Discussion related to cado-nfs

[hermann@stamm-wilbrandt.de]

> Which command is now needed to complete the factorization of RSA-155?
Found it, the sqrt command completed the factorization already, see 
below.

> Where in source tree is the orchestration of the tasks done?
This question is still open.

Regards,

Hermann.



hermann@7950x:~/cado-nfs$ cat /tmp/cado.10b02n9l/c155.sqrt.stdout.2
Failed
Failed
Failed
106603488380168454820927220360012878679207958575989291522270608237193062808643
102639592829741105772054196573991675900716567808038066803341933521790711307779
hermann@7950x:~/cado-nfs$

hermann@7950x:~/RSA_numbers_factored/pari$ gp -q RSA_numbers_factored.gp
? [,n]=RSA.get(155);
? 
p=106603488380168454820927220360012878679207958575989291522270608237193062808643;
? 
q=102639592829741105772054196573991675900716567808038066803341933521790711307779;
? n==p*q
1
?

-------- Original Message --------
Subject: Re: [cado-nfs] problems factoring 155-digit numbers, need help
Date: 2025-07-28 11:47
From: hermann@stamm-wilbrandt.de
To: Paul Zimmermann <Paul.Zimmermann@inria.fr>
Cc: cado-nfs@inria.fr

Hi Paul,

On 2025-07-28 10:03, Paul Zimmermann wrote:
> Hi Hermann,
>
> the fact there is no error message looks suspicious. Maybe there is not 
> enough
> memory on your machine.
it was an oversight by me, the error message appeared further up in the 
log, see below.

> However the 3rd run failed in polynomial selection, which should not 
> take much
> memory.
The error message below shows error is in Square Root.

I started pfree with nohup to watch memory every second:

hermann@7950x:~$ cat /usr/local/bin/pfree
#!/bin/bash
while true
do
free -m | grep Mem: | awk '{print ($3/$2)*100}'
sleep 1
done
hermann@7950x:~$

The I started command as before with 8 threads, and indeed it gets 
killed for ooM:

hermann@7950x:~$ sort -n nohup.out | tail -5
89.7394
89.9317
90.3773
90.3997
94.8392
hermann@7950x:~$


Then I tried with "-t 4" with watching memory, and this time it 
successfully completed:

hermann@7950x:~/cado-nfs$ /home/hermann/cado-nfs/build/7950x/sqrt/sqrt 
-poly /tmp/cado.10b02n9l/c155.poly -prefix 
/tmp/cado.10b02n9l/c155.dep.gz -dep 0 -t 4 -side0 -side1 -gcd > 
/tmp/cado.10b02n9l/c155.sqrt.stdout.2 2> 
/tmp/cado.10b02n9l/c155.sqrt.stderr.2
hermann@7950x:~/cado-nfs$

With maximal memory use of below 95% of 32GB RAM:

hermann@7950x:~$ sort -n nohup.out | uniq -c | tail -5
      1 89.7394
      1 89.9317
      1 90.3773
      1 90.3997
      1 94.8392
hermann@7950x:~$


Which command is now needed to complete the factorization of RSA-155?

Where in source tree is the orchestration of the tasks done?

Regards,

Hermann.


...
^[[32;1mInfo^[[0m:Linear Algebra: Mksol: CPU time 66735.65,  WCT time 
4979.24, iteration CPU time 0.06, COMM 0.01, cpu-wait 0.01, comm-wait 
0.0 (64000 iterations)
^[[32;1mInfo^[[0m:Quadratic Characters: Starting
^[[32;1mInfo^[[0m:Complete Factorization / Discrete logarithm: Quadratic 
Characters
^[[32;1mInfo^[[0m:Quadratic Characters: Total cpu/real time for 
characters: 87.25/19.3318
^[[32;1mInfo^[[0m:Square Root: Starting
^[[32;1mInfo^[[0m:Square Root: Creating file of (a,b) values
^[[33;1mWarning^[[0m:Command: Process with PID 653804 finished with 
return code -9
^[[31;1mError^[[0m:Square Root: Program run on server failed with exit 
code -9
^[[31;1mError^[[0m:Square Root: Command line was: 
/home/hermann/cado-nfs/build/7950x/sqrt/sqrt -poly 
/tmp/cado.10b02n9l/c155.poly -prefix /tmp/cado.10b02n9l/c155.dep.gz -dep 
0 -t 8 -side0 -side1 -gcd > /tmp/cado.10b02n9l/c155.sqrt.stdout.2 2> 
/tmp/cado.10b02n9l/c155.sqrt.stderr.2
^[[31;1mError^[[0m:Square Root: Stderr output (last 10 lines only) 
follows (stored in file /tmp/cado.10b02n9l/c155.sqrt.stderr.2):
^[[31;1mError^[[0m:Square Root:         Alg(6): starting level 3 at 
cpu=1771.24s (wct=80.44s), 32 values to multiply
^[[31;1mError^[[0m:Square Root:         Alg(7): level 2 took cpu=206.33s 
(wct=7.07s)
^[[31;1mError^[[0m:Square Root:         Alg(7): starting level 3 at 
cpu=1777.15s (wct=80.66s), 32 values to multiply
^[[31;1mError^[[0m:Square Root:         Alg(5): level 2 took cpu=210.05s 
(wct=7.21s)
^[[31;1mError^[[0m:Square Root:         Alg(5): starting level 3 at 
cpu=1780.95s (wct=80.80s), 32 values to multiply
^[[31;1mError^[[0m:Square Root:         Alg(0): level 2 took cpu=245.47s 
(wct=8.48s)
^[[31;1mError^[[0m:Square Root:         Alg(0): starting level 3 at 
cpu=1814.61s (wct=82.01s), 32 values to multiply
^[[31;1mError^[[0m:Square Root:         Alg(1): level 3 took cpu=216.96s 
(wct=8.14s)
^[[31;1mError^[[0m:Square Root:         Alg(1): starting level 4 at 
cpu=1910.44s (wct=85.64s), 16 values to multiply
^[[31;1mError^[[0m:Square Root:
^[[32;1mInfo^[[0m:API server: Got notification to stop serving Workunits
^[[32;1mInfo^[[0m:Lattice Sieving: Aggregate statistics:
...

> The 2nd run failed in the square root, which takes a lot of memory. I 
> see you
> used -t 8, i.e., 8 threads. You should try with only one thread at a 
> time
> (tasks.sqrt.threads=1).
>
> Hope this helps,
> Paul
>
> > Date: Mon, 28 Jul 2025 08:50:37 +0200
> > From: hermann@stamm-wilbrandt.de
> >
> > Two years ago I did run into an ARM processor cado-nfs problem and got
> > help from Paul.
> >
> > End of last week I wanted to factor a C155 number and after more than
> > 27h execution stopped with an error. I used cado-nfs code from two 
> > years
> > ago, with only new parameter file:
> > https://github.com/cado-nfs/cado-nfs/blob/master/parameters/factor/params.c155
> >
> > Then I cloned the current repo somewhere else, built and tried to 
> > factor
> > RSA-155 this time. After more than 25h that run stopped with an error.
> >
> > I had started both with "nohup ./cado-nfs.py ...", and have the 
> > complete
> > nohup log output for both. Unfortunately I did shutdown the 
> > workstation
> > then and lost the /tmp/cado... data.
> >
> >
> > Then I found the old thread where Paul provided the command to try to
> > restart computation, and that /tmp data is important for resume:
> > https://sympa.inria.fr/sympa/arc/cado-nfs/2023-07/msg00005.html
> >
> > Again, after more than 25h of a third run (on RSA-155 again) 
> > computation
> > on stopped with error.
> > The first thing I did is to recursively copy "/tmp/cado.10b02n9l" to a
> > save place.
> >
> > So I have the complete nohup output for all three runs, but only the
> > /tmp/cado... data for the last.
> >
> > Workstation is 32GB RAM AMD 16C/32T 7950X CPU.
> >
> > 1st failure happened here:
> >
> > ...
> > ^[[32;1mInfo^[[0m:Linear Algebra: Lingen CPU time 1454.04, WCT time
> > 79.63
> > ^[[32;1mInfo^[[0m:Linear Algebra: Mksol: CPU time 68420.99,  WCT time
> > 5087.03, iteration CPU time 0.06, COMM 0.01, cpu-wait 0.01, comm-wait
> > 0.0 (64000 iterations)
> > ^[[32;1mInfo^[[0m:Quadratic Characters: Total cpu/real time for
> > characters: 88.51/18.3812
> > ^[[32;1mInfo^[[0m:HTTP server: Shutting down HTTP server
> > Traceback (most recent call last):
> >    File "/home/hermann/cado-nfs/./cado-nfs.py", line 202, in <module>
> >      factors = factorjob.run()
> > ...
> >
> >
> > 2nd failure happended here:
> >
> > ...
> > ^[[32;1mInfo^[[0m:Linear Algebra: Aggregate statistics:
> > ^[[32;1mInfo^[[0m:Linear Algebra: Krylov: CPU time 123334.55, WCT time
> > 9481.7, iteration CPU time 0.06, COMM 0.01, cpu-wait 0.01, comm-wait 
> > 0.0
> > (128000 iterations)
> > ^[[32;1mInfo^[[0m:Linear Algebra: Lingen CPU time 432.32, WCT time 
> > 44.08
> > ^[[32;1mInfo^[[0m:Linear Algebra: Mksol: CPU time 67470.41,  WCT time
> > 5092.29, iteration CPU time 0.06, COMM 0.01, cpu-wait 0.01, comm-wait
> > 0.0 (64000 iterations)
> > ^[[32;1mInfo^[[0m:Quadratic Characters: Starting
> > ^[[32;1mInfo^[[0m:Complete Factorization / Discrete logarithm: 
> > Quadratic
> > Characters
> > ^[[32;1mInfo^[[0m:Quadratic Characters: Total cpu/real time for
> > characters: 91.36/18.5425
> > ^[[32;1mInfo^[[0m:Square Root: Starting
> > ^[[32;1mInfo^[[0m:Square Root: Creating file of (a,b) values
> > ^[[33;1mWarning^[[0m:Command: Process with PID 1294735 finished with
> > return code -9
> > ^[[31;1mError^[[0m:Square Root: Program run on server failed with exit
> > code -9
> > ^[[31;1mError^[[0m:Square Root: Command line was:
> > /home/hermann/cado-nfs/build/7950x/sqrt/sqrt -poly
> > /tmp/cado.e4bk0it9/c155.poly -prefix /tmp/cado.e4bk0it9/c155.dep.gz 
> > -dep
> > 0 -t 8 -side0 -side1 -gcd > /tmp/cado.e4bk0it9/c155.sqrt.stdout.2 2>
> > /tmp/cado.e4bk0it9/c155.sqrt.stderr.2
> > ^[[31;1mError^[[0m:Square Root: Stderr output (last 10 lines only)
> > follows (stored in file /tmp/cado.e4bk0it9/c155.sqrt.stderr.2):
> > ^[[31;1mError^[[0m:Square Root:         Alg(7): starting level 3 at
> > cpu=1824.80s (wct=81.60s), 32 values to multiply
> > ...
> >
> >
> > 3rd failure happened here:
> >
> > hermann@7950x:~/cado-nfs$ nohup ./cado-nfs.py
> > 10941738641570527421809707322040357612003732945449205990913842131476349984288934784717997257891267332497625752899781833797076537244027146743531593354333897
> > ...
> > ^[[32;1mInfo^[[0m:Linear Algebra: Aggregate statistics:
> > ^[[32;1mInfo^[[0m:Linear Algebra: Krylov: CPU time 124201.66, WCT time
> > 9341.07, iteration CPU time 0.06, COMM 0.01, cpu-wait 0.01, comm-wait
> > 0.0 (128000 iterations)
> > ^[[32;1mInfo^[[0m:Linear Algebra: Lingen CPU time 428.58, WCT time 
> > 41.28
> > ^[[32;1mInfo^[[0m:Linear Algebra: Mksol: CPU time 66735.65,  WCT time
> > 4979.24, iteration CPU time 0.06, COMM 0.01, cpu-wait 0.01, comm-wait
> > 0.0 (64000 iterations)
> > ^[[32;1mInfo^[[0m:Generate Factor Base: Total cpu/real time for 
> > makefb:
> > 10.18/0.735045
> > ^[[32;1mInfo^[[0m:Filtering - Duplicate Removal, removal pass: Total
> > cpu/real time for dup2: 4142.98/1411.62
> > ^[[32;1mInfo^[[0m:Filtering - Duplicate Removal, removal pass: 
> > Aggregate
> > statistics:
> > ^[[32;1mInfo^[[0m:Filtering - Duplicate Removal, removal pass: CPU 
> > time
> > for dup2: 1318.3s
> > ^[[32;1mInfo^[[0m:Polynomial Selection (root optimized): Aggregate
> > statistics:
> > ^[[32;1mInfo^[[0m:Polynomial Selection (root optimized): Total time:
> > 4745.81
> > ^[[32;1mInfo^[[0m:Polynomial Selection (root optimized): Rootsieve 
> > time:
> > 4803.13
> > ^[[32;1mInfo^[[0m:API server: Shutting down server on exception 
> > Program
> > failed
> > Traceback (most recent call last):
> >    File "/home/hermann/cado-nfs/./cado-nfs.py", line 210, in <module>
> >      factors = factorjob.run()
> >                ^^^^^^^^^^^^^^^
> > ...
> >
> >
> > No command got logged for last as for 2nd run.
> > Which command can I try to resume the computation with intact
> > /tmp/cado... files?
> >
> > Or is 32GB RAM not enough to process RSA-155?
> >
> >
> > Regards,
> >
> > Hermann.