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 2013-04-01 10:55

Target density can now be defined interactively on the command line :smile:
-nc1 "target_density=X".

pinhodecarlos 2013-04-01 11:00

[QUOTE=debrouxl;335681]Target density can now be defined interactively on the command line :smile:
-nc1 "target_density=X".[/QUOTE]

I didn't know that Lionel. Thank you.
Which target density should I use for SF1229, 100...130?

Carlos

debrouxl 2013-04-01 11:29

That's a 15e task, even if it used 31-bit LPs, so I'm not sure I'm knowledgeable (but this is an occasion to learn)...

I'll venture a guess: for 14e, we usually use 210M-220M raw relations for 31-bit LPs. If that's true for 15e tasks as well, than the ~288M raw relations that I can see in NFS@Home's 15e detailed status page means significant oversieving, and you should therefore be able to raise target density above, say, 110. Try 120 ?

henryzz 2013-04-01 13:33

[QUOTE=debrouxl;335685]That's a 15e task, even if it used 31-bit LPs, so I'm not sure I'm knowledgeable (but this is an occasion to learn)...

I'll venture a guess: for 14e, we usually use 210M-220M raw relations for 31-bit LPs. If that's true for 15e tasks as well, than the ~288M raw relations that I can see in NFS@Home's 15e detailed status page means significant oversieving, and you should therefore be able to raise target density above, say, 110. Try 120 ?[/QUOTE]
Duplication rate decreases a bit when increasing the sieve area. 15e will need slightly less raw relations as a result.

jasonp 2013-04-01 15:41

I'd suggest everyone who is using v1.51 to run 'msieve -h' to see the new set of command line flags available, and possibly read the Readme.nfs file. The latest release really has a lot of user-visible API changes which hopefully eliminate the need to custom-compile the demo binary, for just about everybody.

Right now I think the only reason you would need to roll your own is if you want LARGEBLOCKS=1.

henryzz 2013-04-01 16:29

It isn't possible to change the seconds per coefficient in stage 1 as far as I can tell.

pinhodecarlos 2013-04-01 21:49

[QUOTE=jasonp;335705]

Right now I think the only reason you would need to roll your own is if you want LARGEBLOCKS=1.[/QUOTE]

Yes, I want to set largeblocks to 1. Can it be done via command line (flags) under linux?

RichD 2013-04-01 23:38

L1261
 
Oops! I forgot to state I am running L1261. It is humming along nicely and should be done by the weekend.

jasonp 2013-04-02 01:37

[QUOTE=pinhodecarlos;335740]Yes, I want to set largeblocks to 1. Can it be done via command line (flags) under linux?[/QUOTE]
Unfortunately no; LARGEBLOCKS=1 means a lot of code and several data structures have to change, so it can only be turned on at compile time.

pinhodecarlos 2013-04-02 08:39

[QUOTE=RichD;335753]Oops! I forgot to state I am running L1261. It is humming along nicely and should be done by the weekend.[/QUOTE]

And I am already running L1746 and it should also be done by the weekend. I also pretend to run SF1229 but I don't see our names in reservation. Lionel?

So L1261 reserved by Rich Dickerson.
L1746 reserved by Carlos Pinho.
SF1229 reserved by Carlos Pinho.

Carlos

PS( Thank you jasonp)

debrouxl 2013-04-02 15:05

Reservations for L1261 and L1746 updated, SF1229 is out of my control :smile:


All times are UTC. The time now is 23:02.

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