Running GGNFS

[jrk]

Oh, I could have mentioned the size of L1 Cache sooner, but I was having a brain fart and did not draw the line between the "L1_BITS" you were telling me to change and the L1 cache size... What a day.

[jrk]

Originally Posted by Batalov

If still slow, try L1_BITS 14 (which is the old default). Should be no change from old versions (except for various patches),

Actually the old rev 353 I have uses L1_BITS 15 by default, so it makes sense that I would not see a speed increase on my C2D since it is already at max.

So it appears that the benefit of rev 377 will only be for Phenom and Opteron users.

[siew]

Please forgive me for my supid question, but is it possible to generate polynomial, if i know the C154 factors?

For example, i want to collect some stats about possible polys for C154, when the p and q has similar length.

So any algo for making poly from known factors?

tanx a lot!

[Mini-Geek]

Originally Posted by siew

Please forgive me for my supid question, but is it possible to generate polynomial, if i know the C154 factors?

I don't know if there is or not, but wouldn't you get more accurate results by finding a poly as if you didn't know the factors?

[R.D. Silverman]

Originally Posted by siew

Please forgive me for my supid question, but is it possible to generate polynomial, if i know the C154 factors?

For example, i want to collect some stats about possible polys for C154, when the p and q has similar length.

So any algo for making poly from known factors?

tanx a lot!

(1) "the C154 factors". --> What C154 factors??
(2) If you know the factors, then why run the algorithm?
(3) The polynomial used in GNFS is independent of the factors. They
are not related at all.
(4) p & q having similar lengths does not affect the polynomial at all.

May I suggest actually *reading* one of the papers about constructing
the polynomial?????

[siew]

I want to study the factorization algo.
So i generated by myself the p,q,n,d,e and want to see how it works.

If you can post some links for good material, i obviously will read it.
Sorry for disturbing, but i don't want just factor the number, i want to know how it works in details. So kick me to right way, please.

tanx!

[R.D. Silverman]

Originally Posted by siew

I want to study the factorization algo.
So i generated by myself the p,q,n,d,e and want to see how it works.

If you can post some links for good material, i obviously will read it.
Sorry for disturbing, but i don't want just factor the number, i want to know how it works in details. So kick me to right way, please.

tanx!

How good is your math background????
Without some background in algebraic number theory, it will
be impossible to learn the details. Do you know what an ideal is?
Do you know what an extension field (of Q) is? To really understand
the algorithm requires a strong undergraduate degree in math with
a specialization in number theory & algebra. [or the equivalent self-study].

Start by reading Crandall & Pomerance's book: Prime Numbers, A Computational Perspective

The read the book by Lenstra et.al. Development of the Number Field
Sieve (available from Springer)

[Jeff Gilchrist]

Originally Posted by Batalov

I wonder if the similar changes (L1_BITS 16 with corrections to the code) will benefit the plain source; I guess it should. I will merge these changes now (watch for revisions 378+). Jeff, could you build and test them on small polynomials on your computer and then possibly rebuild all of them for public use? Let me know if there will be any problems.

So with the discovery of the AMD vs Intel cache sizes, what are you doing about L1_BITS, is there a way to specify that just for specific platforms or how are you doing it now?

Should I still be testing the latest SVN code or have things essentially "reverted" back to what they were before now?

Jeff.

[Batalov]

TO readers joining in from this message:
This is about the trunk source (trunk/src/lasieve4/), the above was about trunk/src/experimental/lasieve4_64/. These are parallel worlds, so try not get confused.

TO Jeff: To be on the safe side I will drop it to 15 in the trunk source.
(but it was since the dawn of time 14, except in the 16e siever, so there may be acceleration still. Could you please run some tests, old 14 vs. 15, after Win32 builds are done side-by-side; all sievers, small to big, 16e being a neutral test. If you have Win32 Athlons, try 16, too! )

Because it is a #define, it permeates all accelerated source, so there should be two builds for those who wants - AMD-type with 16 and Intel-type with 15. Combining essentially two binaries into one binary and then branching is hairy (just like jasonp's latest GPU source), but it's your call (I will not mess with vc9 stuff, you will )

[Greebley]

I was trying 374 and got the following issue. Any idea why it would occur?
This is on windows 32 bit and I was running aliqueit.exe rather than entering in any commands myself.

<several successful runs with no problems - total yield about 124k each time>

...
milliseconds total: Sieve 89279 Sched 0 medsched 33745
TD 99343 (Init 7231, MPQS 16907) Sieve-Change 99240
TD side 0: init/small/medium/large/search: 992 7620 2092 2014 30515
sieve: init/small/medium/large/search: 5431 19946 3084 2552 13160
TD side 1: init/small/medium/large/search: 337 5993 2099 2069 38358
sieve: init/small/medium/large/search: 2567 11478 3171 2268 25622
=>"c:/bin/cat.exe" spairs.out >> test.dat
Found 1154434 relations, need at least 1936352 to proceed.
-> Q0=750001, QSTEP=40000.
-> makeJobFile(): q0=750000, q1=790000.
-> makeJobFile(): Adjusted to q0=750001, q1=790000.
-> Lattice sieving algebraic q-values from q=750001 to 790000.
=> "/aliquot/ggnfs/gnfs-lasieve4I11e.exe" -k -o spairs.out -v -n0 -a test.job
Cannot open spairs.out for output: m
-> Return value 256. Updating job file and terminating...
=>"c:/bin/cat.exe" spairs.out >> spairs.add
"c:/bin/cat.exe": spairs.out: No such file or directory
-> makeJobFile(): q0=750000, q1=790000.
-> makeJobFile(): Adjusted to q0=750001, q1=790000.
Terminating...

[siew]

is correct paths was defined in .pl files?
copy cat.exe to same directory where you unpacked ggnfs.
cat.exe can be found in ported unix utils pack.
here you can find it:
http://unxutils.sourceforge.net/

also, .out files can be used by other application, so cant be opened/appended.
or maybe disk full...

check with some file monitor what happens in this moment