2022 Queue management of 15e - Page 4

swellman · 2022-03-30, 17:18

QUEUED AS 10p9_283

10+9,283 is a HCN, recently finished with ECM (full t60 plus 1000 curves @B1=850M). It is a GNFS 194, using a polynomial found by Gimarel. It can be run on 15e.

Code:

n: 81743383332836962057695785771313489881126164906792638283009336778727990819875122320852219082832408804312533152511125453405418535403488398501615412734357849727912722309601547604190321551350629161
skew: 73099244.97
type: gnfs
lss: 0
c0: 81927702151939929172622142443110617559424626640
c1: -2636022244986997462244685796892390178827
c2: -42901376517955052267868136139126
c3: 92695303237307143296671
c4: 9240270496236570
c5: 116424000
Y0: -16962426209861537614672116488694011418
Y1: 113869057497149248839901
# norm 6.923660e-19 alpha -9.321044 e 1.232e-14 rroots 3
rlim: 266000000
alim: 134000000
lpbr: 33
lpba: 33
mfbr: 66
mfba: 96
rlambda: 3.0
alambda: 3.7

Results of test sieving on the algebraic side with Q in blocks of 1000:

Code:

MQ         Norm_yield      Speed (sec/rel)
50            3139             0.354
70            3004             0.408
100           2929             0.387
150           2756             0.454
200           2357             0.494
250           2147             0.561
300           2047             0.559
350           2023             0.553
400           1827             0.613
450           1850             0.591

Suggesting a sieving range for Q of 50-465M to generate 950M raw relations.

charybdis · 2022-03-30, 20:07

Quote:

Originally Posted by swellman

QUEUED AS 3p2_1674L

3+2,1674L is a GNFS 192 from the HCN project. It has completed ECM and is ready for sieving. Thanks to Gimarel for the record breaking poly.
...

It's a bit late now, but what was wrong with this poly?

Code:

n: 118774638495343044085764617595746668209262348394154072091504703448929588459098368911352625946762996320778680241182601579702534542835804685795107781414309880804992674911118346161218362744529909
skew: 0.82
c6: 27
c4: -108
c3: -36
c2: 108
c1: 72
c0: 8
Y1: 1871021475612879770007267593649389568 = 3*6^46
Y0: -235655016338368245402588046228913837374112915 = -(3^93+2^93)

Difficulty 268, ~3 times easier than GNFS. Don't forget that L/Ms with exponents divisible by 9 can sometimes produce sextics...

jyb · 2022-03-30, 23:36

Quote:

Originally Posted by charybdis

It's a bit late now, but what was wrong with this poly?

Code:

n: 118774638495343044085764617595746668209262348394154072091504703448929588459098368911352625946762996320778680241182601579702534542835804685795107781414309880804992674911118346161218362744529909
skew: 0.82
c6: 27
c4: -108
c3: -36
c2: 108
c1: 72
c0: 8
Y1: 1871021475612879770007267593649389568 = 3*6^46
Y0: -235655016338368245402588046228913837374112915 = -(3^93+2^93)

Difficulty 268, ~3 times easier than GNFS. Don't forget that L/Ms with exponents divisible by 9 can sometimes produce sextics...

Yes, you're correct, as also noted in this post. And with a slight variation in the polynomials, the difficulty can be dropped to 266.2.

But you're right, it's probably too late to be worth canceling the job at this point, it's nearly done sieving.

charybdis · 2022-03-31, 02:12

Quote:

Originally Posted by jyb

Yes, you're correct, as also noted in this post.

I think that apart from the ones in that post, the only other number left in the HCN tables for which this specific trick works is 8+3,882M, difficulty 266. It works for Cunningham 2LMs too of course, such as 2,2862L which 16e-small did a couple of months ago (and 2,2754L which I'm currently sieving).

[as I think of it: take a quartic where the root is a cube, write it as a degree-12 polynomial, and use the degree-halving trick]

jyb · 2022-03-31, 04:01

Quote:

Originally Posted by charybdis

I think that apart from the ones in that post, the only other number left in the HCN tables for which this specific trick works is 8+3,882M, difficulty 266. It works for Cunningham 2LMs too of course, such as 2,2862L which 16e-small did a couple of months ago (and 2,2754L which I'm currently sieving).

[as I think of it: take a quartic where the root is a cube, write it as a degree-12 polynomial, and use the degree-halving trick]

It also works for Cunningham 6LMs.

swellman · 2022-03-31, 17:01

Quote:

Originally Posted by charybdis

I think that apart from the ones in that post, the only other number left in the HCN tables for which this specific trick works is 8+3,882M, difficulty 266. It works for Cunningham 2LMs too of course, such as 2,2862L which 16e-small did a couple of months ago (and 2,2754L which I'm currently sieving).

[as I think of it: take a quartic where the root is a cube, write it as a degree-12 polynomial, and use the degree-halving trick]

An oversight on my part, as I’m the one who queued it as a GNFS. The technique discussed was in my notes but I didn’t have 3+2,1674L flagged.

Based on the discussion last year and charybdis’s comments above, there seems to be only four remaining such cases in the current project: 3+2,1818L, 3+2,1926L/M and 8+3,882M. And only the last case is even close to completing ECM (which will likely still take months to run based on Yoyo’s queue). However I have now flagged these four HCNs in my notes for their unique SNFS polynomials.

frmky · 2022-03-31, 17:08

Quote:

Originally Posted by swellman

Greg says go for it.

I’ll post it as a 34-bit job.

It doesn't work on the Windows clients. They are limited to 33-bit LPs. I stopped WU generation by lowering the max and cancelled the generated WUs. Please requeue it as a 33-bit job.

VBCurtis · 2022-03-31, 17:19

Good to know about 34LP on 15e.
(edited) Can the new 33LP job have Q-range 35-290M? I'll salvage whatever relations I can from the first job in 25-35M.

swellman · 2022-03-31, 18:25

Quote:

Originally Posted by VBCurtis

Good to know about 34LP on 15e.
(edited) Can the new 33LP job have Q-range 35-290M? I'll salvage whatever relations I can from the first job in 25-35M.

Of course. I’ll do so this afternoon.

Done.

swellman · 2022-04-04, 17:45

QUEUED AS 11m7_251

11-7,251 is a HCN now finishing up ECM. It is a SNFS of difficulty 261 with a sextic polynomial. It would not fit within the bounds of 15e_small, so it goes onto 15e as a 32/31 hybrid with asymmetric lims.

Code:

n: 2245579998294160634668455563717515018621121599329301027081896517966739372354508276945433549547443106236604322045929860646893053214954679995473589135270426482499681481273180827587114797570309159256687304184611452601320219265711039697612773621835453
skew: 1.0782
type: snfs
size: 261
c6: 7
c0: -11
Y1: -311973482284542371301330321821976049
Y0: 54763699237492901685126120802225273763666521
rlim: 134000000
alim: 266000000
lpbr: 32
lpba: 31
mfbr: 94
mfba: 62
rlambda: 3.5
alambda: 2.7

Results of test sieving on the rational side with Q in blocks of 1000:

Code:

MQ       Norm_yield      Speed (sec/rel)
40          3070              0.193
70          2799              0.203
100         2409              0.259
150         1937              0.310
200         1828              0.352

Suggesting a sieving range for Q of 40-200M to generate 360M raw relations.

swellman · 2022-04-06, 09:57

QUEUED AS 5p2_1090M

5+2,1090M is a HCN now ready for sieving. It is a GNFS 194 which can just fit into the 15e siever. (Going forward, I will use 16e for GNFS 195+.) Thanks to Gimarel for yet another record polynomial.

Code:

n: 22210632230246541477713039859446690282652445390183852971394893889999886281468063387054645931605201553060731319129351920515814331556338993760550731909490785594053526090946998578599980383681176081
# norm 7.960015e-19 alpha -7.945662 e 1.361e-14 rroots 3
skew: 19728660.01
type: gnfs
lss: 0
c0: 71147189690970348375082680657192486752131250
c1: 27672129762822669440216286237579758257
c2: -10201676830695417023567117095232
c3: -112200717722234171613079
c4: 14689270664747220
c5: 317520000
Y0: -12284852852780799375166334341274201337
Y1: 19543970466085276110731
rlim: 266000000
alim: 134000000
lpbr: 33
lpba: 33
mfbr: 66
mfba: 96
rlambda: 3.0
alambda: 3.7

Results of test sieving on the algebraic side with Q in blocks of 1000:

Code:

MQ       Norm_yield       Speed (sec/rel)
35          3111               0.207
50          3212               0.210
75          3051               0.243
100         2956               0.256
150         2670               0.291
200         2401               0.337
250         2195               0.347
300         2123               0.341
350         1997               0.382
400         2007               0.352

Suggesting a sieving range for Q of 35-430M to generate 950M raw relations.

2022-03-31, 17:19	#41
VBCurtis "Curtis" Feb 2005 Riverside, CA 53×113 Posts	Good to know about 34LP on 15e. (edited) Can the new 33LP job have Q-range 35-290M? I'll salvage whatever relations I can from the first job in 25-35M. Last fiddled with by VBCurtis on 2022-03-31 at 17:21

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Queue management for 14e queue	VBCurtis	NFS@Home	170	2023-01-02 15:27
2022 - queue management for 15e_small	swellman	NFS@Home	186	2022-12-29 17:58
Queue management for 16e queue	VBCurtis	NFS@Home	154	2022-12-23 21:35
Queue management for e_small and 15e queues	VBCurtis	NFS@Home	254	2022-01-02 01:59
Improving the queue management.	debrouxl	NFS@Home	10	2018-05-06 21:05