mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Operation Kibibit (https://www.mersenneforum.org/forumdisplay.php?f=97)
-   -   Call for volunteers: RSA896 (https://www.mersenneforum.org/showthread.php?t=17460)

debrouxl 2012-12-28 07:20

1 Attachment(s)
I had a Xeon E3-1230 spend:
* more than a CPU-month on `msieve -np1` with norm 2e33, producing 13705 hits (attached) - I still can't use my GPU at the moment;
* a negligible amount of time on `msieve -nps` with norm 1e100;
* another CPU-month or so on `msieve -npr` for [I]all[/I] hits.

The second CPU month aimed at testing the updated version of poily's MPI patch (see post #42, with the modification mentioned in posts #49 and #50) for -nps + -npr - seems to work :smile:

The best polynomial produced by stage 2 is worse than the best hits posted earlier in this topic by a significant margin:
[code]# norm 3.453121e-20 alpha -10.768289 e 1.933e-19 rroots 2
skew: 113809421.14
c0: -777040021595652574801970561643527925908245389976845316223520
c1: -56689368246124675729566462839180299627446395002122964
c2: -1184036534621771421230391418974433516476565936
c3: 13945075463277536309744615078227615403
c4: 162343630391708833196793797373
c5: -762530856893636308781
c6: 1500024500100
Y0: -8062516096125044668396811397254458255803647
Y1: 34136309016942249143093[/code]I'll now make the computer run more stage 1.

Dubslow 2012-12-28 07:41

Just so y'all know, I've been running the last few days with my [STRIKE]460[/STRIKE] 560. I started at 5e12 with norm 2e33, and am averaging around 25K hits per day (give or take a few thousand, it's just a rough estimate). I have ~85.8K total at the moment, and ~100 more every few minutes.

jasonp, perhaps worker's locations should be marked in the first thread? Even if it's hard to duplicate work of the same coefficient, it still seems to me like we'll get better results if we run as many coefficients as possible.

frmky 2012-12-28 08:18

Here's another 2 million stage 1 hits with norm of 1e33. I've now restarted it with norm of 5e33.

[url]https://www.dropbox.com/s/vczyvz0bgmnrgvm/rsa896_2.dat.m.gz[/url]

firejuggler 2012-12-31 22:00

running msieve -v -np1 "stage1_norm=5e33 2345678901234,20000000000000" will post my hits/optimised once I get back home.

Dubslow 2012-12-31 23:39

How many S1 hits have there been, either from this thread alone or total? I'd estimate we're somewhere between 3-5 million, with Greg doing most of it?

I have ~180K hits [URL="http://dubslow.tk/random/batch1.gz"]here[/URL] (4.7 MB) (I would run size opt myself but my CPU's all tied up ATM).

RichD 2013-01-01 00:51

I've sent jasonp, via PM, just under 4M.

Dubslow 2013-01-01 02:24

[QUOTE=RichD;323256]I've sent jasonp, via PM, just under 4M.[/QUOTE]

Sweet! I guess there's more than one big player around here :smile:

So we're easily north of 5M, probably somewhere near 10M?

debrouxl 2013-01-01 08:51

By tomorrow, I'll have another dozen thousand CPU stage 1 hits (norm 2e33) from MPI msieve.

And yeah, we're probably around 10M hits now :smile:

firejuggler 2013-01-01 18:43

1 Attachment(s)
168xx hits min norm seem to be @ 4.xxx +034 but I may have done a bad job at optimising

RichD 2013-01-02 05:24

[QUOTE=Dubslow;323262]... I guess there's more than one big player around here ... probably somewhere near 10M?[/QUOTE]

Not really a big player, just consistent. Been running since my first post.

[QUOTE=jasonp;319061]A trillion hits would probably be too much to deal with, though it would be awesome :)[/QUOTE]

jasonp 2013-01-02 12:10

So far the total contributions amount to 10.4M hits. Keep 'em coming!

Edit: by comparison, the few months that everyone spent on RSA768 produced 30 million hits, the vast majority of which came from CPU Msieve.


All times are UTC. The time now is 04:56.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.