reasons for so few factorers

Printable View

01-29-2005, 05:51 PM
Keroberts1

reasons for so few factorers

I believe the reason thqt so few people use the factoring is because of the difficulty of the interface. If it were to have a easy to use interface such as download the program, click the icon, select a range and go then many moer people would be participating. Also since I don't believe the link on the co-ordination thread gives out an up to dat .dat file or even the newest version of prime 95's factoring code many people shy away from this sub-project. I would surely be doing some factoring on my P4s if not for these complications.
01-29-2005, 06:30 PM
hc_grove

Re: reasons for so few factorers

Quote:

Originally posted by Keroberts1
I believe the reason thqt so few people use the factoring is because of the difficulty of the interface. If it were to have a easy to use interface such as download the program, click the icon, select a range and go then many moer people would be participating.

Well, what's easy to user depends on the user, I'm pretty glad I don't have to deal with a GUI. Having said that, I much preferred the interface of sbfactor for the interface om prime95 (mprime on LInux), and if all the algorithmic improvements are in Georges math routines, I think it' will be quite simple to link those against sbfactor, if we could get the object files.

Actually I have a P4, that could do some work, but it runs Debian Woody (uograding is not an option) which has a glibc 2.2 and mprime is build agianst 2.3, and I don't want to bother with two diffentent interfaces for factoring. (Of course this could be solved by a version of mprime build against glibc 2.2).

Quote:

Also since I don't believe the link on the co-ordination thread gives out an up to dat .dat file

Maybe not, but it has a link to an up to date factors.txt, nd as long as you check that (sbafctor did and make_worktodo.pl does) no work will be wasted.

Quote:

or even the newest version of prime 95's factoring code

That's a real problem, you have to dig through a couple of threads here to find it, and when you find something you have to check carefully that it is the newest version. Last time I checked the newest version (which had a fix for tests with n's above 9M, which I would have liked) were only available for windows.
:swear:

So my point is: the interface does matter (but there none that fits all), and easy access to the newest version matters too.
01-29-2005, 08:38 PM
Frodo42

Quote:

Actually I have a P4, that could do some work, but it runs Debian Woody (uograding is not an option) which has a glibc 2.2 and mprime is build agianst 2.3, and I don't want to bother with two diffentent interfaces for factoring. (Of course this could be solved by a version of mprime build against glibc 2.2).

Isn't the sprime binary with the Prime95-factorer staticly compiled?
Both binaries (sprime and mprime) work very well on my two P4's, but they also have all up-to-date glibc.

And I strongly second hc_grove's wish for a new version of the prime95 linux factorer

Let the windows machines bother with the gui ... I also like working without and am pretty satiesfied with the way mprime works ... especially that it keeps running and don't stop between ranges and that I can just put a number into worktodo.ini that I would want factored in the middle of a range.

Let's also get some links to the newest factorers somewhere where they are easy to find.
01-30-2005, 06:28 AM
Troodon

I think the GUI lack isn't a problem (proth_sieve is command line and we have a lot of sievers) - I prefer command line programs. I would like to see a new sbfactor version with all the latest improvements as it seems to me easier to use for mass-factoring than primer95 (maybe that's because I'm used to sbfactor).
Also, I think there are the following problems with factoring (compared to sieving):
1) It usually takes more to get a factor.
2) You have to play with more parameters.
3) It requires more memory.
4) It requieres more user attention, as factoring is always very close to PRPing and if you don't immediately send the factor, there are many chances it will come too late. Plus you have to frecuently update the data files.
5) It gets less attention and less marketing.
01-30-2005, 07:08 AM
hc_grove

Quote:

Originally posted by Frodo42
Isn't the sprime binary with the Prime95-factorer staticly compiled?

I totally missed that, thanks for pointing that out. Now I'll see if I can make that machine do something useful.

Quote:

I also like working without and am pretty satiesfied with the way mprime works ... especially that it keeps running and don't stop between ranges and that I can just put a number into worktodo.ini that I would want factored in the middle of a range.

Well worktodo.ini is a nice feature, but I liked being able to set B1 and B2 directly, being able to set the available amount of memory on the command-line and having the factors found in their own file.

I guess you avoid a stop between ranges by putting the candidates of a new range into worktodo.ini before the old range is done, but then how do you notice when a range is done?
01-30-2005, 09:34 AM
Frodo42

Quote:

Originally posted by hc_grove
Well worktodo.ini is a nice feature, but I liked being able to set B1 and B2 directly, being able to set the available amount of memory on the command-line and having the factors found in their own file.

I guess you avoid a stop between ranges by putting the candidates of a new range into worktodo.ini before the old range is done, but then how do you notice when a range is done?

I usally just look at the output ... no harm is done if the report get in a bit late ... it's more important to get the factors in before prp "hits".

I also liked being able to set B1 and B2 manually.
Maybe somehow one could combine the two version ... I don't know I'm not the programmer here.
I would like to be able to set B1 and B2 in worktodo.ini.
Another nice feature of worktodo.ini is that I can put in a factor that I just recieved for prp'ing and test it before going on with the range I have going now.
Yet another thing I like about prime95 is that it saves progress very nice when it's killed "softly".
01-30-2005, 03:30 PM
hc_grove

Quote:

Originally posted by Frodo42

Maybe somehow one could combine the two version ... I don't know I'm not the programmer here.

Normally I don't see my self as a programmer either, but as I did make some changes to sbfactor, you know that I do know something about programming.

I probably would try implementing support for a worktodo.ini in sbfactor, but unlike last year when I hacked on sbfactor, I have a job now, and as a consequence at lot less time for that kind of stuff.
01-30-2005, 03:37 PM
Frodo42

blasted jobs ;)

Anyhow apreciate your effort hc_grove ... I'll adjust to whatever Linux P-1 factorer that is fastest out there ... so far your scripts do very well to keep mprime running smoothly ... maybe we should just collect them somewhere.