Python Driver for GGNFS and MSIEVE

[Brian Gladman]

Originally Posted by Karl M Johnson

Cant launch

Traceback (most recent call last):
  File "C:\Program Files (x86)\msieve\ggnfs\example\factmsieve[N].py", line 1984, in <module>
    .format(NUM_CORES, ('' if NUM_CORES == 1 else 's'), THREADS_PER_CORE))
ValueError: unmatched '{' in format

Thanks for testing it - I hope I have corrected this now. I am away from my machine right now so I can't test it.

Brian

[jasonp]

Originally Posted by Brian Gladman

I have not uploaded it here because I worry that my frequent postings of these updates as attachments might be unwelcome from the site owner's perspective as there is no obvious way to delete the outdated ones.

My understanding is that attachments do not take up space in the forum's database, which is at a premium, but use the server's filesystem which has plenty of room. If you want me to delete old attachments I'd be happy to.

[Karl M Johnson]

Works fine.
Here's proof:

Number: fljuger
N = 10516071764181878019570915711193255841747277442931750685301436621536353931969436843 (83 digits)
Divisors found:
r1=100054404774992936538559592036032315433 (pp39)
r2=105103536299385480146611735607955433919086771 (pp45)
Version: Msieve v. 1.45
Total time: 0.19 hours.
Factorization parameters were as follows:
n: 10516071764181878019570915711193255841747277442931750685301436621536353931969436843
Y0: -68415166295548572651
Y1: 12381286097
c0: -377694361883706557375845
c1: 4090294606463989838
c2: -1338990207050
c3: 16832272
c4: 480
skew: 221556.36 
type: gnfs
Factor base limits: 550000/550000
Large primes per side: 3
Large prime bits: 24/24
Sieved algebraic special-q in [0, 0)
Total raw relations: 1080300
Relations: 48816 relations
Pruned matrix : 35196 x 35422
Polynomial selection time: 0.08 hours.
Total sieving time: 0.09 hours.
Total relation processing time: 0.01 hours.
Matrix solve time: 0.00 hours.
time per square root: 0.00 hours.
Prototype def-par.txt line would be: gnfs,82,4,56,1500,0.001,0.3,200,15,10000,500,550000,550000,24,24,40,40,1.9,1.9,10000
total time: 0.19 hours.
Intel64 Family 6 Model 15 Stepping 11, GenuineIntel
Windows-post2008Server-6.1.7600
processors: 4, speed: 3.51GHz

[Brian Gladman]

Originally Posted by jasonp

My understanding is that attachments do not take up space in the forum's database, which is at a premium, but use the server's filesystem which has plenty of room. If you want me to delete old attachments I'd be happy to.

Hi Jason,

In an ideal world I would certainly like to see them deleted as quite a number of them contain bugs that will cause problems for anyone who uses them and doesn't realise that these were being posted to obtain help with debugging.

Will deleting them cause problems for anyone who reads old postings with these attachments? If it doesn't cause problems, I would, on balance, like to see them deleted (I assume we can't rediect old links to a current version).

I think the script is pretty solid now (except for relations estimation) so I am hoping it won't need such frequent updates in future. It would be nice to improve the relations estimates but this really requires someone with the the experience to provide the formulas (or tables) this needs.

Brian

[em99010pepe]

Brian Gladman,

I don't know if this was already posted....
About your script, why not use all cores to search for the polynomial? Right now I don't have a NVIDIA GPU card and it is just a waste of time to see 3 of 4 cores idle. Thank you in advance,

Carlos

[Brian Gladman]

Originally Posted by em99010pepe

Brian Gladman,

I don't know if this was already posted....
About your script, why not use all cores to search for the polynomial? Right now I don't have a NVIDIA GPU card and it is just a waste of time to see 3 of 4 cores idle. Thank you in advance,

Carlos

Hi Carlos,

My script doesn't do any real work - it just calls the either ggnfs or msieve programs to do the hard work.

I could launch multiple copies of these applications but to do this I would have to know that they won't interfere with each other and I would also have to work out how to share the search space among the copies.

I doubt that this would be easy and it would seem better for the application to do this rather than my driver.

Brian

[em99010pepe]

Originally Posted by Brian Gladman

Hi Carlos,

My script doesn't do any real work - it just calls the either ggnfs or msieve programs to do the hard work.

I could launch multiple copies of these applications but to do this I would have to know that they won't interfere with each other and I would also have to work out how to share the search space among the copies.

I doubt that this would be easy and it would seem better for the application to do this rather than my driver.

Brian

Ok Brian I understand. I just don't know if I should occupy the rest of the cores while polynomial search is taking place because I don't know when the sieve starts. My current work has been on the polynomial phase for the last 12 hours....
Thank you.

Carlos

[wreck]

May I ask how to set the priority in factmsieve.py under Linux OS?

[miklin]

Originally Posted by wreck

May I ask how to set the priority in factmsieve.py under Linux OS?

In sense???

Conversation concerning NICE ???

# Run the binaries at low priority?
# $NICE="nice -n 19 ";

[wreck]

Originally Posted by miklin

In sense???

Conversation concerning NICE ???

# Run the binaries at low priority?
# $NICE="nice -n 19 ";

I see these two lines in file factMsieve.pl, but I use factmsieve.py cuttently,
Following lines are concerning priority I notice in the file factmsieve.py

if sys.platform.startswith('win'):
  EXE_SUFFIX = '.exe'
else:
  EXE_SUFFIX = ''
  NICE_PATH = ''
  MSIEVE = './' + MSIEVE


  if sys.platform.startswith('win'):
#   priority_high = 0x00000080
#   priority_normal = 0x00000020
#   priority_idle = 0x00000040
    al['creationflags'] = al.get('creationflags', 0) | 0x00000040
  else:
    al['preexec_fn'] = NICE_PATH

]

[EdH]

Hi Brian,

There's a post on my AliWin thread expressing a difficulty with factmsieve.py. Could you take a quick look? The basics are:

error (line 104): CUDA_ERROR_FILE_NOT_FOUND
'C:\Program' is not recognized as an internal or external command,
operable program or batch file.

Or, is this something JasonP could help with?

Thanks.

Edit: I think the trouble might be due to the use of back instead of forward slashes:

ggnfs_cmd = C:\Program Files\Python\python "C:\MathWork\aliqueit\factmsieve.py"