Msieve QS estimates

[bsquared]

Originally Posted by mataje

Done.

May I ask: what OS and computer specs are you using to do these?

[mataje]

Originally Posted by bsquared

May I ask: what OS and computer specs are you using to do these?

Windows Vista 32 bits and Q6600.

[henryzz]

Originally Posted by bsquared

Excel.

Create a chart, from chart menu select add trendline, create a new exponential fit, and don't forget to select "display equation on chart" under the options tab. I have some more data for C60 < N < C100. I'll post later tonight.

It might not work well if you mix all the data though, because msieve only turns on DLP at 85 digits in size, and the relation accumlation rate (and final full/partial ratio) changes quite a bit at that point.

- ben.

i tried open office but it didnt do very well
i will now try excel on another pc

i realize that and i will do two separate equations

edit:
excel did a better job of <85 but my data is too sparse higher up and it fits the early points and then goes off course further on
excel used the formula format you used
i will glean some more data from my msieve logs today

[henryzz]

Originally Posted by mataje

Done.

thanks
i also have a question
why did you not use gnfs
i could factor a 108 digit number <3 hours using four cores of a Q6600 with gnfs you took several days

[mataje]

Originally Posted by henryzz

thanks
i also have a question
why did you not use gnfs
i could factor a 108 digit number <3 hours using four cores of a Q6600 with gnfs you took several days

Yes, now I´m using GGNFS, but using one core and factlat for a 108 digits it takes me about 10-11 hours.
I´m a newbie, can you give me some help to optimize it?

[henryzz]

Originally Posted by mataje

Yes, now I´m using GGNFS, but using one core and factlat for a 108 digits it takes me about 10-11 hours.
I´m a newbie, can you give me some help to optimize it?

i get <3 hours using factmsieve.pl which has multicore support
that time of yours doesnt look bad factlat.pl has reasonably good parameters
i have uploaded factmsieve.pl so you can do multicore factorizations
it also uses msieve for the postprocessing which is quite a bit faster and uses lots less memory

[mataje]

Thank you henryzz.
I will try factmsieve.pl.

[bsquared]

Here are some more results, C60 through C90, all with msieve-1.38

[henryzz]

Originally Posted by bsquared

Here are some more results, C60 through C90, all with msieve-1.38

thanks

[10metreh]

A C91 for the collection (done in two sessions, both normal priority, on a P4@1.7GHz):

Fri Jan 09 18:49:10 2009  Msieve v. 1.39
Fri Jan 09 18:49:10 2009  random seeds: 12a40d2c e088543e
Fri Jan 09 18:49:10 2009  factoring 1689963259509811912183158401036141439705912908146000729452376919726086631046494327326628521 (91 digits)

<sieve params, end of session 1 snipped>
 
Sat Jan 10 09:05:55 2009  63160 relations (17909 full + 45251 combined from 677855 partial), need 62449

<postprocessing snipped>

The total time was 2:28:18.

[henryzz]

Originally Posted by 10metreh

A C91 for the collection (done in two sessions, both normal priority, on a P4@1.7GHz):

Fri Jan 09 18:49:10 2009  Msieve v. 1.39
Fri Jan 09 18:49:10 2009  random seeds: 12a40d2c e088543e
Fri Jan 09 18:49:10 2009  factoring 1689963259509811912183158401036141439705912908146000729452376919726086631046494327326628521 (91 digits)

<sieve params, end of session 1 snipped>
 
Sat Jan 10 09:05:55 2009  63160 relations (17909 full + 45251 combined from 677855 partial), need 62449

<postprocessing snipped>

The total time was 2:28:18.

i am actually collecting more info than is neccessary for just this experiment
could you post the full log