Polynomial Request Thread - Page 102

swellman · 2018-03-30, 01:07

Quick update: slow is my progress. My normally “fast” script crawls at 1000 c5 per hr. Running 1-16,000 yielded 5400+ polys to be -npr’d!

Still chewing on that batch but I’m abandoning the script.

I’ll run my other script - it gave me a record on the C195 from AS 3366 and was much faster with less garbage retained for -npr.

Max0526 · 2018-03-30, 16:09

I'll be out of the loop for the next four days, sorry. I tried C197 with yafu on CPU, saw a lot of 2's and maybe an occasional 3. If any of you got to a 4e-15, it's already better than CPU yafu. Just post your top E score from time to time to track the progress.

Max0526 · 2018-03-30, 16:16

There is a dive in the top score table for a C191. My old poly 1.581e-14, July 2016, 10,429- was apparently never a record (see c191 links at http://cosweb1.fau.edu/~sbai/). Anybody has a good C191 candidate in mind?

Max0526 · 2018-03-30, 17:22

Please ignore the message above. I found the following from swellman:

Code:

Sat Jun  3 16:03:51 2017  Msieve v. 1.53 (SVN unknown)
Sat Jun  3 16:03:51 2017  random seeds: d3184c2c 10718587
Sat Jun  3 16:03:51 2017  factoring 29460893303338144751360360097976743017149046981259832053501450854362438285630845458294329588136961317820466603439061784124252469966169391148524869909406500896547611862071404959591325864761463 (191 digits)
Sat Jun  3 16:03:52 2017  searching for 15-digit factors
Sat Jun  3 16:03:53 2017  commencing number field sieve (191-digit input)
Sat Jun  3 16:03:53 2017  R0: -1870545265872388890161243166679811449
Sat Jun  3 16:03:53 2017  R1: 554791760382776548817
Sat Jun  3 16:03:53 2017  A0: 826319531224681659341264214653314828053252960
Sat Jun  3 16:03:53 2017  A1: 26940857953335994224321141436849154800
Sat Jun  3 16:03:53 2017  A2: -22688220016211567990602767782164
Sat Jun  3 16:03:53 2017  A3: 15607687492217763886884
Sat Jun  3 16:03:53 2017  A4: 54690847245705303
Sat Jun  3 16:03:53 2017  A5: 1286485200
Sat Jun  3 16:03:53 2017  skew 17653518.44, size 1.011e-18, alpha -8.479, combined = 1.735e-14 rroots = 3

swellman · 2018-03-31, 01:16

Quote:

Originally Posted by Max0526

Please ignore the message above. I found the following from swellman:

Code:

Sat Jun  3 16:03:51 2017  Msieve v. 1.53 (SVN unknown)
Sat Jun  3 16:03:51 2017  random seeds: d3184c2c 10718587
Sat Jun  3 16:03:51 2017  factoring 29460893303338144751360360097976743017149046981259832053501450854362438285630845458294329588136961317820466603439061784124252469966169391148524869909406500896547611862071404959591325864761463 (191 digits)
Sat Jun  3 16:03:52 2017  searching for 15-digit factors
Sat Jun  3 16:03:53 2017  commencing number field sieve (191-digit input)
Sat Jun  3 16:03:53 2017  R0: -1870545265872388890161243166679811449
Sat Jun  3 16:03:53 2017  R1: 554791760382776548817
Sat Jun  3 16:03:53 2017  A0: 826319531224681659341264214653314828053252960
Sat Jun  3 16:03:53 2017  A1: 26940857953335994224321141436849154800
Sat Jun  3 16:03:53 2017  A2: -22688220016211567990602767782164
Sat Jun  3 16:03:53 2017  A3: 15607687492217763886884
Sat Jun  3 16:03:53 2017  A4: 54690847245705303
Sat Jun  3 16:03:53 2017  A5: 1286485200
Sat Jun  3 16:03:53 2017  skew 17653518.44, size 1.011e-18, alpha -8.479, combined = 1.735e-14 rroots = 3

This is not my work - I merely nominated it for NFS@Home.

The finder of this poly is Gimarel.

swellman · 2018-03-31, 02:06

Quote:

Originally Posted by swellman

Quick update: slow is my progress. My normally “fast” script crawls at 1000 c5 per hr. Running 1-16,000 yielded 5400+ polys to be -npr’d!

Still chewing on that batch but I’m abandoning the script.

I’ll run my other script - it gave me a record on the C195 from AS 3366 and was much faster with less garbage retained for -npr.

Well, my other script is running c5 at 300/hr.

I’m giving it a week, and then back to my “fast” script. I’ve made a few decisions

- I’m only working 1-500K. 500k-2M would take months. So 500k-2M is still unclaimed. Sorry but my old/slow hardware limits my productivity up here north of C190.

- Only posting e-scores > 6.0. So far my high was 5.77, which is quite terrible.

- Lastly, I’ve got to play with alternate scripts

VBCurtis · 2018-03-31, 18:14

Roughly two GPU-days to cover 5-5.35M. Best is 6.37, with four or five polys over 6.0.
I'm using stage1norm near 6e28 (still experimenting for best production), which on my 750ti yields 200+ hits a day with stage2norm= 6e26. I've had a handful of *e25 hits so far, less than ten.

RichD · 2018-04-01, 12:57

The best I've seen is 6.704 through 4.25M.

VBCurtis · 2018-04-02, 15:47

Quote:

Originally Posted by VBCurtis

Thanks Max! I'm testing the 1.226 now for best parameters. I'll then use the same parameters to test the other two, will let you know what I chose.

The poly with score 1.226 sieved most quickly, and had best yield over most Q (for small Q, the 1.214 had better yield and same sieve speed, but performance fell off more steeply over the likely Q-range).

I haven't tried editing factmsieve.py yet, but assuming I get it to use 134M for all alim/rlim .job files I'll be running 34LP/67mfba-r, alim=rlim=134M. I'll use 16f from Q=3-9M, and 15f from 9M-50M or so. Yield is around 50 for the 16f region, and around 20 for the 15f region. sec/rel is around 0.025 from 3M to 20M, 0.030 on avg from 20-50M. This will be my first full factorization with the f sievers; I will not be sieving composite Q. If duplicate-relation rates are similar to e sievers, this GNFS-179 project will take about the same time as the GNFS-175 I just completed.

15e/33 is forecast to take 50M thread-seconds for 600M raw relations. 15f-16f combo and 34LP is forecast to take 32M thread-seconds for 1100M raw relations.

swellman · 2018-04-02, 22:34

Quote:

Originally Posted by RichD

The best I've seen is 6.704 through 4.25M.

Best I’ve seen is 6.07! My speed is increasing as I get out of the mud, i.e. c5 < 10^5. Still panning for gold...

RichD · 2018-04-02, 23:32

Sadly,

Code:

expecting poly E from 7.84e-15 to > 9.01e-15

2018-03-30, 16:09	#1113
Max0526 "Max" Jun 2016 Toronto 19·47 Posts	C197 I'll be out of the loop for the next four days, sorry. I tried C197 with yafu on CPU, saw a lot of 2's and maybe an occasional 3. If any of you got to a 4e-15, it's already better than CPU yafu. Just post your top E score from time to time to track the progress.

2018-03-30, 16:16	#1114
Max0526 "Max" Jun 2016 Toronto 19×47 Posts	looking for a C191 candidate There is a dive in the top score table for a C191. My old poly 1.581e-14, July 2016, 10,429- was apparently never a record (see c191 links at http://cosweb1.fau.edu/~sbai/). Anybody has a good C191 candidate in mind?

2018-04-01, 12:57	#1119
RichD Sep 2008 Kansas 3,391 Posts	C197 The best I've seen is 6.704 through 4.25M.

2018-04-02, 23:32	#1122
RichD Sep 2008 Kansas 3,391 Posts	Sadly, Code: expecting poly E from 7.84e-15 to > 9.01e-15

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
GIMPS wiki account request thread	ixfd64	mersennewiki	169	2018-09-21 05:43
Polynomial Discriminant is n^k for an n-1 degree polynomial	carpetpool	Miscellaneous Math	14	2017-02-18 19:46
Lost Prime Raider password request thread	cheesehead	Forum Feedback	6	2009-07-28 13:02
Polynomial	R.D. Silverman	NFSNET Discussion	13	2005-09-16 20:07
Deutscher Thread (german thread)	TauCeti	NFSNET Discussion	0	2003-12-11 22:12

2018-03-30, 01:07	#1112
swellman Jun 2012 3091₁₀ Posts	Quick update: slow is my progress. My normally “fast” script crawls at 1000 c5 per hr. Running 1-16,000 yielded 5400+ polys to be -npr’d! Still chewing on that batch but I’m abandoning the script. I’ll run my other script - it gave me a record on the C195 from AS 3366 and was much faster with less garbage retained for -npr.

2018-03-31, 18:14	#1118
VBCurtis "Curtis" Feb 2005 Riverside, CA 130A₁₆ Posts	Roughly two GPU-days to cover 5-5.35M. Best is 6.37, with four or five polys over 6.0. I'm using stage1norm near 6e28 (still experimenting for best production), which on my 750ti yields 200+ hits a day with stage2norm= 6e26. I've had a handful of *e25 hits so far, less than ten.