2^821-1 - Page 2

fivemack · 2008-05-14, 18:16

Yes, that's right. You do need to specify a new output filename - after one embarrassing incident, I hacked the gnfs-lasieve that I use at home to fail if the output file exists, and there's at least one version which conveniently uses as default filename {polynomial file}.lasieve-{side}.{start}-{end}.

You're right: concatenate the relations files, gzip and upload when you're finished. Good luck with the sieving!

bdodson · 2008-05-14, 18:22

Quote:

Originally Posted by fivemack

70-100M on the R side would be fantastic, thanks!

Let's make that 70M-100M A+R; adjusting my scripts for writing condor
submit files to use "-a" wasn't as bad as I thought. -Bruce

(I'm alternating 10M ranges for 821- and 949+, 70M-80M "R" was just
about done.)

mdettweiler · 2008-05-14, 20:22

I was just thinking: should I be using gnfs-lasieve4I15.exe instead of gnfs-lasieve4I14.exe? I know I15 produces more relations in a given range, but would it make my range take longer?

bsquared · 2008-05-14, 20:26

Quote:

Originally Posted by Anonymous

I was just thinking: should I be using gnfs-lasieve4I15.exe instead of gnfs-lasieve4I14.exe? I know I15 produces more relations in a given range, but would it make my range take longer?

Yes, roughly twice as long. I believe the use of 15e has been requested, yes.

mdettweiler · 2008-05-14, 20:31

Quote:

Originally Posted by bsquared

Yes, roughly twice as long. I believe the use of 15e has been requested, yes.

Hmm...okay. I'm at about 2.5% through my range right now; I'll switch over to 15e shortly.

mdettweiler · 2008-05-14, 21:40

Considering the 2x slowdown from using 15e for my range, I think I'll only do 40-40.5 instead of 40-41. This should bring the time to complete the range back in line with what I had originally planned for.

smh · 2008-05-16, 13:17

Extended my own reservation to 45M on both sides

Andi47 · 2008-05-17, 10:04

Hmmmm... Where did I say that I reduce my reserved range from 40-41M R to 40-40.5M R?

I still plan to do 40-41M on the R side.

Edit: @ fivemack: I think that you have entered the reducing of Anonymous' range in post #17 into the wrong line.

fivemack: Yes, I did, sorry; fixed now

mdettweiler · 2008-05-22, 18:18

40M-40.5M on the A side is complete, now uploading.

Edit: Upload complete.

mdettweiler · 2008-05-24, 04:22

I'll do 45M-45.5M on the R side, if nobody minds me having to delay starting it for a couple of days.

fivemack · 2008-05-24, 09:25

Thanks to the enormous resources that bdodson has applied to sieving, we're doing pretty well with relations; I'm doing a run at the moment which I expect to generate a matrix. Further relations will reduce the size of the matrix, and I suspect reduce the run-time by more than the time it takes to produce the relations, so we should definitely keep sieving for the next couple of weeks.

I intend to start the matrix job on or about June 7th, so please don't make reservations that you are not sure you will be able to complete by then; I'll do a final call for uploads of relations from unfinished ranges on June 5th.

2008-05-17, 10:04	#19
Andi47 Oct 2004 Austria 2482₁₀ Posts	Hmmmm... Where did I say that I reduce my reserved range from 40-41M R to 40-40.5M R? I still plan to do 40-41M on the R side. Edit: @ fivemack: I think that you have entered the reducing of Anonymous' range in post #17 into the wrong line. fivemack: Yes, I did, sorry; fixed now Last fiddled with by fivemack on 2008-05-17 at 11:27

2008-05-22, 18:18	#20
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3·2,083 Posts	40M-40.5M on the A side is complete, now uploading. Edit: Upload complete. Last fiddled with by mdettweiler on 2008-05-22 at 19:05

2008-05-24, 09:25	#22
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 2³·11·73 Posts	Dates for the matrix job Thanks to the enormous resources that bdodson has applied to sieving, we're doing pretty well with relations; I'm doing a run at the moment which I expect to generate a matrix. Further relations will reduce the size of the matrix, and I suspect reduce the run-time by more than the time it takes to produce the relations, so we should definitely keep sieving for the next couple of weeks. I intend to start the matrix job on or about June 7th, so please don't make reservations that you are not sure you will be able to complete by then; I'll do a final call for uploads of relations from unfinished ranges on June 5th.

2008-05-14, 18:16	#12
fivemack (loop (#_fork)) Feb 2006 Cambridge, England 2³·11·73 Posts	Yes, that's right. You do need to specify a new output filename - after one embarrassing incident, I hacked the gnfs-lasieve that I use at home to fail if the output file exists, and there's at least one version which conveniently uses as default filename {polynomial file}.lasieve-{side}.{start}-{end}. You're right: concatenate the relations files, gzip and upload when you're finished. Good luck with the sieving!

2008-05-14, 20:22	#14
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	I was just thinking: should I be using gnfs-lasieve4I15.exe instead of gnfs-lasieve4I14.exe? I know I15 produces more relations in a given range, but would it make my range take longer?

2008-05-14, 21:40	#17
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	Considering the 2x slowdown from using 15e for my range, I think I'll only do 40-40.5 instead of 40-41. This should bring the time to complete the range back in line with what I had originally planned for.

2008-05-16, 13:17	#18
smh "Sander" Oct 2002 52.345322,5.52471 29·41 Posts	Extended my own reservation to 45M on both sides

2008-05-24, 04:22	#21
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3·2,083 Posts	I'll do 45M-45.5M on the R side, if nobody minds me having to delay starting it for a couple of days.