Skip to Content.
Sympa Menu

cado-nfs - Re: [Cado-nfs-discuss] Newer Commit Takes Longer to filter

Subject: Discussion related to cado-nfs

List archive

Re: [Cado-nfs-discuss] Newer Commit Takes Longer to filter


Chronological Thread 
  • From: Emmanuel Thomé <emmanuel.thome@inria.fr>
  • To: Ed Hall <ed_ka2fwj@yahoo.com>, <cado-nfs-discuss@lists.gforge.inria.fr>
  • Subject: Re: [Cado-nfs-discuss] Newer Commit Takes Longer to filter
  • Date: Tue, 15 Sep 2020 07:36:20 +0200
  • List-archive: <http://lists.gforge.inria.fr/pipermail/cado-nfs-discuss/>
  • List-id: A discussion list for Cado-NFS <cado-nfs-discuss.lists.gforge.inria.fr>

Hi,

Can you post the full log files if you happen to still have them ? (Or a link, or by private mail if you prefer). A one-year difference entails many changes in many places. In particular, I'd like to know the size of the data set that filtering had to deal with in both cases, and how many filtering attempts were necessary.

E.


On September 15, 2020 02:54:17 Ed Hall <ed_ka2fwj@yahoo.com> wrote:

Dear Team,

I noticed a slow down with a newer commit of CADO-NFS vs. and older one.  After some runs back and forth, I believe this is due to Dup2 in the filtering process.  I'm wondering if it's possible that this is somehow connected to the occasional exit -6 troubles I posted before.  Here is a comparison between an older commit and a newer commit for two c130 composites:

===================================================
c130: 305259...71  (CADO-NFS commit 5 Sep 2019)

Generate Factor Base: Total cpu/real time for makefb: 12.51/1.31892
Generate Free Relations: Total cpu/real time for freerel: 1710.06/75.9012
Filtering - Duplicate Removal, splitting pass: Total cpu/real time for dup1: 149.51/91.2899
Filtering - Duplicate Removal, removal pass: Total cpu/real time for dup2: 779.21/150.602
Filtering - Singleton removal: Total cpu/real time for purge: 212.51/80.0519
Filtering - Merging: Total cpu/real time for merge: 463.24/356.988
Filtering - Merging: Total cpu/real time for replay: 49.6/39.403
Linear Algebra: Total cpu/real time for bwc: 22856.9/2666.63
Quadratic Characters: Total cpu/real time for characters: 72.54/14.7304
Square Root: Total cpu/real time for sqrt: 1599/225.13
Complete Factorization / Discrete logarithm: Total cpu/elapsed time for entire factorization: 292087/5553.8


c130: 144332...19  (CADO-NFS commit 22 Aug 2020)

Generate Factor Base: Total cpu/real time for makefb: 12.81/1.3224
Generate Free Relations: Total cpu/real time for freerel: 2063.21/87.8803
Filtering - Duplicate Removal, splitting pass: Total cpu/real time for dup1: 157.73/88.2725
Filtering - Duplicate Removal, removal pass: Total cpu/real time for dup2: 1718.14/1214.41
Filtering - Singleton removal: Total cpu/real time for purge: 223.15/86.8347
Filtering - Merging: Total cpu/real time for merge: 252.18/21.9253
Filtering - Merging: Total cpu/real time for replay: 49.75/40.1859
Linear Algebra: Total cpu/real time for bwc: 28400.6/3258.21
Quadratic Characters: Total cpu/real time for characters: 73.65/16.0334
Square Root: Total cpu/real time for sqrt: 3010.42/184.822
Complete Factorization / Discrete logarithm: Total cpu/elapsed time for entire factorization: 293111/6367.65

===================================================
Here is a comparison of the same c137 composite run with both commits.  Although this time the overall time was shorter, the dup2 time seemed quite long again:

===================================================
c137: 710996...63  (CADO-NFS commit 5 Sep 2019)

Generate Factor Base: Total cpu/real time for makefb: 11.18/1.25212
Generate Free Relations: Total cpu/real time for freerel: 1706.95/75.8451
Filtering - Duplicate Removal, splitting pass: Total cpu/real time for dup1: 168.36/107.156
Filtering - Duplicate Removal, removal pass: Total cpu/real time for dup2: 880.34/161.948
Filtering - Singleton removal: Total cpu/real time for purge: 284.92/95.1331
Filtering - Merging: Total cpu/real time for merge: 640.69/514.905
Filtering - Merging: Total cpu/real time for replay: 60.93/50.7255
Linear Algebra: Total cpu/real time for bwc: 45730.7/5365
Quadratic Characters: Total cpu/real time for characters: 84.82/17.6833
Square Root: Total cpu/real time for sqrt: 2191.02/308.938
Complete Factorization / Discrete logarithm: Total cpu/elapsed time for entire factorization: 965737/13157.5


c137: 710996...63  (CADO-NFS commit 22 Aug 2020)

Generate Factor Base: Total cpu/real time for makefb: 12.6/1.36344
Generate Free Relations: Total cpu/real time for freerel: 2073.91/88.4133
Filtering - Duplicate Removal, splitting pass: Total cpu/real time for dup1: 169.22/106.052
Filtering - Duplicate Removal, removal pass: Total cpu/real time for dup2: 1998.11/1346.91
Filtering - Singleton removal: Total cpu/real time for purge: 302.02/103.426
Filtering - Merging: Total cpu/real time for merge: 334.62/26.8521
Filtering - Merging: Total cpu/real time for replay: 63.76/52.2779
Linear Algebra: Total cpu/real time for bwc: 47545.8/5544.35
Quadratic Characters: Total cpu/real time for characters: 84.73/16.9589
Square Root: Total cpu/real time for sqrt: 3561.49/210.118
Complete Factorization / Discrete logarithm: Total cpu/elapsed time for entire factorization: 994830/12298.8

===================================================
The commits shown were for the server and all the clients when running the 2020 commit.  For the earlier runs, the commit shown (2019) was for the server and many of the clients, but some clients were running an even earlier commit.

Thanks for all your work.

Sincerely,
Edwin Hall

_______________________________________________
Cado-nfs-discuss mailing list




Archive powered by MHonArc 2.6.19+.

Top of Page