Home My Page Projects cado-nfs
Summary Activity Forums Tracker Lists Tasks Docs News SCM Files

[#21311] The MurphyEs of two polynomials for RSA-155 give wrong indication about which one is the best.

Date:
2017-04-21 07:38
Priority:
5
State:
Open
Submitted by:
Pierrick Gaudry (gaudry)
Assigned to:
Nobody (None)
Hardware:
none
Product:
none
Operating System:
none
Component:
none
Version:
none
Severity:
none
Resolution:
none
URL:
Summary:
The MurphyEs of two polynomials for RSA-155 give wrong indication about which one is the best.

Detailed description
The two attached polynomial files (produced by the 2.2 and by the 2.3 candidates) have the following MurphyE values:
rsa155-2.2.poly:
# MurphyE (Bf=3.09e+07,Bg=1.72e+07,area=4.15e+15) = 8.8e-11
rsa155-2.3.poly:
# MurphyE (Bf=3.09e+07,Bg=1.72e+07,area=4.15e+15) = 9.51e-11
But, during sieving the first one is measurably better than the second one.
Sieving parameters are the default ones.
Message  ↓
Date: 2017-04-24 15:48
Sender: Paul Zimmermann

more precisely with the following command we get (with revision 439f889f2):

/tmp/cado.d6m7qdnq/client/download/las -I 14 -poly /tmp/cado.d6m7qdnq/client/download/c155.poly -q0 30940618 -q1 30950000 -lim0 17246818 -lim1 30940618 -lpb0 29 -lpb1 29 -mfb0 62 -mfb1 61 -lambda0 2.24 -lambda1 2.2 -ncurves0 13 -ncurves1 23 -fb /tmp/cado.d6m7qdnq/client/download/c155.roots.gz -out /tmp/cado.d6m7qdnq/client/localhost.work/c155.30940618-30950000.gz -t 1 -stats-stderr

With the 2.2 polynomial:
# Average J=7595 for 542 special-q's, max bucket fill 0.625827
# Discarded 0 special-q's out of 542 pushed
# Total cpu time 1080.35s [norm 2.65+9.3, sieving 905.5 (519.2 + 70.5 + 315.7), factor 163.0 (161.7 + 1.3)]
# Total elapsed time 1100.49s, per special-q 2.03042s, per relation 0.0715346s
# PeakMemusage (MB) = 851
# Total 15384 reports [0.0702s/r, 28.4r/sq]

With the 2.3 polynomial:
# Average J=7783 for 550 special-q's, max bucket fill 0.626654
# Discarded 0 special-q's out of 550 pushed
# Total cpu time 1115.28s [norm 2.65+11.1, sieving 931.9 (532.6 + 72.4 + 326.9), factor 169.6 (168.4 + 1.2)]
# Total elapsed time 1136.19s, per special-q 2.0658s, per relation 0.0760197s
# PeakMemusage (MB) = 848
# Total 14946 reports [0.0746s/r, 27.2r/sq]

We see the number of relations per special-q is better for the 2.2 polynomial.

Attachments:
Size Name Date By Download
700 bytesrsa155-2.2.poly2017-04-21 07:38gaudryrsa155-2.2.poly
688 bytesrsa155-2.3.poly2017-04-21 07:38gaudryrsa155-2.3.poly
Field Old Value Date By
File Added6037: rsa155-2.2.poly2017-04-21 07:38gaudry
File Added6038: rsa155-2.3.poly2017-04-21 07:38gaudry