Nice progress!

[pinhodecarlos]

Soon I'll be joining you guys so where should I point my cores, 98 digits?

[EdH]

Originally Posted by pinhodecarlos

Soon I'll be joining you guys so where should I point my cores, 98 digits?

If you run yoyo's script as written, it gets the composites from the db for you. It grabs a random one from the lowest available (above a minimum, usually set to 70 digits).

[Jeff Gilchrist]

I guess we spoke too soon, there is a pile of composites in the 70+ digits now...

[pinhodecarlos]

Originally Posted by pinhodecarlos

Soon I'll be joining you guys so where should I point my cores, 98 digits?

Joined.

I'm running yoyo's script but what's better, use factor or siqs? Just finished my first siqs on a 98 digits and got P66*P32.

[RichD]

Originally Posted by pinhodecarlos

I'm running yoyo's script but what's better, use factor or siqs? ...

I can think of three DB worker processes off the top of my head.

- A perl script which you might be using.
- Another perl script. <one of my favorites at times>
- A BOINC project that works in the larger range.

To each his own. :-)

[pinhodecarlos]

Originally Posted by RichD

I can think of three DB worker processes off the top of my head.

- A perl script which you might be using.
- Another perl script. <one of my favorites at times>
- A BOINC project that works in the larger range.

To each his own. :-)

I'm talking about yafu flags. Using factor it starts with some ecm curves and at 98 digits probably it will be faster than using siqs.

[RobertS]

Originally Posted by pinhodecarlos

Joined.

I'm running yoyo's script but what's better, use factor or siqs?.

The numbers are not ecm'ed, or only partially. So you should use factor.

Originally Posted by RichD

I can think of three DB worker processes off the top of my head.

- A perl script which you might be using.
- Another perl script. <one of my favorites at times>
- A BOINC project that works in the larger range.

I recommend using the second perl script (the original DB helper). Here the status of the number is checked in the DB directly when you start to factor the current number.
While Yoyos script gets a list of 100 or even 1000 unfactored numbers in the beginning, factors number by number without checking whether this number is already done or not. So if you start factoring the second number of your list it's probably done, the 100'th is definitely done. (At least in the low digit range, where many people are working.)

But in the second (DB helper 1.4.1) script there is a mistake:
line 76:
if ( $1 != "C" ) {
should be
if ( $1 ne "C" ) {

[yoyo]

Originally Posted by RobertS

I recommend using the second perl script (the original DB helper). Here the status of the number is checked in the DB directly when you start to factor the current number.
While Yoyos script gets a list of 100 or even 1000 unfactored numbers in the beginning, factors number by number without checking whether this number is already done or not.

There you are wrong. My script fetches exatly one composite and factors it and reports it back and than fetches the next.
yoyo

[RobertS]

Originally Posted by yoyo

There you are wrong. My script fetches exatly one composite and factors it and reports it back and than fetches the next.
yoyo

Yes, you are right! I probably had a different script in my mind.

[RichD]

Originally Posted by pinhodecarlos

I'm talking about yafu flags. Using factor it starts with some ecm curves and at 98 digits probably it will be faster than using siqs.

If you are using your own (or modified) script, then factor is better. It performs some ECM and then lets yafu decide what to use after that, sigs or nfs. One may also consider "-plan deep" as a switch.

[Jeff Gilchrist]

Originally Posted by pinhodecarlos

I'm talking about yafu flags. Using factor it starts with some ecm curves and at 98 digits probably it will be faster than using siqs.

In my experience using factor() has been beneficial because a fair amount of composites were found early with P+1/P-1/ecm before the siqs() stage started.