2022 Queue management of 15e - Page 11

charybdis · 2022-09-01, 01:36

QUEUED AS L1992

Lucas(1992) is an SNFS candidate of difficulty 278. It has had 30000 curves at 260M and is ready for 15e.

Code:

n: 10347436847870478097473793996844728322517625552949268810869325393228670032192139315003359513466518229018036726512764912259199493310792806318998768068141666522005734059417473442606468803049317915990798736047
skew: 1.14
type: snfs
c6: 4
c5: 3
c4: 30
c3: -15
c2: 45
c1: -27
c0: 6
Y1: 6867260041627791953052057353082063289320596021
Y0: -11111460156937785151929026842503960837766832936
rlim: 134000000
alim: 266000000
lpbr: 33
lpba: 33
mfbr: 96
mfba: 66
rlambda: 3.8
alambda: 2.6

Test-sieving 1k ranges on the rational side:

Code:

MQ      Norm_yield    Speed (sec/rel)
25         5693            0.079
50         4955            0.096
75         4504            0.106
100        3763            0.127
150        3400            0.143
200        2918            0.160
250        2860            0.161
300        2620            0.173

Suggest sieving 25-290M for 950M raw relations.

pinhodecarlos · 2022-09-03, 06:30

Just a quick note. That guy who was running only the 15e was on his goal to reach 100M overall, he is now running yoyo. Probably you already noticed the output dropping significantly. Good while it lasted.

https://stats.free-dc.org/user/yoy/16779

swellman · 2022-09-04, 21:59

QUEUED AS 6p5_374

6+5,374 is a quintic SNFS of difficulty 264 from the HCN project. It is ready for sieving on 15e.

Code:

n: 772195493570538574472133350374480967574844577842826289795771391382764100945904744802912346588848659710233586306614014080071147694295400643959678427771190206502436816042476011847587401095077205239802931427496801156439464669390355574847473
skew: 1.0000
type: snfs
size: 264
c5: 1
c4: -1
c3: -4
c2: 3
c1: 3
c0: -1
Y1: -166771816996665690000000000000000000000000000000000
Y0: 82089350328392269627321951212068476903210635085853041
rlim: 134000000
alim: 266000000
lpbr: 32
lpba: 32
mfbr: 94
mfba: 64
rlambda: 3.5
alambda: 2.8

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

Code:

MQ       Norm_yield      Speed (sec/rel)
35          2442              0.354
50          2600              0.324
75          2728              0.337
100         2656              0.356
150         2457              0.430
200         2458              0.362
250         2210              0.428

Suggesting a sieving range for q of 35-210M to generate 450M raw relations.

kruoli · 2022-09-06, 14:26

GCW 3,543- is an SNFS 262 which is ready for sieving.

Code:

n: 2097026643802877507269943929206861145366651524876715588205870797295371435262177200032311581577065062943201951040730730939818168279185975155039206178294169390003508531914739075399798610623509784910133338777
skew: 0.26122
type: snfs
size: 262
c6: 14661
c0: -1
Y0: -8727963568087712425891397479476727340041449
Y1: 1
rlim: 134000000
alim: 266000000
lpbr: 32
lpba: 31
mfbr: 91
mfba: 64
lss: 1
rlambda: 3.8
alambda: 2.7

To be sieved on the rational side. Test sieving blocks of 1k gave

Code:

  Q    norm. yield
 40M   2677
 70M   2216
125M   1880
200M   1720
300M   1481

It would have not quite fit into 15e_small (e.g. Q 37M-217M would have given ~340M relations with the smaller lims). Therefore I suggest sieving Q 40M-230M for 370M relations.

I will do the LA.

VBCurtis · 2022-09-06, 14:37

Quote:

Originally Posted by kruoli

rlim: 134000000
alim: 266000000
lpbr: 32
lpba: 31
mfbr: 91
mfba: 64
[....]It would have not quite fit into 15e_small (e.g. Q 37M-217M would have given ~340M relations with the smaller lims). Therefore I suggest sieving Q 40M-230M for 370M relations.

I will do the LA.

Your lpb's and mfb's don't match up. 31/64 doesn't make sense, and 32/91 is unusual but not crazy. Looks like you reversed something. If you test-sieved this combo, you ought to test again with mfb's 92/62 or 93/61. But, since your average yield is below 2.0, 32/32 with 92 or 93 and 64 is likely better anyway.

kruoli · 2022-09-06, 14:39

It looks like my test-sieving-script has a bug and I was too blind to catch it, thanks!

If you say 32/32 is appropriate here I will re-sieve with 32/32 and move it to 15e_small.

swellman · 2022-09-06, 14:59

Quote:

Originally Posted by kruoli

It looks like my test-sieving-script has a bug and I was too blind to catch it, thanks!

If you say 32/32 is appropriate here I will re-sieve with 32/32 and move it to 15e_small.

If you want to sieve this with 15e_small, don’t forget to make both lim = 134M.

kruoli · 2022-09-06, 15:09

Making it 32/32 at lower lims was not good enough (around 390M relations), I will re-sieve for normal 15e.

kruoli · 2022-09-06, 16:27

QUEUED AS 3_543m

GCW 3,543- is an SNFS 262 which is ready for sieving.

Code:

n: 2097026643802877507269943929206861145366651524876715588205870797295371435262177200032311581577065062943201951040730730939818168279185975155039206178294169390003508531914739075399798610623509784910133338777
skew: 0.26122
type: snfs
size: 262
c6: 14661
c0: -1
Y0: -8727963568087712425891397479476727340041449
Y1: 1
rlim: 134000000
alim: 266000000
lpbr: 32
lpba: 32
mfbr: 94
mfba: 64
lss: 1
rlambda: 3.9
alambda: 2.7

To be sieved on the rational side. Test sieving blocks of 1k gave

Code:

  Q    norm. yield
 40M   3655
 70M   3104
125M   2689
200M   2511
300M   2310

I suggest sieving Q 40M-210M for 470M+ relations.

I will do the LA.

VBCurtis · 2022-09-06, 18:31

Quote:

Originally Posted by kruoli

I suggest sieving Q 40M-210M for 470M+ relations.

I will do the LA.

10% smaller sieve range than 31/32.

It's worth noting that using higher-than-strictly-necessary LP bounds means jobs don't need the absolute maximum relations count to build a good matrix. I've run 15e/32LP jobs that only needed 320M raw relations, because the jobs were a size that others might use 30LP (say, GNFS-167). That's an extreme, but I bet this job needs only 440M raw relations, which would save another 10MQ.
If you have time to filter twice, please try one run with filter_maxrels=440M and let's compare matrix sizes.

swellman · 2022-09-09, 12:49

QUEUED AS 9p7_283

9+7,283 is a SNFS 270 composite from the HCN project. It is sieved on 15e.

Code:

n: 170846439673200184435309938385385060653003635613117738799480834616922885468214755375809745242758989975794297370256548104543596886967251753546920228795598287182875795718376567439213341811486929467
skew: 0.9590
type: snfs
size: 270
c6: 9
c0: 7
Y1: -5243338316756303634461458718861951455543
Y0: 706965049015104706497203195837614914543357369
rlim: 134000000
alim: 266000000
lpbr: 32
lpba: 32
mfbr: 94
mfba: 64
rlambda: 3.5
alambda: 2.8

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

Code:

MQ       Norm_yield      Speed (sec/rel)
35          3380              0.173
50          3187              0.188
75          2907              0.289
100         2693              0.257
150         2264              0.312
200         2109              0.316
250         1936              0.317

Suggesting a sieving range for Q of 35-215M to generate 460M raw relations.

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

2022-09-03, 06:30	#112
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 7·733 Posts	Just a quick note. That guy who was running only the 15e was on his goal to reach 100M overall, he is now running yoyo. Probably you already noticed the output dropping significantly. Good while it lasted. https://stats.free-dc.org/user/yoy/16779

2022-09-06, 14:39	#116
kruoli "Oliver" Sep 2017 Porta Westfalica, DE 3²·149 Posts	It looks like my test-sieving-script has a bug and I was too blind to catch it, thanks! If you say 32/32 is appropriate here I will re-sieve with 32/32 and move it to 15e_small.

2022-09-06, 15:09	#118
kruoli "Oliver" Sep 2017 Porta Westfalica, DE 3²·149 Posts	Making it 32/32 at lower lims was not good enough (around 390M relations), I will re-sieve for normal 15e.