BOINC NFS sieving - NFS@Home - Page 89

frmky · 2013-10-28, 17:46

No. It terminated before the end of the file. I'll look into it.

Batalov · 2013-10-28, 18:17

Quote:

Originally Posted by wombatman

Here's what I get running your command (limited to the end part, of course), Frmky:

Code:

Mon Oct 28 12:20:16 2013  31.5M unique relns  1.59M duplicate relns (+0.05M, avg D/U ratio in block was 9.4%)
Found 31809992 unique, 1617420 duplicate (4.8% of total), and 181942 bad relations.
Largest dimension used: 147 of 800
Average dimension used: 97.1 of 800
Terminating program at Mon Oct 28 12:20:18 2013

Does that seem right to you?

Based on this report, you don't need dim value (hash depth) to be above 150. (you may have run out of memory with dim=800.)

Could you retry with
zcat rels.dat.gz | remdups4 150 -v > msieve.dat

wombatman · 2013-10-28, 18:32

Here's the result:

Code:

Found 31809992 unique, 1617420 duplicate (4.8% of total), and 181942 bad relations.
Largest dimension used: 147 of 150
Average dimension used: 97.1 of 150
Terminating program at Mon Oct 28 13:31:32 2013

Just confirmed that the remdups binary works properly on an older, smaller relations file (17M or so relations). It found exactly the same number of duplicates that msieve did.

VictordeHolland · 2013-10-28, 19:34

I got my relations to work with:

Code:

remdups 600 GC_12_206.dat msieve.dat

Copied the remdups.c from the link into the cross platform IDE "Code::Block" and pressed compile. "Code::Block" used "mingw32-gcc" with parameter "-march=core2" for compiling I believe, so it's 32bit unfortunately. "800" therefore gave me an out of memory message, but "600" worked just fine too. I attached the file, hope it runs for others too. Linear Algebra is now at 5.8% .

wombatman · 2013-10-28, 22:53

I tried Victor's exe and a rebuilt exe using both "-m64" and "-march=corei7". All of them end at the same point with the same number of dupes and unique relations.

VictordeHolland · 2013-10-29, 00:56

31.8M unique relations out of 120M+ relations indeed seems incorrect.
Wombatman are you able to construct a matrix and proceed with the LA, or are you stuck at the relations filtering?

wombatman · 2013-10-29, 01:22

Stuck on the filtering as far as I can tell unless there's a way to skip that part.

frmky · 2013-10-29, 01:39

Worked for me. Odd...

Uploading msieve.dat.gz now. Should be ready to download in a half hour or so.

wombatman · 2013-10-29, 03:16

Very strange. Maybe a Windows/Linux line ending issue or something? Don't know why it would pop up only now.

Batalov · 2013-10-29, 06:00

One doesn't simply ~~walk into Mordor~~ debug Windows.

xilman · 2013-10-29, 09:22

Ben & Victor:

Could I ask you to email me your NFS@Home results in future please? Somewhat more trouble for you but much more reliable and much less hassle for me than the alternatives. I thjnk I managed to scrape your factors from the NFS@Home site and the factordb can't be certain that I got all the details correct. Neither are the reports made particularly promptly.

Thanks!

The GCW web pages will be updated later today, all being well.

Paul

2013-10-28, 18:32	#971
wombatman I moo ablest echo power! May 2013 13×137 Posts	Here's the result: Code: Found 31809992 unique, 1617420 duplicate (4.8% of total), and 181942 bad relations. Largest dimension used: 147 of 150 Average dimension used: 97.1 of 150 Terminating program at Mon Oct 28 13:31:32 2013 Just confirmed that the remdups binary works properly on an older, smaller relations file (17M or so relations). It found exactly the same number of duplicates that msieve did. Last fiddled with by wombatman on 2013-10-28 at 18:48

2013-10-29, 00:56	#974
VictordeHolland "Victor de Hollander" Aug 2011 the Netherlands 2³×3×7² Posts	31.8M unique relations out of 120M+ relations indeed seems incorrect. Wombatman are you able to construct a matrix and proceed with the LA, or are you stuck at the relations filtering? Last fiddled with by VictordeHolland on 2013-10-29 at 00:58

2013-10-29, 06:00	#978
Batalov "Serge" Mar 2008 Phi(4,2^7658614+1)/2 9,497 Posts	One doesn't simply ~~walk into Mordor~~ debug Windows. Attached Thumbnails

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Boinc Statistics for NFS@Home borked ?	thomasn	NFS@Home	1	2013-10-02 15:31
BOINC NFS sieving - RSALS	debrouxl	NFS@Home	621	2012-12-14 23:44
BOINC?	masser	Sierpinski/Riesel Base 5	1	2009-02-09 01:10
BOINC?	KEP	Twin Prime Search	212	2007-04-25 10:29
BOINC	bebarce	Software	3	2005-12-15 18:35

2013-10-28, 17:46	#969
frmky Jul 2003 So Cal 2²×3²×59 Posts	No. It terminated before the end of the file. I'll look into it.

2013-10-28, 22:53	#973
wombatman I moo ablest echo power! May 2013 13·137 Posts	I tried Victor's exe and a rebuilt exe using both "-m64" and "-march=corei7". All of them end at the same point with the same number of dupes and unique relations.

2013-10-29, 01:22	#975
wombatman I moo ablest echo power! May 2013 13×137 Posts	Stuck on the filtering as far as I can tell unless there's a way to skip that part.

2013-10-29, 01:39	#976
frmky Jul 2003 So Cal 2²×3²×59 Posts	Worked for me. Odd... Uploading msieve.dat.gz now. Should be ready to download in a half hour or so.

2013-10-29, 03:16	#977
wombatman I moo ablest echo power! May 2013 13·137 Posts	Very strange. Maybe a Windows/Linux line ending issue or something? Don't know why it would pop up only now.

2013-10-29, 09:22	#979
xilman Bamboozled! "𒉺𒌌𒇷𒆷𒀭" May 2003 Down not across 10786₁₀ Posts	Ben & Victor: Could I ask you to email me your NFS@Home results in future please? Somewhat more trouble for you but much more reliable and much less hassle for me than the alternatives. I thjnk I managed to scrape your factors from the NFS@Home site and the factordb can't be certain that I got all the details correct. Neither are the reports made particularly promptly. Thanks! The GCW web pages will be updated later today, all being well. Paul