Bernoulli(200) c204

[jrk]

Originally Posted by henryzz

What's the difference between the different versions of lasieve?
4e, 5e, 5f, 5g?

One difference is that the f siever allows composite special_Q with prime factors greater than 256 while the others use only prime special_Q. This increases the number of lattices sieved over a given range of Q values.

[jrk]

Over the weekend I did a trial run of Shi Bai's 3.428e-15 polynomial using lasieve4I16e. Although Greg is using the f variant of the siever now which allows composite special_Q to be sieved (thus affecting total yield), I thought the results might still be interesting to some.

3.428e-15 polynomial from Shi Bai tested with lasieve4I16e and the following parameters:
rlim: 400000000
alim: 400000000
lpbr: 32
lpba: 33
mfbr: 64
mfba: 96
rlambda: 2.6
alambda: 3.6

start Q     rels in   Q roots in   rels per   Q roots in   est. rels in   est. cumulative
            Q..Q+2k   Q..Q+2k      Q root     Q..Q+50M     Q..Q+50M       rels
150000000      3359           86      39.06     2637012        1.03E+08          1.03E+08
200000000      4295          113      38.01     2603558        9.90E+07          2.02E+08
250000000      4383          117      37.46     2575156        9.65E+07          2.98E+08
300000000      4606          126      36.56     2550438        9.32E+07          3.92E+08
350000000      3421           99      34.56     2535727        8.76E+07          4.79E+08
400000000      2945           88      33.47     2514663        8.42E+07          5.63E+08
450000000      3286          100      32.86     2504409        8.23E+07          6.46E+08
500000000      3637          118      30.82     2492256        7.68E+07          7.23E+08
550000000      2778           91      30.53     2481334        7.57E+07          7.98E+08
600000000      3154          105      30.04     2467955        7.41E+07          8.72E+08
650000000      3186          109      29.23     2461102        7.19E+07          9.44E+08
700000000      3120          110      28.36     2451406        6.95E+07          1.01E+09
750000000      3068          109      28.15     2444230        6.88E+07          1.08E+09
800000000      2027           72      28.15     2437199        6.86E+07          1.15E+09
850000000      2210           84      26.31     2429513        6.39E+07          1.22E+09
900000000      2478           93      26.65     2421418        6.45E+07          1.28E+09

Assuming about 900M raw rels would be desired (judging by the number of rels gathered for 7^374+1 c197), this suggests a range of special_Q from 150M to about 700M using lasieve4I16e.

[Batalov]

My mistake. I've been testing degree 6 for too long, and for it I played with 3LP on the other side (for some polynomials, both variants, but for this batch these were used):

lpbr: 33
lpba: 32
mfbr: 96
mfba: 64
alambda: 2.6
rlambda: 3.6
alim: 400000000
rlim: 400000000

Anyway, Greg's poly has 3LP on the a side and 33:

lpbr: 33
lpba: 33
mfbr: 67
mfba: 96
rlambda: 2.6
alambda: 3.6

[frmky]

Originally Posted by jrk

One difference is that the f siever allows composite special_Q with prime factors greater than 256 while the others use only prime special_Q. This increases the number of lattices sieved over a given range of Q values.

Unfortunately there is some confusion between the e, f, & g variants of gnfs-lasieve and the applications at NFS@Home. At the time NFS@Home was established, we only had the e variant of gnfs-lasieve. In addition, the script for updating application versions would sometimes become confused if there was a number in the applications' name. Therefore, I replaced the numbers referring to the size of the sieve region with letters.

lasievea = gnfs-lasieve4I11e
lasieveb = gnfs-lasieve4I12e
lasievec = gnfs-lasieve4I13e
lasieved = gnfs-lasieve4I14e
lasievee = gnfs-lasieve4I15e
lasievef = gnfs-lasieve4I16e

The first four applications have never been used and are therefore suppressed in the app list on the website.

Bottom line...the lasievef and lasieve5f applications use gnfs-lasieve4I16e, not gnfs-lasieve4I16f.

[debrouxl]

And lasieved / gnfs-lasieve4I14e would become used if RSALS were merged into NFS@Home

[R.D. Silverman]

Originally Posted by debrouxl

And lasieved / gnfs-lasieve4I14e would become used if RSALS were merged into NFS@Home

I hope not.

[debrouxl]

Well, there are obvious advantages to having a single NFS grid
For instance, a single flavor of binaries, and a single server whose BOINC software needs to be updated once in a while.

[jrk]

Originally Posted by frmky

Bottom line...the lasievef and lasieve5f applications use gnfs-lasieve4I16e, not gnfs-lasieve4I16f.

Thanks for clearing that up.

[henryzz]

Originally Posted by frmky

Unfortunately there is some confusion between the e, f, & g variants of gnfs-lasieve and the applications at NFS@Home. At the time NFS@Home was established, we only had the e variant of gnfs-lasieve. In addition, the script for updating application versions would sometimes become confused if there was a number in the applications' name. Therefore, I replaced the numbers referring to the size of the sieve region with letters.

lasievea = gnfs-lasieve4I11e
lasieveb = gnfs-lasieve4I12e
lasievec = gnfs-lasieve4I13e
lasieved = gnfs-lasieve4I14e
lasievee = gnfs-lasieve4I15e
lasievef = gnfs-lasieve4I16e

The first four applications have never been used and are therefore suppressed in the app list on the website.

Bottom line...the lasievef and lasieve5f applications use gnfs-lasieve4I16e, not gnfs-lasieve4I16f.

So the lasieve5f application is gnfs-lasieve4I16e from lasieve5 which you have modified to input and output like ggnfs? What is the difference to the 16e siever we have been using for ages? Have you been able to compile the f or g variants(not your system of lettering) from lasieve5? Are the changes you made available in the svn or is the boinc binary the only one available?

[frmky]

Originally Posted by henryzz

So the lasieve5f application is gnfs-lasieve4I16e from lasieve5 which you have modified to input and output like ggnfs? What is the difference to the 16e siever we have been using for ages? Have you been able to compile the f or g variants(not your system of lettering) from lasieve5? Are the changes you made available in the svn or is the boinc binary the only one available?

Yep. Primary changes for users are the ability to sieve larger q and lower memory usage. Neither f nor g compile out of the box. Both fail in linking due to undefined functions. I have spent no time trying to diagnose it. I've sent the patches to Serge to place in the SVN, but I don't think he's had time to do so. So I'll place them here.

The compile currently only works on Linux. Make sure cweb is installed on your computer. Then, download the original source and extract it. Then from the file attached, apply the patch lasieve5_ggnfs.patch and compile with
make gnfs-lasieve4I16e
and you should have a GGNFS compatible binary. If you wish to create a BOINC binary (which you generally wouldn't) then also apply the patch lasieve5_boinc.patch to add in the BOINC API.

[akruppa]

Is B200 already mostly sieved? I'm getting WUs for 2^1019-1 now, but the highest WUs for B200 I saw were about 140k.