Python Driver for GGNFS and MSIEVE - Page 30

Brian Gladman · 2010-03-17, 21:12

Quote:

Originally Posted by sleigher

Tested that and it worked. Thanks.

I cannot switch to the python script now after running the perl script for this long though can I? Are they interchangeable to the point that I could sop my perl job and start where I left off with a python job?

I'm afraid not - the way they record restart information is different.

Brian

scalabis · 2010-03-18, 16:21

Quote:

Originally Posted by Brian Gladman

Here is an updated version that will (I hope) remove some of these errors. Thanks to all for the bug reports.

Brian

Hi!

For my 1st post at this Forum , i just hope i'm not writing something stupid.

It goes like this:

I used your updated version, and tried to test run your script twice, at 2 different Command prompts at the same time.

I used this command on the first:

..\factmsieve.py test 1 2

And, on the second Command prompt:

..\factmsieve.py test 2 2

After a while, i noticed on the first Command Prompt this info:

-> Lattice sieving algebraic q from 450000 to 460000.

The values started at 31000 to 320000 and were increasing ...

On the second Command Prompt this info:

-> Lattice sieving algebraic q from 300000 to 310000.

And these values never change.

A few minutes later the 1st command prompt showed:

Found 4141493 relations, 258.8% of the estimated minimum (1600000).

Msieve is invoked, but it doest work ...

A few seconds latter the process starts again on the 1st Command prompt, with sometinhg like:

-> Lattice sieving algebraic q from 460000 to 470000

What am i doing wrong ?

Thanks,
scalabis

scalabis · 2010-03-18, 21:51

Hi again!

I've been looking around here:

Quote:

# For multiple clients, the q search space is divided
# into major blocks of length n_clients * fact_p['qstep'] so
# that major block i starts and ends at:
#
# QSTART + i * n_clients * fact_p['qstep']
# QSTART + (i + 1) * n_clients * fact_p['qstep'].
#
# Within each such major block, client k sieves the
# (k - 1)'th fact_p['qstep'] block. It then proceeds to it's
# fact_p['qstep'] block within the next major block

k0 = (fact_p['q0'] - fact_p['qstart']) // (n_clients * fact_p['qstep'])
q0 = fact_p['qstart'] + (k0 * n_clients +
((client - 1) % n_clients)) * fact_p['qstep']

But it is not easy to spot something on such a large script ...

I also uninstalled Python 3.5.1, and tried 2.6.4

Got the same result ...

Any ideas ?

Thanks,
scalabis

Brian Gladman · 2010-03-18, 22:07

I am not the best person to answer on the use of the script to drive paralllel invocations of the underlying tools. However if I were to try this, I would be inclined to drive the two versions from different working directories as I am unsure of the interactions that will occur when two versions are run in the same working directory.

scalabis · 2010-03-18, 22:14

Hi again!

First of all thanks for replying.

And let me add, thank you, for all your work.

I have already tried using two different working directories.

But the result is the same.

The "q" range on client 2 of 2, starts before the range on client 1 of 2.

After this range is exausted, client 1 of 2 goes on, and client 2 of 2, stays the same ...

Any ideas ?

Thanks,
scalabis

Batalov · 2010-03-18, 22:58

Point of order:
this thread is about msieve-GPU. There's a separate thread about the python script - people read/write and exchange ideas there. While it is ok-ish (but still inconvenient) for those who read all threads, it would be better if you posted in the appropriate thread.
I've opened this thread half a dozen times already expecting to find news about the GPU, only to find this sub-thread unrolling and unrolling and unrolling...

Brian Gladman · 2010-03-19, 08:27

Quote:

Originally Posted by scalabis

Hi again!

First of all thanks for replying.

And let me add, thank you, for all your work.

I have already tried using two different working directories.

It is a bug - my apologies and my thanks for your excellent report, which gave exactly the information I needed to find it.

A revised version is attached.

Brian

tmorrow · 2010-03-19, 08:45

Brian did you deliberately revert the changes from #288: prepending ./ for unix systems in the msieve invocation? If so, I'm curious to know what the problem with it was.

Brian Gladman · 2010-03-19, 09:09

No, it wasn't deliberate. I am posting from a different machine and I didn't realise that I had a different version loaded here. I'll update the master version on my site later today when I start my development machine.

scalabis · 2010-03-19, 16:24

Quote:

Originally Posted by Brian Gladman

It is a bug - my apologies and my thanks for your excellent report, which gave exactly the information I needed to find it.

A revised version is attached.

Brian

Hi!

Sorry to inform that the problem persists.

Using two Command prompts on different directories:

1st screen:

q from 300000 to 310000 and then repeats que q range on and on.

2nd screen (client 2 of 2) ok:

q from 310000 to 320000 then

q from 330000 to 340000 and so on ...

Regards,
scalabis

Brian Gladman · 2010-03-19, 18:37

Here is another try.

Brian

2010-03-19, 18:37	#330
Brian Gladman May 2008 Worcester, United Kingdom 1000010100₂ Posts	Here is another try. Brian Last fiddled with by jasonp on 2010-07-22 at 03:31 Reason: remove old attachment

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Msieve & ggnfs on MacOS	xilman	Msieve	8	2017-05-20 00:12
Factorizing with MSIEVE, GGNFS & Factmsieve.py	Romuald	Msieve	24	2015-11-09 20:16
Infinite loop for ggnfs or msieve	Greebley	Aliquot Sequences	4	2013-02-06 19:28
Error running GGNFS+msieve+factmsieve.py	D. B. Staple	Factoring	6	2011-06-12 22:23
A new driver? (or type of driver?)	10metreh	Aliquot Sequences	3	2010-02-15 15:57

2010-03-18, 22:07	#323
Brian Gladman May 2008 Worcester, United Kingdom 2²×7×19 Posts	I am not the best person to answer on the use of the script to drive paralllel invocations of the underlying tools. However if I were to try this, I would be inclined to drive the two versions from different working directories as I am unsure of the interactions that will occur when two versions are run in the same working directory.

2010-03-18, 22:14	#324
scalabis Mar 2010 5² Posts	Hi again! First of all thanks for replying. And let me add, thank you, for all your work. I have already tried using two different working directories. But the result is the same. The "q" range on client 2 of 2, starts before the range on client 1 of 2. After this range is exausted, client 1 of 2 goes on, and client 2 of 2, stays the same ... Any ideas ? Thanks, scalabis

2010-03-18, 22:58	#325
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 3⁶×13 Posts	Point of order: this thread is about msieve-GPU. There's a separate thread about the python script - people read/write and exchange ideas there. While it is ok-ish (but still inconvenient) for those who read all threads, it would be better if you posted in the appropriate thread. I've opened this thread half a dozen times already expecting to find news about the GPU, only to find this sub-thread unrolling and unrolling and unrolling...

2010-03-19, 08:45	#327
tmorrow Jan 2004 147₈ Posts	Brian did you deliberately revert the changes from #288: prepending ./ for unix systems in the msieve invocation? If so, I'm curious to know what the problem with it was.

2010-03-19, 09:09	#328
Brian Gladman May 2008 Worcester, United Kingdom 2²×7×19 Posts	No, it wasn't deliberate. I am posting from a different machine and I didn't realise that I had a different version loaded here. I'll update the master version on my site later today when I start my development machine.