quadratic sieve (79-digit input)
Fri Jan 16 16:55:30 2009  using multiplier of 11
Fri Jan 16 16:55:30 2009  using 64kb Pentium 4 sieve core
Fri Jan 16 16:55:30 2009  sieve interval: 6 blocks of size 65536
Fri Jan 16 16:55:30 2009  processing polynomials in batches of 17
Fri Jan 16 16:55:30 2009  using a sieve bound of 1106029 (43235 primes)
Fri Jan 16 16:55:30 2009  using large prime bound of 110602900 (26 bits)
Fri Jan 16 16:55:30 2009  using trial factoring cutoff of 27 bits
Fri Jan 16 16:55:30 2009  polynomial 'A' values have 10 factors

At this point the cpu went idle, while msieve.exe was still present (but using 0% of cpu time) when I looked in the windows task manager. (My system: Pentium 4 with Windows XP pro)

[MatWur-S530113]

Hello,

I run this number with msieve v1.38 without problems:

 
Fri Jan 16 17:07:36 2009  Msieve v. 1.38
Fri Jan 16 17:07:36 2009  random seeds: a6aed848 a65264b6
Fri Jan 16 17:07:36 2009  factoring 1984969981693391894358019559704813808103157758211906517455430577734283035099891 (79 digits)
Fri Jan 16 17:07:36 2009  searching for 15-digit factors
Fri Jan 16 17:07:37 2009  commencing quadratic sieve (79-digit input)
Fri Jan 16 17:07:37 2009  using multiplier of 11
Fri Jan 16 17:07:37 2009  using 32kb Intel Core sieve core
Fri Jan 16 17:07:37 2009  sieve interval: 12 blocks of size 32768
Fri Jan 16 17:07:37 2009  processing polynomials in batches of 17
Fri Jan 16 17:07:37 2009  using a sieve bound of 1106029 (43235 primes)
Fri Jan 16 17:07:37 2009  using large prime bound of 110602900 (26 bits)
Fri Jan 16 17:07:37 2009  using trial factoring cutoff of 27 bits
Fri Jan 16 17:07:37 2009  polynomial 'A' values have 10 factors
Fri Jan 16 17:14:40 2009  43331 relations (22261 full + 21070 combined from 234871 partial), need 43331
Fri Jan 16 17:14:40 2009  begin with 257132 relations
Fri Jan 16 17:14:40 2009  reduce to 61693 relations in 2 passes
Fri Jan 16 17:14:40 2009  attempting to read 61693 relations
Fri Jan 16 17:14:41 2009  recovered 61693 relations
Fri Jan 16 17:14:41 2009  recovered 50381 polynomials
Fri Jan 16 17:14:41 2009  attempting to build 43331 cycles
Fri Jan 16 17:14:41 2009  found 43331 cycles in 1 passes
Fri Jan 16 17:14:41 2009  distribution of cycle lengths:
Fri Jan 16 17:14:41 2009     length 1 : 22261
Fri Jan 16 17:14:41 2009     length 2 : 21070
Fri Jan 16 17:14:41 2009  largest cycle: 2 relations
Fri Jan 16 17:14:41 2009  matrix is 43235 x 43331 (5.6 MB) with weight 1301016 (30.03/col)
Fri Jan 16 17:14:41 2009  sparse part has weight 1301016 (30.03/col)
Fri Jan 16 17:14:41 2009  filtering completed in 3 passes
Fri Jan 16 17:14:41 2009  matrix is 30793 x 30857 (4.4 MB) with weight 1029927 (33.38/col)
Fri Jan 16 17:14:41 2009  sparse part has weight 1029927 (33.38/col)
Fri Jan 16 17:14:41 2009  saving the first 48 matrix rows for later
Fri Jan 16 17:14:41 2009  matrix is 30745 x 30857 (2.6 MB) with weight 736447 (23.87/col)
Fri Jan 16 17:14:41 2009  sparse part has weight 489056 (15.85/col)
Fri Jan 16 17:14:41 2009  matrix includes 64 packed rows
Fri Jan 16 17:14:41 2009  using block size 12342 for processor cache size 4096 kB
Fri Jan 16 17:14:41 2009  commencing Lanczos iteration
Fri Jan 16 17:14:41 2009  memory use: 3.2 MB
Fri Jan 16 17:14:44 2009  lanczos halted after 487 iterations (dim = 30743)
Fri Jan 16 17:14:44 2009  recovered 16 nontrivial dependencies
Fri Jan 16 17:14:44 2009  prp24 factor: 344235009495368018409517
Fri Jan 16 17:14:44 2009  prp55 factor: 5766322212848897905873019141260276733507925310789893023
Fri Jan 16 17:14:44 2009  elapsed time 00:07:08

maybe msieve has problems if the working-directory is not the same as the calling-directory, maybe the dat-file is stored wrong...
try to use the doscommand 'cd msieve-directory' before calling msieve... maybe it helps.

Best regards,

Matthias

[Andi47]

I retried and it succeeded too. I wonder why it failed before. (btw.: working directory is the same as the calling directory, and msieve had succeeded many times before when started from this script.)

Fri Jan 16 16:58:46 2009  
Fri Jan 16 16:58:46 2009  
Fri Jan 16 16:58:46 2009  Msieve v. 1.39
Fri Jan 16 16:58:46 2009  random seeds: 9ecb288c 3469ba3f
Fri Jan 16 16:58:46 2009  factoring 1984969981693391894358019559704813808103157758211906517455430577734283035099891 (79 digits)
Fri Jan 16 16:58:47 2009  searching for 15-digit factors
Fri Jan 16 16:58:48 2009  commencing quadratic sieve (79-digit input)
Fri Jan 16 16:58:49 2009  using multiplier of 11
Fri Jan 16 16:58:49 2009  using 64kb Pentium 4 sieve core
Fri Jan 16 16:58:49 2009  sieve interval: 6 blocks of size 65536
Fri Jan 16 16:58:49 2009  processing polynomials in batches of 17
Fri Jan 16 16:58:49 2009  using a sieve bound of 1106029 (43235 primes)
Fri Jan 16 16:58:49 2009  using large prime bound of 110602900 (26 bits)
Fri Jan 16 16:58:49 2009  using trial factoring cutoff of 27 bits
Fri Jan 16 16:58:49 2009  polynomial 'A' values have 10 factors
Fri Jan 16 17:11:46 2009  43345 relations (22338 full + 21007 combined from 234309 partial), need 43331
Fri Jan 16 17:11:46 2009  begin with 256647 relations
Fri Jan 16 17:11:46 2009  reduce to 61692 relations in 2 passes
Fri Jan 16 17:11:46 2009  attempting to read 61692 relations
Fri Jan 16 17:11:48 2009  recovered 61692 relations
Fri Jan 16 17:11:48 2009  recovered 50394 polynomials
Fri Jan 16 17:11:48 2009  attempting to build 43345 cycles
Fri Jan 16 17:11:48 2009  found 43345 cycles in 1 passes
Fri Jan 16 17:11:48 2009  distribution of cycle lengths:
Fri Jan 16 17:11:48 2009     length 1 : 22338
Fri Jan 16 17:11:48 2009     length 2 : 21007
Fri Jan 16 17:11:48 2009  largest cycle: 2 relations
Fri Jan 16 17:11:48 2009  matrix is 43235 x 43345 (5.6 MB) with weight 1300681 (30.01/col)
Fri Jan 16 17:11:48 2009  sparse part has weight 1300681 (30.01/col)
Fri Jan 16 17:11:48 2009  filtering completed in 3 passes
Fri Jan 16 17:11:48 2009  matrix is 30851 x 30914 (4.4 MB) with weight 1028939 (33.28/col)
Fri Jan 16 17:11:48 2009  sparse part has weight 1028939 (33.28/col)
Fri Jan 16 17:11:48 2009  saving the first 48 matrix rows for later
Fri Jan 16 17:11:48 2009  matrix is 30803 x 30914 (2.6 MB) with weight 732982 (23.71/col)
Fri Jan 16 17:11:48 2009  sparse part has weight 488274 (15.79/col)
Fri Jan 16 17:11:48 2009  matrix includes 64 packed rows
Fri Jan 16 17:11:48 2009  using block size 12365 for processor cache size 2048 kB
Fri Jan 16 17:11:49 2009  commencing Lanczos iteration
Fri Jan 16 17:11:49 2009  memory use: 3.2 MB
Fri Jan 16 17:11:56 2009  lanczos halted after 488 iterations (dim = 30801)
Fri Jan 16 17:11:56 2009  recovered 17 nontrivial dependencies
Fri Jan 16 17:11:56 2009  prp24 factor: 344235009495368018409517
Fri Jan 16 17:11:56 2009  prp55 factor: 5766322212848897905873019141260276733507925310789893023
Fri Jan 16 17:11:56 2009  elapsed time 00:13:10

[jasonp]

I was about to say what Matthias just said. Others have reported trouble when relying on the msieve binary being in a different directory than the 'working directory'. It's possible that ubasic's 'doscmd' function does weird things with the system path.

I also dimly remember that you could run into trouble if worktodo.ini doesn't have newlines in place.

PS: Andi, I wonder too. Perhaps a file caching issue? Nothing made it to disk the first time?

[henryzz]

i have run msieve quite a bit from the windows directory
msieve.dat is created in the directory that i am currently using
that was with a c# program using manual input so no worktodo.ini

[MatWur-S530113]

Originally Posted by Andi47

I retried and it succeeded too. I wonder why it failed before. (btw.: working directory is the same as the calling directory, and msieve had succeeded many times before when started from this script.)
snipped

ah, ok, I thought it was while developing the UBasic-script.
as Jason said, maybe the script hasn't wrote a chr$(13) at the end of input-line in worktodo.ini, this could cause an error. (GMP-ECM doesn't like this, too). But no real idea about this error. Maybe you should run the torture-test of prime95 for some time on this computer to exclude a hardware-failure.

Best regards,

Matthias

[miklin]

mik@debian:~/test_msieve$ ./msieve -v
Msieve v. 1.39
Fri Jan 16 19:45:38 2009
random seeds: d1cfb374 a3719c05
factoring 1984969981693391894358019559704813808103157758211906517455430577734283035099891 (79 digits)
searching for 15-digit factors
commencing quadratic sieve (79-digit input)
using multiplier of 11
using 32kb Intel Core sieve core
sieve interval: 12 blocks of size 32768
processing polynomials in batches of 17
using a sieve bound of 1106029 (43235 primes)
using large prime bound of 110602900 (26 bits)
using trial factoring cutoff of 27 bits
polynomial 'A' values have 10 factors

sieving in progress (press Ctrl-C to pause)
43567 relations (22568 full + 20999 combined from 233707 partial), need 43331
43567 relations (22568 full + 20999 combined from 233707 partial), need 43331
sieving complete, commencing postprocessing
begin with 256275 relations
reduce to 61821 relations in 2 passes
attempting to read 61821 relations
recovered 61821 relations
recovered 50542 polynomials
attempting to build 43567 cycles
found 43567 cycles in 1 passes
distribution of cycle lengths:
   length 1 : 22568
   length 2 : 20999
largest cycle: 2 relations
matrix is 43235 x 43567 (6.3 MB) with weight 1305257 (29.96/col)
sparse part has weight 1305257 (29.96/col)
filtering completed in 3 passes
matrix is 30708 x 30770 (4.8 MB) with weight 1023192 (33.25/col)
sparse part has weight 1023192 (33.25/col)
saving the first 48 matrix rows for later
matrix is 30660 x 30770 (3.0 MB) with weight 730103 (23.73/col)
sparse part has weight 486197 (15.80/col)
matrix includes 64 packed rows
using block size 12308 for processor cache size 4096 kB
commencing Lanczos iteration
memory use: 3.2 MB
lanczos halted after 486 iterations (dim = 30656)
recovered 14 nontrivial dependencies
prp24 factor: 344235009495368018409517
prp55 factor: 5766322212848897905873019141260276733507925310789893023
elapsed time 00:08:59
mik@debian:~/test_msieve$

[Andi47]

Originally Posted by MatWur-S530113

Maybe you should run the torture-test of prime95 for some time on this computer to exclude a hardware-failure.

After running the torture-test for 12 hours and 40 minutes, I got 0 errors and 0 warnings.

[MatWur-S530113]

hm, this sounds strange.
The good news: most probable your computer is ok
Bad news: no more ideas about this fault. (I still look at the random seed msieve choosed at your first attempt. It starts with 'ff', maybe some 1's to much...but this might be only a crazy thought).
Is it somehow possible to re-run the number with the same random seed? If yes I would do it and look what happens at my computer.

best regards,

Matthias

[jasonp]

Good idea, but I tried that and the sieving started normally.

PS: Post #1000. w00t!

[MatWur-S530113]

Originally Posted by jasonp

Good idea, but I tried that and the sieving started normally.

Maybe it only was a choke of his computer

Originally Posted by jasonp

PS: Post #1000. w00t!

Congratulations! Awaiting msieve 1.1000 then

best regards,

Matthias