Fast Breeding (guru management) - Page 110

RichD · 2017-10-19, 11:18

QUEUED C180 from the OPN t600 file.
Sieve on algebraic side.

Code:

n: 446968172410122872992196388563123554199753477810992828395877112575518306619324701349248558096445378768374270606892728430431750550825937414107999299175967531699289648699961746620947
# 2387363771591^17-1, difficulty: 222.80, skewness: 115.61, alpha: 0.00
# cost: 9.68461e+17, est. time: 461.17 GHz days (not accurate yet!)
lss: 0
skew: 115.608
c6: 1
c0: -2387363771591
Y1: -1
Y0: 13606793610144465216226908028995378071
m: 13606793610144465216226908028995378071
rlim: 134000000
alim: 134000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.7
alambda: 2.7
type: snfs

Trial sieving 5K blocks.

Code:

swellman · 2017-10-20, 00:29

queued C229_149_54 is ready for SNFS.

Code:

n: 5653173533248994794247447276562651826449584825317819743946400192042140761541282322664794589728154821510962718342738281880966175531184298521350949426181114974944647080208693460157710313314350009561304996793884968965918238898780389
# 149^54+54^149, difficulty: 259.86, anorm: 1.47e+037, rnorm: 1.46e+049
# scaled difficulty: 261.86, suggest sieving rational side
# size = 7.920e-013, alpha = 0.000, combined = 9.797e-014, rroots = 0
type: snfs
size: 259
skew: 1.9442
c6: 1
c0: 54
Y1: -36197319879620191349
Y0: 20410046566186296742332216391818083643162624
rlim: 268000000
alim: 268000000
lpbr: 32
lpba: 32
mfbr: 64
mfba: 64
rlambda: 2.8
alambda: 2.8

Test sieving on the -r side with Q in blocks of 2K

Code:

Q=20M    4040
Q=60M    3588
Q=120    3053
Q=200M   2967
Q=300M   2510

Suggesting a sieveing range for Q of 20M-320M with a target # rels of 460M.

swellman · 2017-10-24, 01:02

queued C243_127_113 can be run on 14e, as it has survived a full t50, plus a few thousand more curves @B1=260M.

Code:

n: 107076943618137683967379943913650054643076848188518968335031265248775750535227821878160399025435757089546341568930171803072959527050772771064083256433250761400530764914052992400174726920350324042650111608058679402327460191506455452234578759081
# 127^113+113^127, difficulty: 260.74, anorm: 2.40e+038, rnorm: -2.89e+049
# scaled difficulty: 262.59, suggest sieving rational side
# size = 6.358e-013, alpha = 0.000, combined = 8.314e-014, rroots = 0
type: snfs
size: 260
skew: 4.9296
c6: 1
c0: 14351
Y1: -13021089174137413266744892374538813705886513
Y0: 9381465766970461872264625493558368225663
rlim: 268000000
alim: 268000000
lpbr: 32
lpba: 32
mfbr: 64
mfba: 64
rlambda: 2.8
alambda: 2.8

Test sieving on the -r side with Q in blocks of 2K.

Code:

Q=20M    3939
Q=80M    2826
Q=150M   2707
Q=250M   2607
Q=350M   2353

Suggesting a sieving range of 20M-370M for Q with target # rels=480M.

fivemack · 2017-10-24, 09:14

15e is running out of work to do. The 13xx-index Fibonacci numbers have been generally a bit tough to post-process; C180s are more reasonable but I can't generate reasonable polynomials for them fast enough to feed both my home equipment and nfs@home.

So what do you suggest? I can push more Fibonacci SNFS jobs, but basically that creates work for Greg and I don't know if he'd be OK with that. If people have 26x-difficulty SNFS from XYYXF, those would be handy. I would be delighted if someone else was willing to do ECM and polynomial selection for numbers of 183 digits or more from http://mersennus.net/fibonacci/smallest.txt

swellman · 2017-10-24, 10:52

Well, there is a C180 GNFS job ready to go, with my thanks to Max, VBCurtis and YuL for the poly search. It’s from AS 3408 and it’s been well covered with ECM courtesy of yoyo@Home. It looks like a 15e/32 job to be sieved on the -a side presumably. Only drawback is I’m traveling and can’t test sieve it until later this week. If someone is willing to test sieve it now please be my guest. Or submit it with a short range of Q - say 30M-250M - and adjust later.

I have several xyyx numbers ready for 15e that I’m happy to post but see above about test sieving.

3408:1671 test-sieved with guessed parameters and queued

swellman · 2017-10-24, 11:09

Quote:

Originally Posted by fivemack

I would be delighted if someone else was willing to do ECM and polynomial selection for numbers of 183 digits or more from http://mersennus.net/fibonacci/smallest.txt

I’ll take the C184 from L3865B if it’s available. How much ECM is ultimately needed at the t60 level before switching to NFS methods?

swellman · 2017-10-24, 11:37

queued C219_129_92 is ready for 15e

Code:

n: 251038195988045898600953315032429333639309555389212960413681197162218022181654170988435137749941172615502169056985465271548235503369507875117086731486571359512513617883253795784956569734210177673119844434553825369404441
# 129^92+92^129, difficulty: 255.89, anorm: 3.50e+033, rnorm: 1.95e+056
# scaled difficulty: 259.69, suggest sieving rational side
# size = 8.405e-018, alpha = 0.000, combined = 4.700e-014, rroots = 1
type: snfs
size: 255
skew: 8.6291
c5: 8
c0: 382743
Y1: -97862157334118736160267353892330031361
Y0: 572075719290556693384652005928556009754405080399872
rlim: 268000000
alim: 268000000
lpbr: 32
lpba: 32
mfbr: 64
mfba: 64
rlambda: 2.8
alambda: 2.8

queued C200_140_73 is ready for 15e:

Code:

n: 55280931276473990415949299833125217386743133024402737332484429638043667425816937651741743528557876250953079039653997690121186758330917881325758280261184663442707700950318206865968248329065834553826459
# 140^73+73^140, difficulty: 260.87, anorm: 5.86e+032, rnorm: 4.78e+057
# scaled difficulty: 265.02, suggest sieving rational side
# size = 3.407e-018, alpha = 0.000, combined = 2.492e-014, rroots = 1
type: snfs
size: 260
skew: 9.6972
c5: 1
c0: 85750
Y1: -2222401365111603200000000000000
Y0: 14894985451961941943846586557477384171894488107677281
rlim: 268000000
alim: 268000000
lpbr: 32
lpba: 32
mfbr: 64
mfba: 64
rlambda: 2.8
alambda: 2.8

fivemack · 2017-10-24, 11:57

Quote:

Originally Posted by swellman

I’ll take the C184 from L3865B if it’s available. How much ECM is ultimately needed at the t60 level before switching to NFS methods?

I make it about 15000 curves @ B1=260M (this is a bit more than previous versions of ecm-toy suggest, because I've updated the prior for factor distribution based on experience with the brilliant-numbers search)

A slightly more optimal search would be 15000 @ 43M followed by 12000 @ 260M if the number survived the first lot.

Please mail marin DOT mersennus AT gmail DOT com to reserve the number.

swellman · 2017-10-24, 12:30

Understood and reservation sent via email.

pinhodecarlos · 2017-10-24, 21:58

Quote:

Originally Posted by fivemack

15e is running out of work to do. The 13xx-index Fibonacci numbers have been generally a bit tough to post-process; C180s are more reasonable but I can't generate reasonable polynomials for them fast enough to feed both my home equipment and nfs@home.

What would be the ideal system to post process them?

OS
No. Cores
GB

VBCurtis · 2017-10-25, 01:46

Quote:

Originally Posted by pinhodecarlos

What would be the ideal system to post process them?
OS
No. Cores
GB

"ideal" : Bigger is better. If by ideal you mean fastest possible, then a Xeon server with 32+GB. More cores is more work is less time! YuL has been using Amazon server instances to blast through medium-sized LA jobs in hours. I don't think OS matters, but I don't have any experience with windows machines over 16GB. msieve is msieve regardless.

However, anyone with the minimum system can do the work, if they're patient. 32GB is enough to solve a matrix for anything the 15e queue can generate; but a 4-core might take 3-4 months to solve the matrix from a GNFS195.

The smaller 15e jobs (say, GNFS 185 or lower, SNFS 265 or lower) are similar in difficulty to the big 14e jobs; matrices 15M-25M, 3-6 quadcore-weeks. A 6-core merely cuts 1/3rd from the time; changing a 6 week job into a 4 week job isn't going to alter the course of any project, so a patient worker can run anything he wants on a quad core.

Example: My current job, the second 14e/33 job we've tried, is a 23M matrix at low-ish TD (due to oversieving). top claims msieve is using 9GB on a 16GB machine. So, a 25M matrix should be solvable on any 16GB machine if it's not used for much else, and 24GB can handle all but the biggest jobs while still using the machine for regular life.

A GNFS 180-185 should build a matrix smaller than 20M, and should be solvable on a 16GB machine (that's what I plan to do, anyway!).

2017-10-24, 09:14	#1203
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 2³·11·73 Posts	What next for 15e? 15e is running out of work to do. The 13xx-index Fibonacci numbers have been generally a bit tough to post-process; C180s are more reasonable but I can't generate reasonable polynomials for them fast enough to feed both my home equipment and nfs@home. So what do you suggest? I can push more Fibonacci SNFS jobs, but basically that creates work for Greg and I don't know if he'd be OK with that. If people have 26x-difficulty SNFS from XYYXF, those would be handy. I would be delighted if someone else was willing to do ECM and polynomial selection for numbers of 183 digits or more from http://mersennus.net/fibonacci/smallest.txt

2017-10-24, 10:52	#1204
swellman Jun 2012 3091₁₀ Posts	Well, there is a C180 GNFS job ready to go, with my thanks to Max, VBCurtis and YuL for the poly search. It’s from AS 3408 and it’s been well covered with ECM courtesy of yoyo@Home. It looks like a 15e/32 job to be sieved on the -a side presumably. Only drawback is I’m traveling and can’t test sieve it until later this week. If someone is willing to test sieve it now please be my guest. Or submit it with a short range of Q - say 30M-250M - and adjust later. I have several xyyx numbers ready for 15e that I’m happy to post but see above about test sieving. 3408:1671 test-sieved with guessed parameters and queued Last fiddled with by fivemack on 2017-10-24 at 17:19

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

2017-10-24, 12:30	#1208
swellman Jun 2012 3091₁₀ Posts	Understood and reservation sent via email.