2022 Queue management of 15e - Page 10

swellman · 2022-08-21, 13:25

Quote:

Originally Posted by pinhodecarlos

Hey guys, thank you so much for keeping this queue loaded. I need a favour. There are thousands of cores only allocated to this queue who are idle. Can we move one of the 16e smaller numbers (C214_139_87) here taking into consideration that each Wu's should fit under 1GB/thread. Thank you.

That’s a big ask only because of the terrible quintic polynomial for C214_139_87. Assuming Greg et al don’t object, I’m willing to reload it on 15e with maxed out parameters, but let me test sieve it and reexamine the sextic. I’m in the Wilderness right now (i.e. anywhere > 50 meters from my computers) but I’ll look at it tonight or in the morning.

If memory serves it will be a long hard sieving phase…

charybdis · 2022-08-21, 13:48

Quote:

Originally Posted by pinhodecarlos

Hey guys, thank you so much for keeping this queue loaded. I need a favour. There are thousands of cores only allocated to this queue who are idle. Can we move one of the 16e smaller numbers (C214_139_87) here taking into consideration that each Wu's should fit under 1GB/thread. Thank you.

I would guess 13_2_938m1 is the most feasible on 15e. Will test-sieve.

pinhodecarlos · 2022-08-21, 14:50

Some numbers to help decision making. On 15e queue we were daily sieving 33k units (wu size 4k) , 16e small 11k units (Wu size 1k). By himself one particular member was doing 22k per day. How it looks now: https://stats.free-dc.org/user/nfs/1084

charybdis · 2022-08-21, 15:13

QUEUED AS 13_2_938m1_2nd_try

13_2_938m1, SNFS difficulty 283, originally queued on 16e-small by VBCurtis. It's easier than several of the recent large 15e jobs.

Code:

n: 592255508891273891433051852817055982895416134239342155873928742987353815658249290934043752709866193555986011313495975786026642407845703866460808748600831192866992062678216248812957209799199466277975169888804187612710806776636410574327275925996708182572199337343868592730053600937221
skew: 0.68
type: snfs
c6: 52
c0: -1
Y1: -1
Y0: 91343852333181432387730302044767688728495783936
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 with 15e:

Code:

MQ       Norm_yield      Speed (sec/rel)
25          3981              0.106
40          3696              0.120
70          3237              0.139
100         2841              0.160
150         2587              0.179
200         2293              0.197
300         1920              0.226
400         1727              0.246
500         1574              0.264

Suggest sieving 25-430M for 950M raw relations.

pinhodecarlos · 2022-08-21, 17:57

Thank you, honestly.

VBCurtis · 2022-08-21, 19:11

Carlos-
Moving these jobs to 15e means about 30% wasted computation compared to running it on f-small.
The memory requirement for f-small is roughly 100MB larger than 15e, and still well under 1GB/process.

Could you suggest to this user/team that f-small is where we need the firepower? Or find out what we need to adjust to make f-small get more attention, since that's where we have a queue that will only get deeper with future jobs.

pinhodecarlos · 2022-08-21, 19:21

I cannot get hold of him, I did try a few weeks ago with a direct message flagging he would even increase his daily output by running the small numbers. Maybe Mike knows him from Team Ars Tecnhica forum.

My perspective is different than yours since I'm trying to feed his fleet whilst it is here. No hard feelings.
I don't have the powers to let people know where to deploy their resources.

VBCurtis · 2022-08-22, 03:44

QUEUED AS 13_2_929m1

13*2^929-1 is ready for, I guess, 15e:

Code:

n: 799516953194561531712600847044965086853762294809183841465586073526277593444419475948738754398198663425583431614513989774894294069622414410719883789187637509550397952345330238533458118721984680047155623834137969111765002149652207435060863633209
type: snfs
skew: 0.75
deg: 6
c6: 13
c0: -2
Y1: -1
Y0: 45671926166590716193865151022383844364247891968
rlim: 134000000
alim: 266000000
lpbr: 33
lpba: 33
mfbr: 96
mfba: 66
rlambda: 3.6
alambda: 2.5

Test-sieving 1kQ blocks on the r side indicates Q=25-365M should be enough for 950M relations:

Code:

#Q=30M	4351 rels, 0.092 sec/rel 62 sqQ
#Q=60M	3783 rels, 0.104 sec/rel 54 spQ
#Q=90M	3156 rels, 0.131 sec/rel 52 spQ
#Q=120M 3332 rels, 0.136 sec/rel 55 spQ
#Q=150M 3266 rels, 0.147 sec/rel 58 spQ
#Q=180M 2215 rels, 0.168 sec/rel 46 spQ
#Q=210M 2627 rels, 0.160 sec/rel 51 spQ
#Q=240M 2821 rels, 0.168 sec/rel 59 spQ
#Q=270M 2501 rels, 0.168 sec/rel 54 spQ
#Q=300M 2276 rels, 0.174 sec/rel 51 spQ
#Q=330M 1942 rels, 0.209 sec/rel 43 spQ
#Q=360M 1740 rels, 0.204 sec/rel 47 spQ

I would like to handle the matrix.

unconnected · 2022-08-24, 10:16

QUEUED AS C189_11040_10209

C189 from 11040:i10209 is ready for 15e queue:

Code:

n: 142511357330466243476746690497051575928345729751366129422971474654704075568089404245440258486192592405061316915235821208519879617537091746369391499418464177318972566719274808703086801429397 
skew: 181203923.855
# e-score 2.67097335e-14
c0: 8753585702118798264017971574180247062870008545 
c1: 1462380903005071847006946283271961507733 
c2: -3609286928071149997822765954299 
c3: -216398824337073239972533 
c4: 139051083328194
c5: 1081080
Y0: -3306911518669156528704633740527063178 
Y1: 1412443136271580297601 
rlim: 266000000
alim: 134000000
lpbr: 32
lpba: 33
mfbr: 64
mfba: 96
rlambda: 2.8
alambda: 3.7
type: gnfs
lss: 0

Suggesting sieving range is 30M-250M.
Thanks Plutie for the decent poly!

swellman · 2022-08-25, 20:28

QUEUED AS C195_145_122

C195_145_122 is a GNFS 195 from the XYYXF project now ready for sieving on 15e.

Code:

n: 328606880159851653901498068502945699522341371334506631710815644380739452303853012248598043817669999935646198807225558536316875879511592524393207275859568918684662490812329849077724697066606457917
skew: 23248876.16
type: gnfs
lss: 0
c0: 16734616657131278714808079418694212811411912
c1: 28182385282593881106959852573533945574
c2: 11436240044684009125565858850309
c3: -209544207391865601662151
c4: -17097385717580408
c5: 51236640
Y0: -34836735077907242766383210973608962666
Y1: 296155417798931427557711
# MurphyE (Bf=8.590e+09,Bg=4.295e+09,area=2.684e+16) = 9.666e-09
# cownoise 9.77038349e-15
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)
30          2342              0.398
50          2477              0.388
75          2370              0.493
100         2253              0.512
150         2072              0.664
200         1897              0.610
250         1776              0.642
300         1642              0.713
350         1568              0.654
400         1497              0.701
450         1472              0.694
500         1409              0.695
550         1357              0.764

Suggesting a sieving range for Q of 25-560M to generate 950M raw relations.

swellman · 2022-08-30, 14:02

QUEUED AS 11p7_290

11+7,290 is an SNFS 241 quartic from the HCN project now ready for sieving on 15e.

Code:

n: 1839624425573616146312628717991470174207268935078285319013117054513237081929144011280827172424615480449030281367897694330731503297800668697968119327449895250611804079681149825590263833917365601
skew: 1.0000
type: snfs
size: 241
c4: 1
c3: -1
c2: 1
c1: -1
c0: 1
Y1: -10367793076318844190248738727596255138212949486449
Y0: 2516377186292711566730985912068419625116019959228909823321881
rlim: 134000000
alim: 266000000
lpbr: 33
lpba: 31
mfbr: 96
mfba: 62
rlambda: 3.7
alambda: 2.7

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

Code:

MQ      Norm_yield    Speed (sec/rel)
35         1725            0.513
50         1900            0.492
75         2104            0.487
100        2187            0.473
150        2195            0.484
200        2014            0.523
250        1895            0.568
300        1927            0.517
350        1883            0.563

Suggesting a sieving range for Q of 35-340M to generate 610M raw relations.

2022-08-21, 14:50	#102
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 2³·641 Posts	Some numbers to help decision making. On 15e queue we were daily sieving 33k units (wu size 4k) , 16e small 11k units (Wu size 1k). By himself one particular member was doing 22k per day. How it looks now: https://stats.free-dc.org/user/nfs/1084 Last fiddled with by pinhodecarlos on 2022-08-21 at 14:54

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

2022-08-21, 17:57	#104
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 2³·641 Posts	Thank you, honestly.

2022-08-21, 19:11	#105
VBCurtis "Curtis" Feb 2005 Riverside, CA 15FE₁₆ Posts	Carlos- Moving these jobs to 15e means about 30% wasted computation compared to running it on f-small. The memory requirement for f-small is roughly 100MB larger than 15e, and still well under 1GB/process. Could you suggest to this user/team that f-small is where we need the firepower? Or find out what we need to adjust to make f-small get more attention, since that's where we have a queue that will only get deeper with future jobs.

2022-08-21, 19:21	#106
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 2³·641 Posts	I cannot get hold of him, I did try a few weeks ago with a direct message flagging he would even increase his daily output by running the small numbers. Maybe Mike knows him from Team Ars Tecnhica forum. My perspective is different than yours since I'm trying to feed his fleet whilst it is here. No hard feelings. I don't have the powers to let people know where to deploy their resources.