Fast Breeding (guru management) - Page 120

fivemack · 2018-01-09, 16:07

Quote:

Originally Posted by xilman

Thanks. I'll set up some candidates.

Added in edit: 265*(9^265+1).C232 aka GC(9,265) aka 9,265+ now added to the 15e queue. It's been years since I put anything there so may have mis-remembered the procedure. Could someone cast a critical eye over it please? If all is well I'll add 9,266+, 12,236+ and 12,238+

That looks reasonable, but it would be nice to have a sieving range - you've presumably done some trial sieving. I usually use the Y0/Y1 notation rather than m, but I think that's just an idiosyncrasy on my part and the siever will work fine.

Default is to sieve on the rational side, 'lss: 0' for the algebraic one.

RichD · 2018-01-18, 15:31

C160 from the t600 file.
( a.k.a. Phi_3(Phi_13(44611351)/225629*3919943)/3 )

Code:

n: 1645198996771698287519863000103286845983679086526137651458314721270137400493141395353109314720582263630795645925431899925091171439403445582834679495769727909407
# 70253804098533303996256039060114483059484940828633741987845788959049023444144179^3-1
Y0: -8510992183097608908062342669389
Y1: 3119774805895455331
c0: -2458879178066279636329775588966198643600
c1: 175013919598807213096374655733910
c2: 184284825279548288432637647
c3: 1077827593448971645
c4: -992173422658
c5: 73680
skew: 9223168.66838
# lognorm 50.25, E 42.78, alpha -7.47 (proj -1.69), 3 real roots
# MurphyE = 1.53414751e-12
lss: 0
type: gnfs
rlim: 80000000
alim: 80000000
lpbr: 30
lpba: 30
mfbr: 60
mfba: 60
rlambda: 2.6
alambda: 2.6

Trial sieving 5K blocks.

Code:

  Q Yield
 20M 9207
 50M 6194
 80M 4735
110M 4844

RichD · 2018-01-20, 18:48

C227 from the OPN t600 file.
( a.k.a. Phi_11(Phi_83(3)/167*12119*1036745531) )

Code:

n: 26203677826730851377939430233626534532641385178649456490321172522050584861944654221947297095921282393175735198139983340765314029246333295994370171982113084465746386028525242503793016869208931881500957725963368575548045513021493
# 950996059627210897943351^11-1, difficulty: 239.78, skewness: 1.00, alpha: 2.22
# cost: 3.76853e+18, est. time: 1794.54 GHz days (not accurate yet!)
skew: 1.000
c5: 1
c4: 1
c3: -4
c2: -3
c1: 3
c0: 1
Y1: -950996059627210897943351
Y0: 904393505426481665605349523006322663681605109202
type: snfs
rlim: 134000000
alim: 134000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.7
alambda: 2.7

Trial sieving 5K blocks.

Code:

RichD · 2018-01-22, 15:40

C160 from the OPN t600 file.
( a.k.a. Phi_3(Phi_11(Phi_3(794191)/2/3^/5/23/307/1151)/23/858001)/3/151 )

Code:

n: 2919357762182025637085736884361041528026147286452771585642240169286735537985256859236432072193315178520639434000278148158555949326472316399110665642570661007011
# 1149986550472855579648408239822856912298581122489969299470029337872846236797481461^3-1
# expecting poly E from 1.35e-12 to > 1.56e-12
lss: 0
Y0: -4722891534396009461283724343702
Y1: 5216350505029673
c0: 578637001754130307843070131474029555
c1: -23476954085250857643262788342
c2: -10969905444374948895614757
c3: -36272430452838801366
c4: 4272682082006
c5: 1242360
# skew 1539269.06, size 1.694e-15, alpha -6.538, combined = 1.549e-12 rroots = 3
skew: 1539269.06
type: gnfs
rlim: 80000000
alim: 80000000
lpbr: 30
lpba: 30
mfbr: 60
mfba: 60
rlambda: 2.6
alambda: 2.6

Trial sieving 5K blocks.

Code:

RichD · 2018-01-24, 20:21

C199 from the OPN t1000 file.
3LPs provide a better yield but the times are a little slower.
Not having to go as deep into the special-Qs may be quicker overall.
I will look closer in the future since I have several of these quartics.

Code:

n: 5262199667347704370383382573882546842844132668115527189076062581874638943044817276421916975296385935474146837843853199024201903372763704218014620151293853622962990584469731632691969324944794952060901
# 441047607640944329101719685655443319185854243052422221^5-1, difficulty: 214.58, skewness: 1.00, alpha: 1.45
# cost: 4.90392e+17, est. time: 233.52 GHz days (not accurate yet!)
skew: 1.000
c4: 1
c3: 1
c2: 1
c1: 1
c0: 1
Y1: -1
Y0: 441047607640944329101719685655443319185854243052422221
type: snfs
rlim: 67000000
alim: 67000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 93
rlambda: 2.6
alambda: 3.6

Trial sieving 5K blocks.
Perhaps starting at Q=30M.

Code:

VBCurtis · 2018-01-24, 20:48

Quote:

Originally Posted by RichD

C199 from the OPN t1000 file.
3LPs provide a better yield but the times are a little slower.
Not having to go as deep into the special-Qs may be quicker overall.
lpbr: 31
lpba: 31
mfbr: 62
mfba: 93

It is my experience with 3LP that trying mfba = 3 * lpba -2 improves speed quite a bit without hurting yield much. Maybe give mfba = 91 a try? I found 32/93 quicker than 32/95, but I didn't experiment much with 31bit 3LP.
The idea (I think) is that 93-bit cofactors are unlikely to split into 31*31*31, so we shouldn't bother trying to split them.

RichD · 2018-01-24, 22:07

Quote:

Originally Posted by VBCurtis

It is my experience with 3LP that trying mfba = 3 * lpba -2 improves speed quite a bit without hurting yield much. Maybe give mfba = 91 a try? I found 32/93 quicker than 32/95, but I didn't experiment much with 31bit 3LP.
The idea (I think) is that 93-bit cofactors are unlikely to split into 31*31*31, so we shouldn't bother trying to split them.

I was going to try mfba = 92 but it slip my mind until you mentioned the above. I will try 90 & 91 going forward. I have about 8-10 of these quartics that are ready for SNFS. It just takes a while to perform trial sieving on my antiquated MacBook Pro (C2D processor era).

wombatman · 2018-01-25, 23:41

This C156 is a current blocker for Home Primes base 10 #10319 and seems like a good candidate for 14e:

Code:

n: 829947270509843783968285132730360242554211003188405745637764693710273647973469773749403752968599120365370119962963727902520256786907451772898125027846577241
#norm 4.318399e-015 alpha -7.710025 e 2.361e-012 rroots 5
skew: 1818364.43
c0: 15404695912880251969672235365489751625
c1: 96169763706945968510973435993989
c2: -55451960683080945242365464
c3: -52672332961861307650
c4: 15731934889652
c5: 7819728
Y0: -638515978727085600665939620354
Y1: 3485497284917351

rlim: 33600000
alim: 33600000
lpbr: 29
lpba: 29
mfbr: 58
mfba: 60
rlambda: 2.6
alambda: 2.6

Test sieving yielded the following (note that the sec/rel would really be about 2x as fast--these were done on a computer carrying out numerous tasks):

Code:

q=8M: total yield: 2740, q=8002003 (0.07685 sec/rel)
q=12M: total yield: 2970, q=12002059 (0.07785 sec/rel)
q=16M: total yield: 3047, q=16002001 (0.07880 sec/rel)
q=20M: total yield: 2630, q=20002007 (0.08051 sec/rel)
q=24M: total yield: 2884, q=24002051 (0.08107 sec/rel)
q=28M: total yield: 2822, q=28002019 (0.08472 sec/rel)
q=32M: total yield: 3291, q=32002021 (0.08572 sec/rel)
q=40M: total yield: 3255, q=40002023 (0.08797 sec/rel)
q=50M: total yield: 2498, q=50002009 (0.09430 sec/rel)

Based on previous numbers, I would expect to need ~50M relations to solve, so a q-range of ~10M-45M should do it, but it's feasible to increase to 50M or 60M and drop down to 8M as needed.

I would be happy to run the linear algebra on this once it's done.

wombatman · 2018-01-26, 05:01

Shoot. Forgot to mention that this was on the a-side. Apologies.

fivemack · 2018-01-26, 11:40

I have queued wombatman's number and RichD's first three numbers; waiting for information about trial sieving with different mfb figures before queuing the C199 quartic.

chris2be8 · 2018-01-26, 16:48

Quote:

Originally Posted by RichD

C199 from the OPN t1000 file.
3LPs provide a better yield but the times are a little slower.
Not having to go as deep into the special-Qs may be quicker overall.
I will look closer in the future since I have several of these quartics.

Code:

n: 5262199667347704370383382573882546842844132668115527189076062581874638943044817276421916975296385935474146837843853199024201903372763704218014620151293853622962990584469731632691969324944794952060901
# 441047607640944329101719685655443319185854243052422221^5-1, difficulty: 214.58, skewness: 1.00, alpha: 1.45
# cost: 4.90392e+17, est. time: 233.52 GHz days (not accurate yet!)
skew: 1.000
c4: 1
c3: 1
c2: 1
c1: 1
c0: 1
Y1: -1
Y0: 441047607640944329101719685655443319185854243052422221
type: snfs
rlim: 67000000
alim: 67000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 93
rlambda: 2.6
alambda: 3.6

A C199 quartic will be harder on the rational side, so I would expect

Code:

mfbr: 92
mfba: 62
rlambda: 3.6
alambda: 2.6

to work better. At least try trial sieving it.

Chris

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
System management notes	kriesel	kriesel	7	2020-10-21 18:52
Improving the queue management.	debrouxl	NFS@Home	10	2018-05-06 21:05
Script-based Primenet assignment management	ewmayer	Software	3	2017-05-25 04:02
Do normal adults give themselves an allowance? (...to fast or not to fast - there is no question!)	jasong	jasong	35	2016-12-11 00:57
Power Management settings	PrimeCroat	Hardware	3	2004-02-17 19:11

2018-01-26, 05:01	#1318
wombatman I moo ablest echo power! May 2013 13·137 Posts	Shoot. Forgot to mention that this was on the a-side. Apologies.

2018-01-26, 11:40	#1319
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 2³·11·73 Posts	I have queued wombatman's number and RichD's first three numbers; waiting for information about trial sieving with different mfb figures before queuing the C199 quartic.