mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   NFS@Home (https://www.mersenneforum.org/forumdisplay.php?f=98)
-   -   BOINC NFS sieving - NFS@Home (https://www.mersenneforum.org/showthread.php?t=12388)

debrouxl 2014-07-01 19:59

The factors appeared thanks to msieve SVN r965, but the invocation was the same for both the faulty msieve SVN r966 runs (different target densities) and the successful msieve SVN r965 run:
[code]ionice -c3 ~/msieve/msieve -i GC_2_757.ini -s GC_2_757.dat.gz -nf GC_2_757.fb -nc target_density=90 -v -t 4[/code]

Tonight, with msieve SVN r967 invoked the same way as before, filtering the original raw dataset for (G)C_2_757 succeeded, discarding 53 composite relations. LA has now successfully started as well, currently at ~0.3% with ~42h ETA, I'll let it run at least until tomorrow morning.

Mini-Geek 2014-07-02 00:55

1 Attachment(s)
[URL="http://factordb.com/index.php?id=1000000000014251926"]L1902[/URL]:
[CODE]prp69 factor: 448770026822784402728514481476646952184358752399132088821837198314401
prp163 factor: 2158668299839204964177220860280160089054160108028268766415913271405733919803799561621194297561963745657006174295990940277990189527281507877632561174088825422119289[/CODE]
Log is attached. It took right at a week, as expected. The memory usage for most of the LA was around 10.4 GB.

pinhodecarlos 2014-07-02 02:04

Just a quick note. For Fibonacci and Lucas Factorizations please send an email to Marin Mersennus so he can update the tables (4 are already missing, the ones processed by Thomas Womack).

[URL]http://mersennus.net/fibonacci/[/URL]
[URL]http://mersennus.net/fibonacci/history.html[/URL]

I think it is time to my a little guide (sticky thread) to whom post-processing guys should send an email with factors after submitting the factors to [URL]http://factordb.com/[/URL].
Something like:

Fibonacci and Lucas Factorizations - Send factors by email to Marin Mersennus at marin dot mersennus at gmail dot com

Cullen and Woodall Factorizations (GCW)- Send factors by email to Paul Leyland at paul at leyland dot vispa dot com

XYYXF Factorizations - Send factors by email to YYY

Aliquot Factorizations - Send factors by email to ZZZ

OddPerfect Factorizations - Send factors by email to William

etc


Carlos

fivemack 2014-07-02 04:55

3270.692 done
 
[code]
Wed Jul 2 05:08:46 2014 prp65 factor: 26104815813483108835535737500572575760729863017539546098929503971
Wed Jul 2 05:08:46 2014 prp111 factor: 703208770372701754201233527160876951691285989719117138065477844378890314944280482644067125018481775627630422297
[/code]

34.5 hours for 8.6M matrix (target_density=120) on i7/4930K -t6 with svn952

Log at [url]http://pastebin.com/Kk1VtmQ2[/url] ; it took eleven dependencies to get a nontrivial square root

With 455.1 million relations, it was also possible to run target_density=150, which gave an 8.0M matrix. Running it for a couple of hours at -t 6 suggested that a full run would take about 35 hours, which sounds as if we may have gone through the optimum density.

fivemack 2014-07-03 09:53

GW_11_219 done
 
[code]
Wed Jul 2 20:32:01 2014 prp79 factor: 1111311316618589879708928196739584948763935183044655049280766013394787738757327
Wed Jul 2 20:32:01 2014 prp113 factor: 42934460362066500167065925004394957756913613111999126336558671696666934817362174794343354829669816609500980859347
[/code]

51.5 hours for 7.9M matrix on i7/2600 -t3

Log at [url]http://pastebin.com/GRLMWKLi[/url]

fivemack 2014-07-03 18:04

I am tempted to take GC_7_292 and GC_3_516, download the current data, and throw 48 cores of Opteron at doing a bit more sieving with 15e and probably 3LP down at Q~100M to get the yield up a bit. I think that's a better use of electricity than continuing sieving with the current amazingly low yields.

debrouxl and frmky, would you be OK with this?

Current position:

[code]
7C292
Thu Jul 3 17:57:53 2014 found 33556973 hash collisions in 120036767 relations
Thu Jul 3 18:05:41 2014 found 40380979 duplicates and 80876128 unique relations
Thu Jul 3 18:31:09 2014 begin with 80876128 relations and 79926683 unique ideals
Thu Jul 3 18:32:00 2014 reduce to 44271755 relations and 38868073 ideals in 25 passes
Thu Jul 3 18:49:33 2014 begin with 44271755 relations and 48871946 unique ideals
Thu Jul 3 18:50:40 2014 reduce to 44241787 relations and 48841972 ideals in 16 passes
(not enough)
3C516
Thu Jul 3 19:01:05 2014 found 30265028 hash collisions in 114189683 relations
Thu Jul 3 19:08:05 2014 found 35014846 duplicates and 80394517 unique relations
Thu Jul 3 19:25:48 2014 begin with 80394517 relations and 78913758 unique ideals
Thu Jul 3 19:26:24 2014 reduce to 42387306 relations and 35975518 ideals in 19 passes
Thu Jul 3 19:41:22 2014 begin with 42387306 relations and 45903758 unique ideals
Thu Jul 3 19:42:13 2014 reduce to 42359184 relations and 45875632 ideals in 13 passes
(not enough)
[/code]

fivemack 2014-07-04 16:36

2340.672 should be done Monday evening

debrouxl 2014-07-04 20:05

No objection from me on the plan you've outlined in your post from yesterday evening, I've reserved GC_7_292 and GC_3_516 for you.
Sieving for GC_3_516 was done up to "only" 340M for now, so we [i]could[/i] probably still get a matrix with 14e alone, without having to sieve to such heights as the current 480M for GC_7_292.

swellman 2014-07-05 19:32

I'll take GC_4_410 if it's available, once sieving is complete.

swellman 2014-07-07 19:09

I'll grab C169_123_109 once sieving is completed.

Eta for GC_4_410 is next Wednesday.

fivemack 2014-07-07 20:07

C179_2340_732 done
 
[code]
Mon Jul 7 18:09:18 2014 prp66 factor: 248635147377542896113261749274786147758650114902431479778854478749
Mon Jul 7 18:09:18 2014 prp113 factor: 60975148432651109377757083060587480547895242027357946487127667414482584280459095338341228292954632498997540145211
[/code]

73 hours for 12.0M matrix (target_density=120) on i7/4930K -t6

Log at [url]http://pastebin.com/iqmKd89p[/url] - note failed filtering with 270.7M and 292.5M relations before success with 389M.


All times are UTC. The time now is 22:58.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.