[QUOTE=VBCurtis;486766]Sean-
How about starting at 8M? I'm about done with 1-2M and will continue to 3M. Rich has 3-5M covered (or was it 4-6M?) and is likely to continue beyond that.[/quote]

Certainly. I’ll run C205 starting with c5=8M. I’ll go until the search is called off by consensus.

[quote]My search recipe is to divide default stage 1 norm by 10 to 12, while the stage 2 norm can be divided by 30 for smaller jobs; for something this size I experimented to find a stage 2 norm that produced 40-60 hits per day. I run root-opt every 2-3 days.
I believe on these large jobs stage 1 norm should not be set so tightly that msieve stops searching a subset of the range.[/QUOTE]

Could be - I do fear leaving good polys in my wake with an overly tight stage 1 norm and my results in these high ranges do not disabuse me of this notion. While my stated recipe does well in the lower sized composites, above C199 it’s been not so good (read terrible). I’ll pull back a bit on the stage_1 limit.

Just a follow up on the C205 - I tweaked my recipe a bit and have gotten better speed with about 200 hits/day. Ran it over the range 8M-8.08M in 15 hours and a top hit above 1.6. Not a great result but better than I was seeing with C200.

Currently plowing through 8-9M, hoping to finish it by next weekend. Restating that I will run this poly search until group consensus agrees we are finished. Might be late summer!:smile:

I have in mind something like another month for each of us on the C205; I think we each spent ~3 weeks on C200, so a total of 17-20 GPU weeks on C205 would be double the effort for a job expected to be twice as hard to sieve.

I think we're also expecting a poly score half as big; a 2.2 will be good and a 2.3 will be terrific for this candidate. I'll be tempted to quit the search early if we luck into a 2.3x somewhat quickly.

I'll be pausing the search here and there for other poly-select tasks.

I reached 3M on the C200, completing my GPU search. I ran CADO from 500k to 1.15M as well. I'll start that GPU on C205 at 2M this evening.
EDIT: Noticed GPU2 is nearing 2M already, so GPU1 will take 7-8M on the C205.

C205
 

My best poly for the C205:
[code]
# norm 4.360225e-20 alpha -8.482317 e 2.160e-15 rroots 3
skew: 576643112.81
c0: -49831972907768980741847754899694896009872578368904
c1: -1208252898570059997960254995437853104752250
c2: 2301054153094287847095810813500019
c3: -768701320508910349722448
c4: 20217122428639892
c5: 10090080
Y0: -3417765234685637144252856994227250140533
Y1: 52603066613593727963[/code]

C205 poly
 

[QUOTE=Gimarel;486920]My best poly for the C205:
[code]
# norm 4.360225e-20 alpha -8.482317 e 2.160e-15 rroots 3
skew: 576643112.81
c0: -49831972907768980741847754899694896009872578368904
c1: -1208252898570059997960254995437853104752250
c2: 2301054153094287847095810813500019
c3: -768701320508910349722448
c4: 20217122428639892
c5: 10090080
Y0: -3417765234685637144252856994227250140533
Y1: 52603066613593727963[/code][/QUOTE]
That's a new record! Great job, Gimarel! It could be a bit better with 8 times smaller c5:
[code]
# norm 4.351299e-20 alpha -8.482473 e 2.175e-15 rroots 3
skew: 1154570345.17
c0: -198771074233024288059859432513101941810047271748000
c1: -2417566055475912315880932883284311864116300
c2: 2301321393253189240440876621682995
c3: -389007449816032310022648
c4: 5052827812216373
c5: 1261260
Y0: -6835530469371286406779375895268798573302
Y1: 52603066613593727963
[/code]

The score may be a bit better, but the skew is much higher.
If you substitute the skew with the skew of the original poly, the score becomes lower.

The ggnfs sievers limit the skew to the special-q that is sieved, so the poly with the lower skew seems better to me. But maybe testsieving proves me wrong.

I need a poly for the 11040:i10080, here is my best so far. I assume it'll run on 14e queue because 15e is pretty big ATM.

[CODE]# norm 1.052085e-16 alpha -7.985825 e 2.606e-13 rroots 3
n: 1773609810640326137506048934005720326001842445891839878160657321987999204446671681151317000989104105836960808112036515218167435411817552189170383677070033522690218104145549
skew: 78814766.32
c0: -4558081567105030790214396465597896874454805
c1: -1660751272020321736972482556298515782
c2: 383825676724416825221147215
c3: -46817502651796682034
c4: 11630953749286
c5: 34320
Y0: -2201213421276728294659199649512778
Y1: 636233426236388287
rlim: 134000000
alim: 134000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.6
alambda: 2.6
[/CODE]

[QUOTE=Gimarel;486920]My best poly for the C205:
[code]
# norm 4.360225e-20 alpha -8.482317 e 2.160e-15 rroots 3
skew: 576643112.81
[/code][/QUOTE]
What range did you search?

This poly came from a search with msieve on CPU. The leading coefficients were a multiple of 120120. The search started at 10090080 and ended at 10570560.

I do search a few other selected leading coefficients.

C172 poly
 

[QUOTE=unconnected;486925]I need a poly for the 11040:i10080, here is my best so far. I assume it'll run on 14e queue because 15e is pretty big ATM.

[CODE]# norm 1.052085e-16 alpha -7.985825 e 2.606e-13 rroots 3
n: 1773609810640326137506048934005720326001842445891839878160657321987999204446671681151317000989104105836960808112036515218167435411817552189170383677070033522690218104145549
skew: 78814766.32
c0: -4558081567105030790214396465597896874454805
c1: -1660751272020321736972482556298515782
c2: 383825676724416825221147215
c3: -46817502651796682034
c4: 11630953749286
c5: 34320
Y0: -2201213421276728294659199649512778
Y1: 636233426236388287
rlim: 134000000
alim: 134000000
lpbr: 31
lpba: 31
mfbr: 62
mfba: 62
rlambda: 2.6
alambda: 2.6
[/CODE][/QUOTE]
If you are planning to test-sieve it, below is the one to compare it to. I'll post more if they pop up.
[code]
# norm 9.462739e-17 alpha -8.090827 e 2.459e-13 rroots 3
skew: 130476842.35
c0: -56278469432930368674449473430682146320145344
c1: -2891567916845332428354889462570246062
c2: 2181732010706563780821734321
c3: -153868280263726979898
c4: 5614582094843
c5: 8580
Y0: -4402426845532849924170585025462950
Y1: 636233426236388287
[/code]

C172 poly
 

Another one (skew should be adjusted):
[code]
Y0: -4402426834857753938455540295417342
Y1: 636233426236388287
c0: -7127964805653396655921588903694243919997200
c1: -2838824185316088581935755119099863710
c2: 4325667169106708664020901473
c3: 247105160542815794150
c4: 6334383348443
c5: 8580
skew: 78529324.433
# lognorm 54.91, E 46.77, alpha -8.14 (proj -1.63), 3 real roots
# MurphyE = 2.50e-13
[/code]