Queue management for e_small and 15e queues - Page 4

RichD · 2020-10-30, 14:36

Quote:

Originally Posted by swellman

I have no issue enqueuing this job even though it’s a touch over the difficulty and Q range limits for e_small. Any objections?

Where ever you think is best for balancing the grid.

VBCurtis · 2020-10-30, 16:07

I don't think the Q range nor LP size should be the deal-killer for e-small; but since memory use is advertised to BOINCers and overshooting the claims of memory use on 14e was a topic of complaint from the users previously. We should be hesitant about putting lim=134/268 jobs onto e-small.

That said, with one of these BOINC-competition things firing up, feeding the grid is more important this week than strictly following rules, so I point this out just for future "normal" times. E-small is more likely to run out right now, so we feed it.

I'll test-sieve a couple more jobs around SNFS-265 today, since it looks like the BOINCer army has arrived.

jyb · 2020-10-30, 16:29

Quote:

Originally Posted by VBCurtis

I don't think the Q range nor LP size should be the deal-killer for e-small; but since memory use is advertised to BOINCers and overshooting the claims of memory use on 14e was a topic of complaint from the users previously. We should be hesitant about putting lim=134/268 jobs onto e-small.

That said, with one of these BOINC-competition things firing up, feeding the grid is more important this week than strictly following rules, so I point this out just for future "normal" times. E-small is more likely to run out right now, so we feed it.

I'll test-sieve a couple more jobs around SFNS-265 today, since it looks like the BOINCer army has arrived.

And FWIW, I have several jobs with difficulties in the high 250s that are probably actually within bounds for 15e-small. I'll work on getting a few test-sieved.

unconnected · 2020-10-30, 18:36

QUEUED AS C172_785232_11530

C172 from 785232:i11530 for 15_e small queue:

Code:

n: 2017146075209695116896365398417236105967681538512158722562374886741873756695928091960832628298102885487202336164778114259060003668462472752614297229623400985805685015158967
# norm 9.651545e-17 alpha -7.661741 e 2.471e-13 rroots 3
skew: 34977867.85
c0: 4447182967122057907765606348536577374031232
c1: 552797606118231906377542190590976096
c2: -34420768132857712903595321496
c3: -1016128504754010736787
c4: 48736323796768
c5: 194220
Y0: -1596940251429871087366953519990759
Y1: 907168370488505929
rlim: 134000000
alim: 134000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.6
alambda: 2.6
type: gnfs
lss: 0

Suggesting sieve range is 40M-140M, I'll take the LA.
P.S. This task is slightly big for 14e queue within new guidelines, but you decide because 14e is almost empty.

VBCurtis · 2020-10-30, 20:46

We like 14e empty, and C172 is definitely within e-small guidelines. Good choice.

RichD · 2020-10-30, 21:31

Quote:

Originally Posted by VBCurtis

We should be hesitant about putting lim=134/268 jobs onto e-small.

That was my fault for suggesting e_small. I seem to remember the 200 average lim but that is for 15e, not e_small.

I won't have anymore for e_small and 15e is currently building a long line.

VBCurtis · 2020-10-30, 22:05

During BOINC-challenge time, e-small was the right choice there for the job; with the benefit of hindsight we might just choose 134/200 or 134/180 for such a job, giving up a bit of speed to keep e-small fed without going too far over memory guidelines.

Your job surely did less damage than my bad poly on 5_791m1!!
Here's a corrected job for that number, with the full composite on the "n:" line:

Code:

n: 111812867780754346560415289999565367225454303594734581562227132753730403501180897472097054990558568023911409057075955845067437060818207472286533090719208303228335541944193280571602322588666038416993870470159838517777152295693
type: snfs
skew: 0.87
deg: 6
c6: 5
c0: -2
Y1: -1
Y0: 5444517870735015415413993718908291383296
rlim: 67000000
alim: 50000000
lpbr: 32
lpba: 31
mfbr: 59
mfba: 58
rlambda: 2.7
alambda: 2.7

As noted previously, please sieve Q=10-72M on the -a side. A particularly lousy time to make such a mistake... :-/
EDIT: I'll PM Greg to kill the first try at this job.

frmky · 2020-10-31, 03:20

I requeued it.

chris2be8 · 2020-10-31, 17:08

QUEUED AS f91_127m1

(91^127-1)/8885379019009572114580296807875592974052066631170 from the Brent tables:

Code:

# Built Fri Oct 30 17:09:16 2020
# Estimated SNFS difficulty 249, GNFS equivalent 180, GNFS difficulty 200, degree 6
n: 70726288261859730096617803617430582262211312855340530291518042364729425580221365347709790195080930080009121691682197490771547106020923401882286489424688568288976730740133730938827386478255699769759209
type: snfs
c6: 91
c0: -1
# Y0 = 91^21
Y0: 137996870875659993023030601717979081222891
# Y1 = 1^0
Y1: 1
# msieve rating: skew 0.47, size 3.482e-12, alpha 0.164, combined = 2.903e-13 rroots = 2
skew: 0.47
rlim: 120000000
alim: 120000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.6
alambda: 2.6
lss: 0

Sieve on 15e_small from 40M to 120M on the algebraic side.

Test sieving 10k ranges:

Code:

I'll do the LA.

Chris

chris2be8 · 2020-11-08, 16:43

QUEUED AS f20_193p1

(20^193+1)/21 from the Brent tables:

Code:

# Built Sat Nov  7 16:58:32 2020
# Estimated SNFS difficulty 252, GNFS equivalent 181, GNFS difficulty 250, degree 6
n: 5978192128939695965557894688769206110573671851870509059900952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952380952381
type: snfs
c6: 20
c0: 1
# Y0 = 20^32
Y0: 429496729600000000000000000000000000000000
# Y1 = 1^0
Y1: 1
# msieve rating: skew 0.61, size 2.889e-12, alpha 0.209, combined = 2.534e-13 rroots = 0
skew: 0.607
rlim: 120000000
alim: 120000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.6
alambda: 2.6

Sieve on 15e_small from 40M to 120M on the rational side.

Test sieving 10k ranges:

Code:

I also test sieved on the algebraic side. That's slower but there's not much in it.

I'll do the LA.

Chris

RichD · 2020-11-08, 19:47

QUEUED AS 7489_67m1

C254 from the OPN t550 file.

Code:

n: 19154840215330208503787115744593094531540002908539926854298646600649592315484715365499344342915572111612181784107602272399610890987759942058078070777366850467164293154732836191675565637168096234709535909916854863497821052789709584606400148747284118769999
# 7489^67-1, difficulty: 259.59, skewness: 0.23, alpha: 0.00
# cost: 1.71251e+19, est. time: 8154.82 GHz days (not accurate yet!)
skew: 0.226
c6: 7489
c0: -1
Y1: -1
Y0: 4155871746805483237658018945361617561661889
type: snfs
rlim: 134000000
alim: 134000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.7
alambda: 2.7

Trial sieving 5K blocks.

Code:

2020-10-30, 16:07	#35
VBCurtis "Curtis" Feb 2005 Riverside, CA 13262₈ Posts	I don't think the Q range nor LP size should be the deal-killer for e-small; but since memory use is advertised to BOINCers and overshooting the claims of memory use on 14e was a topic of complaint from the users previously. We should be hesitant about putting lim=134/268 jobs onto e-small. That said, with one of these BOINC-competition things firing up, feeding the grid is more important this week than strictly following rules, so I point this out just for future "normal" times. E-small is more likely to run out right now, so we feed it. I'll test-sieve a couple more jobs around SNFS-265 today, since it looks like the BOINCer army has arrived. Last fiddled with by VBCurtis on 2020-10-30 at 16:52 Reason: SFNS -> SNFS typo

2020-10-30, 22:05	#40
VBCurtis "Curtis" Feb 2005 Riverside, CA 2·5·7·83 Posts	During BOINC-challenge time, e-small was the right choice there for the job; with the benefit of hindsight we might just choose 134/200 or 134/180 for such a job, giving up a bit of speed to keep e-small fed without going too far over memory guidelines. Your job surely did less damage than my bad poly on 5_791m1!! Here's a corrected job for that number, with the full composite on the "n:" line: Code: n: 111812867780754346560415289999565367225454303594734581562227132753730403501180897472097054990558568023911409057075955845067437060818207472286533090719208303228335541944193280571602322588666038416993870470159838517777152295693 type: snfs skew: 0.87 deg: 6 c6: 5 c0: -2 Y1: -1 Y0: 5444517870735015415413993718908291383296 rlim: 67000000 alim: 50000000 lpbr: 32 lpba: 31 mfbr: 59 mfba: 58 rlambda: 2.7 alambda: 2.7 As noted previously, please sieve Q=10-72M on the -a side. A particularly lousy time to make such a mistake... :-/ EDIT: I'll PM Greg to kill the first try at this job. Last fiddled with by VBCurtis on 2020-10-30 at 22:07

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
System management notes	kriesel	kriesel	10	2021-09-01 15:56
Run down the queue on MPRIME without quitting GIMPS	Rodrigo	Software	7	2018-05-25 13:26
Improving the queue management.	debrouxl	NFS@Home	10	2018-05-06 21:05
split a prime95 queue & client installation	joblack	Information & Answers	1	2009-01-06 08:45
Power Management settings	PrimeCroat	Hardware	3	2004-02-17 19:11

2020-10-30, 20:46	#38
VBCurtis "Curtis" Feb 2005 Riverside, CA 16B2₁₆ Posts	We like 14e empty, and C172 is definitely within e-small guidelines. Good choice.

2020-10-31, 03:20	#41
frmky Jul 2003 So Cal A57₁₆ Posts	I requeued it.