mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Msieve (https://www.mersenneforum.org/forumdisplay.php?f=83)
-   -   Polynomial Request Thread (https://www.mersenneforum.org/showthread.php?t=18368)

VBCurtis 2013-07-08 18:28

5 days on a 460M for Lorgix' second C157 has produced no strong polys. Here's the best, with msieve suggesting 2.01 to 2.31 as the expected range:
[CODE]# norm 2.916149e-015 alpha -7.119644 e 1.883e-012 rroots 5
skew: 1639936.64
c0: 8532868550138516311046827336336400640
c1: 70998128199694820805749331912482
c2: -52696319235656576134337513
c3: -75162003432935559781
c4: 19930163724423
c5: 8036784
Y0: -860303108391035434159057424509
Y1: 44662544129823973
[/CODE]
I'm starting 39^158+1, the second of Chris' C155s. I'll let it run while I'm in Vegas the next 4 days, and report a poly Saturday for it. Once that's done I'll return to the rest of Lorgix' numbers. In the meantime, could someone beat this poly please?

lorgix 2013-07-08 18:35

Thanks for the polynomials! I haven't gotten around to poly selection on GPU yet, so this saves days of CPU-time.

Hopefully someone will improve on the second one.

You seem to know your way around the parameters, so I guess this is what's called 'bad luck'.

I might get a used GTX560Ti or something like that (ideas?) so that I have some GPU-power to experiment with.

VBCurtis 2013-07-08 18:49

You're quite welcome. Now that I know you don't have a GPU, I volunteer to find you at least decent polys for the next few months of your factoring. They may not hit msieve's GPU-target range every time, but doing your poly selects beats LLR-cuda as a good use of my GPU.

Repeated searches in the C155+ region give me some incentive to learn how to test-sieve, which leads to fiddling with the knobs as Jason puts it. All useful knowledge, and possibly heading for a beginner-guide mid-level GNFS entry.

GPU ideas: Go cheap, just get anything that supports the right flavor of CUDA. Even a low-end current card will be 50-100x faster for stage 1, while using fewer watts than a $100+ card.

firejuggler 2013-07-08 18:55

The GTX5xx are 2 generation behind. It might be hard to get "new" one. I guess a GTX 650 or 640 (if they are not only for laptop) might work for you.

lorgix 2013-07-08 19:31

[QUOTE=VBCurtis;345720]You're quite welcome. Now that I know you don't have a GPU, I volunteer to find you at least decent polys for the next few months of your factoring. They may not hit msieve's GPU-target range every time, but doing your poly selects beats LLR-cuda as a good use of my GPU.

Repeated searches in the C155+ region give me some incentive to learn how to test-sieve, which leads to fiddling with the knobs as Jason puts it. All useful knowledge, and possibly heading for a beginner-guide mid-level GNFS entry.

GPU ideas: Go cheap, just get anything that supports the right flavor of CUDA. Even a low-end current card will be 50-100x faster for stage 1, while using fewer watts than a $100+ card.[/QUOTE]
Well, I actually have a CUDA-capable GPU. But it is weak, and has few cores. And it is hooked up to my main screen.
There are currently five p^q-1-GNFS-candidates in the range 154-156. I could pretest some above 160 as well. I also have my eyes on a HP-number in the low 160s. So there's no shortage of candidates, if you're up for it.
I really should be getting a new GPU some time soon (in a few moths, at least). More powerful ones should be very useful for ECM. Does poly select saturate any GPU? (can it use any number of cores?)
Maybe I'll opt for a cheaper one; a GT630 based on GK208 (meaning 28nm) was released fairly recently.
Should I look for the latest Compute Capability? (I don't really know the significance of that)
[QUOTE=firejuggler;345721]The GTX5xx are 2 generation behind. It might be hard to get "new" one. I guess a GTX 650 or 640 (if they are not only for laptop) might work for you.[/QUOTE]
I was under the impression that the Fermi architecture was still very competitive in GPGPU. I'm not sure though.
The GTX650 should be close to peak performance/$ (in the sense of MHz*cores/$), in the 600-series, and so should be a good choice I suppose.
The 6- & 700-series use an entirely different architecture than the 4- & 500, so I don't really know how to compare them.
Specifically I'd like to know what correction factor I need in order to compare MHz*cores between Fermi and Kepler. In terms of poly select performance (and ECM performance).

firejuggler 2013-07-08 19:38

VBcurtis, what msieve version did you use to find that poly? it seems that Msieve150 and below sur-evaluate the poly score a bit.

VBCurtis 2013-07-08 20:55

1.51, svn 845 (I think). April 2013 download, premade binary. It should be the newer/lower E-values.

My GPU also drives my main screen, and the only time response has slowed was on the C210 sextic search. I've run cuda-LLR for 12 months straight, and now poly select, with no apparent effect on the screen. I suggest you get CUDA drivers installed and test things out- if nothing else, you can see how many hits you get overnight on the card. My mobile-GPU part isn't all that fast, and still can produce 8 hrs' worth of -nps work in just an hour or two. So, just overnight GPU -np1 runs might produce more than enough data for stage 2 needs.

Yes, I am game for at least a half-dozen candidates after those already posted by you and Chris.

firejuggler 2013-07-09 02:10

For now I have one *good* hit , score is equivalent to VBCurtis but skew is a bit lower
[code]
Tue Jul 09 02:19:43 2013 factoring 3787388118933512063249259173231597770049112435945032094640277541632460519045009491164974276597271170346757143570242916437315466248541421375702840232794402591 (157 digits)
Tue Jul 09 02:37:40 2013 R0: -1063466059487867106617472794654
Tue Jul 09 02:37:40 2013 R1: 64668823176851
Tue Jul 09 02:37:40 2013 A0: -3962907102148251811365883799925121155
Tue Jul 09 02:37:40 2013 A1: -989212107945990307577657150068
Tue Jul 09 02:37:40 2013 A2: 6390045909351316737980130
Tue Jul 09 02:37:40 2013 A3: -9592336224573633921
Tue Jul 09 02:37:40 2013 A4: 5859137926178
Tue Jul 09 02:37:40 2013 A5: 2784336
Tue Jul 09 02:37:40 2013 skew 1408059.13, size 2.287e-015, alpha -6.085, combined = 1.833e-012 rroots = 3
[/code]

wombatman 2013-07-09 02:14

For the 1st C155 listed, this is the best I've gotten thus far. It's a bit below the expected range, but not by too much:

[CODE]
expecting poly E from 2.62e-012 to > 3.02e-012

R0: -1982277912715175423679197808784
R1: 69837075648803257
A0: -44322257284138334787861242542456972633815
A1: 9286106559194619360096878548752378
A2: 108520072080442627462348267
A3: -10202649139317912268
A4: -61572653402
A5: 1740
skew 45960569.14, size 4.023e-015, alpha -7.466, combined = 2.541e-012 rroots = 5[/CODE]

firejuggler 2013-07-09 06:34

[code]
factoring 3787388118933512063249259173231597770049112435945032094640277541632460519045009491164974276597271170346757143570242916437315466248541421375702840232794402591 (157 digits)

R0: -1378574698621825663558022013546
R1: 214052541470369
A0: -175040708004969558266567903993405593325
A1: -220312394373520757850174317763772
A2: 49282988466830844032614321
A3: 25532977516064005444
A4: -3884830113524
A5: 760656
skew 3526801.51, size 2.601e-015, alpha -7.664, combined = 1.978e-012 rroots = 3
[/code]

VBCurtis 2013-07-13 17:19

Here is the poly for 39^158+1:
[CODE] # norm 5.644615e-015 alpha -7.931357 e 2.775e-012 rroots 3
skew: 1813413.83
c0: -26576343119721934097475724487682275175
c1: 58618135712883968651879051388681
c2: -21402685978619238283492841
c3: -24030041819615310289
c4: 13579142843224
c5: 4598160
Y0: -415948918490177005559476926358
Y1: 30809158838486467 [/CODE]

I am now working on the third of Chris' numbers. Or was it already done?
Edit: Ah, seems post 121 refers to this third number. I'll stop my search now, run npr on the overnight data, and report if I found a better poly.
Edit2: Nothing of use. I am now working on the 3rd number from post #85. Firejuggler, the 4th number from post 85 has had no work yet.


All times are UTC. The time now is 20:32.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.