Thank you VBCurtis! I’ll run some test sieving.

I need a poly for C175 from 11040:i10071.
[CODE]n: 6484689970303129020517057103365894793216912303102493240993925500641398178923457820535718262847740804273820043327446358642419268046237801717129218758317768624690555974293740733
[/CODE]My top3 polys selected with YAFU:



[CODE]# norm 4.035263e-17 alpha -8.851023 e 1.436e-13 rroots 5
skew: 65219425.07
c0: -449534111689309867074964193519151777172539105
c1: 25231819248819016723754441786731581483
c2: 751643722621067850341305990825
c3: -30118940469212794494451
c4: -247836514699712
c5: 925680
Y0: -5876052790831920271062058409489722
Y1: 5426620801011945613[/CODE][CODE]# norm 3.948039e-17 alpha -8.412913 e 1.435e-13 rroots 3
skew: 45703855.84
c0: -96325606841672808635154298191532973371726080
c1: -115990067080939429387448618880048848
c2: 392690905993394826587110458596
c3: -5876921656850812985468
c4: 33337968483471
c5: 1634040
Y0: -5244756218842432827594087985964773
Y1: 638224320659838283[/CODE][CODE]# norm 3.889138e-17 alpha -9.343686 e 1.409e-13 rroots 3
skew: 94781238.28
c0: -1988140802798733403389611175557189568702282240
c1: -21556439140549492999887403559052687132
c2: 1815756715132146915463530764020
c3: -10455282893178476394131
c4: -326999349205712
c5: 925680
Y0: -5876052883647295864642087769742017
Y1: 5426620801011945613
[/CODE]

[QUOTE=unconnected;472005]I need a poly for C175 from 11040:i10071.
[CODE]n: 6484689970303129020517057103365894793216912303102493240993925500641398178923457820535718262847740804273820043327446358642419268046237801717129218758317768624690555974293740733
[/CODE]My top3 polys selected with YAFU:



[CODE]# norm 4.035263e-17 alpha -8.851023 e 1.436e-13 rroots 5
skew: 65219425.07
c0: -449534111689309867074964193519151777172539105
c1: 25231819248819016723754441786731581483
c2: 751643722621067850341305990825
c3: -30118940469212794494451
c4: -247836514699712
c5: 925680
Y0: -5876052790831920271062058409489722
Y1: 5426620801011945613[/CODE][CODE]# norm 3.948039e-17 alpha -8.412913 e 1.435e-13 rroots 3
skew: 45703855.84
c0: -96325606841672808635154298191532973371726080
c1: -115990067080939429387448618880048848
c2: 392690905993394826587110458596
c3: -5876921656850812985468
c4: 33337968483471
c5: 1634040
Y0: -5244756218842432827594087985964773
Y1: 638224320659838283[/CODE][CODE]# norm 3.889138e-17 alpha -9.343686 e 1.409e-13 rroots 3
skew: 94781238.28
c0: -1988140802798733403389611175557189568702282240
c1: -21556439140549492999887403559052687132
c2: 1815756715132146915463530764020
c3: -10455282893178476394131
c4: -326999349205712
c5: 925680
Y0: -5876052883647295864642087769742017
Y1: 5426620801011945613
[/CODE][/QUOTE]

My GPU is only a GTX 660 (old/slow) but it can likely produce a better poly than Yafu with a CPU. Will try tonight, results in a day or two.

Some early hits. I tried small coefficients for a short time just to see what would pop out.
[CODE]# norm 4.125469e-017 alpha -8.900162 e 1.449e-013 rroots 5
skew: 1455572238.87
c0: 1255951937143183892247587563448703306283587744
c1: 1610562828004031564735677166769887509440
c2: -2587378187602388086616529399098
c3: -8776620964640000450927
c4: 1062089487868
c5: 228
Y0: -30959281516969606515538242506506235
Y1: 61387989459757867

# norm 4.033509e-017 alpha -7.758342 e 1.416e-013 rroots 5
skew: 335209405.03
c0: -1877089398285646595575713589530894513678037360
c1: 21580558545552247740323991835103574800
c2: 443140388707160344149515351887
c3: -4195992110369430155825
c4: -4627682860947
c5: 600
Y0: -25512181371732002625413634242206431
Y1: 11955329559207683

# norm 3.899808e-017 alpha -7.662554 e 1.387e-013 rroots 5
skew: 331470702.63
c0: -399591888047527503672089903033880935333080525
c1: 14949818605279881695292579336505180355
c2: 527774355335310572226910953562
c3: -4069352063341906248845
c4: -4648161865947
c5: 600
Y0: -25512181371813613710020188220938236
Y1: 11955329559207683

# norm 3.784764e-017 alpha -8.014537 e 1.385e-013 rroots 5
skew: 583222572.53
c0: -16465543459384525592003479010867553391519828320
c1: 60780120096652329881167514910078682906
c2: 1746503451365672636510945071495
c3: 926831799121457179447
c4: -5394412342947
c5: 600
Y0: -25512181374787503838770499281209833
Y1: 11955329559207683

# norm 3.911842e-017 alpha -7.631940 e 1.385e-013 rroots 5
skew: 320268316.96
c0: -67968006702340873907618558994465707256146659
c1: 14831568914972281955763683118300369731
c2: 529139867967078359061114922264
c3: -4067271805652871175829
c4: -4648497511947
c5: 600
Y0: -25512181371814951296201931494927642
Y1: 11955329559207683

# norm 3.887526e-017 alpha -7.643214 e 1.382e-013 rroots 5
skew: 334156025.38
c0: -1912492952474814339546852085839170659366350525
c1: 22009575103169310440335007853753688531
c2: 436997896441859312851612407370
c3: -4205013574222750736333
c4: -4626220537947
c5: 600
Y0: -25512181371726175107617964489997228
Y1: 11955329559207683[/CODE]

From the [url=http://www.mersenneforum.org/showpost.php?p=457286&postcount=59]Best Polys Table[/url] the best score found to date for a C175 is 1.800e-13. With my outdated hardware, I’ll be very happy to find any e > 1.65e-13.

[QUOTE=swellman;472094]From the [url=http://www.mersenneforum.org/showpost.php?p=457286&postcount=59]Best Polys Table[/url] the best score found to date for a C175 is 1.800e-13. With my outdated hardware, I’ll be very happy to find any e > 1.65e-13.[/QUOTE]

I just got 1.91e-13 for 13*2^848-1, also a C175. I have a free GPU presently; if requester has patience until Thursday, I can give it a couple GPU-days of search. If Sean posts his search range, I'll extend it upward a few million.

My search is continuing. I'm working from A5=2e6.
Apparently a job this size has 30 slices, so overlapping ranges wouldn't be much of a problem.

Curtis, what settings are you using for a C175?

[QUOTE=lorgix;472106]My search is continuing. I'm working from A5=2e6.
Apparently a job this size has 30 slices, so overlapping ranges wouldn't be much of a problem.

Curtis, what settings are you using for a C175?[/QUOTE]

I divide stage 1 norm by 10 or so, which leaves just one or two slices; overlapping ranges is a problem for my settings. Generally, for any number 165+, I try a few stage 1 norms near the default divided by ten, looking for the fastest rate of hit-generation with stage 2 norm better than default/30. I run -npr on all such hits, usually around 100 per day. If I get more than 150 hits per day, I tighten the stage2 norm a bit.
When I run -npr, I use default settings.

My best overnight
[code]
R0: -11081373450779752303630745969260644
R1: 20873457748874587
A0: -81836203280471835744157447593921935421125
A1: -1967787734761642912005459062564610615
A2: -967689896601975648676261529
A3: 1144234060063424373879
A4: 63635741062
A5: 38808
skew 62883697.54, size 3.720e-17, alpha -7.156, combined = 1.581e-13 rroots = 3
[/code]

[QUOTE=VBCurtis;472099]If Sean posts his search range, I'll extend it upward a few million.[/QUOTE]

I used the following line to invoke the poly search
[code]
msieve -v -g 0 -np1 -nps "stage1_norm=3.5e25 stage2_norm=3.73e23" -s output[/code]

So basically msieve ran from 12 to ~86000 before I killed it. Clearly I am not well versed in the finer points of msieve poly select (e.g. search range).

eta: I'm stepping back from this effort - my hardware just can’t keep up. Lorix and VBCurtis will doubtless surpass my score in a few hours.

[QUOTE=swellman;472136]
I used the following line to invoke the poly search
[code]
msieve -v -g 0 -np1 -nps "50000,1000000 stage1_norm=3.5e25 stage2_norm=3.73e23" -s output[/code]

So basically msieve ran from 12 to ~86000 before I killed it. Clearly I am not well versed in the finer points of msieve poly select (e.g. search range).

eta: I'm stepping back from this effort - my hardware just can’t keep up. Lorgix and VBCurtis will doubtless surpass my score in a few hours.[/QUOTE]
OK, I'm running from 86000 to 2M. I chose the same stage 1 norm as swellman, and 3.2e23 for stage2. Searching tiny coefficients is time-consuming; 0 to 86000 is probably as much effort as 100-500k.

I edited the quoted command to show how to force msieve to start at a specific coeff, in case it's useful to others.

I have only found 1.494e-013, 1.492 and so on. Eight at or above 1.464.

Curtis
For smaller jobs I have found that I can reduce stg1norm by a factor of three or more.
I think I get the most -nps hits when the GPU keeps the CPU busy.
Have you had a similar experience?
For this job I seem to get the most hits when using something like stg1norm=2.25e26.
Does the stg2norm used to determine the "winning" stg1norm matter a lot? Because I tend to set it closer to default. (I sort the output, of course. And I try to kill -npr after a reasonable amount of time)
Have I reached a point where the GPU should be more particular (right word?) even if that means that a CPU-core that could be doing -nps will be idle?