mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Cunningham Tables

Reply
 
Thread Tools
Old 2020-03-20, 13:29   #100
swellman
 
swellman's Avatar
 
Jun 2012

112×23 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
The memory requirement is fixed by the siever size, not by the number of cores. If nobody else joins Sean and I, then we can consider running with A=31 rather than 32, which would reduce memory from ~18GB to ~9GB (same as C207 job from last summer, roughly). I want to wait on that a couple weeks, since these lowest Q values are most valuable to search with the bigger siever area; once we reach Q=10M it seems reasonable to allow folks to use A=31 and less memory. Yield will still be 3.5 or so with the smaller siever.
How close are we to 10M for Q? Is the plan still to go to A=31 for Q > 10M thereby lowering the memory requirements?

Guess it all hinges on Greg’s plans for this composite, i.e. when it starts sieving and Q range.
swellman is online now   Reply With Quote
Old 2020-03-20, 16:32   #101
axn
 
axn's Avatar
 
Jun 2003

4,591 Posts
Default

Currently @ 9.5m. At the current rate, about 2 days to reach 10m, I think.
axn is offline   Reply With Quote
Old 2020-03-20, 17:17   #102
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

105316 Posts
Default

I agree that it depends on Greg's plans. I don't see a compelling reason to change to A=31 at 10M specifically until we see what our max-Q is.
Then again, we've put a month into this and only sieved 5MQ; even if our target was just Q=20M that would take about another month on A=31.
There's only 3 of us working. If seth isn't coming along soon, I suppose we're better off shifting to A=31 and <10GB memory per process sooner rather than later, like 10M rather than 15M.
VBCurtis is offline   Reply With Quote
Old 2020-03-22, 13:43   #103
axn
 
axn's Avatar
 
Jun 2003

107578 Posts
Default

Looks like server is down.
axn is offline   Reply With Quote
Old 2020-03-22, 15:23   #104
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

3×7×199 Posts
Default

I've lost access from home, name no longer resolves. I'll try again later today, but we may be dark a while.
VBCurtis is offline   Reply With Quote
Old 2020-03-22, 19:11   #105
swellman
 
swellman's Avatar
 
Jun 2012

112·23 Posts
Default

My machine lost contact but seems to have since gotten a new WU.
swellman is online now   Reply With Quote
Old 2020-03-23, 00:37   #106
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

3·7·199 Posts
Default

Seems it was a simple internet outage; I didn't touch the server nor CADO, but have since reconnected and found CADO running normally.
VBCurtis is offline   Reply With Quote
Old 2020-03-26, 17:34   #107
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

417910 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
There's only 3 of us working. If seth isn't coming along soon, I suppose we're better off shifting to A=31 and <10GB memory per process sooner rather than later, like 10M rather than 15M.
We will reach Q=11M later today; sometime around that mark I will restart the server with I=16 (that is, A=31 rather than A=32). This will cut memory use in half, from 18GB to 9GB; it will also cut yield by ~30%. yield is currently around 3.5, so a cut to 2.5 is expected.

We have generated ~27M raw relations so far.
VBCurtis is offline   Reply With Quote
Old 2020-03-27, 03:21   #108
axn
 
axn's Avatar
 
Jun 2003

11EF16 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
We will reach Q=11M later today; sometime around that mark I will restart the server with I=16 (that is, A=31 rather than A=32). This will cut memory use in half, from 18GB to 9GB; it will also cut yield by ~30%. yield is currently around 3.5, so a cut to 2.5 is expected.
How does it impact the runtime for the WU?

Side question: Is it worth it, for those who have sufficient RAM, to override the -A parameter back to 32?
axn is offline   Reply With Quote
Old 2020-03-27, 04:10   #109
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

3·7·199 Posts
Default

I don't know about runtime- we will find out tonight!
I'll time a couple of my own workunits, then time the new batch. sec/rel should fall a tiny bit, so I expect WU time to drop by a bit more than 40%.

Let's see what the sec/rel comparison is like before we decide whether to override. We 3 all have sufficient memory, but if A=31 is faster we should try to complete Q to Greg's starting value (which I think will be 40M).

I'll post a timing comparison tomorrow; if A=32 is faster there's no reason to change to 31 at all. Not likely, though.
VBCurtis is offline   Reply With Quote
Old 2020-03-27, 18:45   #110
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

3×7×199 Posts
Default

Using CADO's ETA as metric, we're 65% faster with A=31 than A=32. ETA for 600M relations (CADO default, I didn't set it) is now 12 months rather than 20.

Using the 'top' CPU time versus # of relations found in a workunit, the host machine on A=32 was finding relations at 6.04 thread-sec/rel; it is running 30-threaded on a 20-core machine, so let's call that 4 cpusec/rel.

I'll update this post in a bit with the 'top' metric on A=31.
Update: With A=31, I get yield around 3.2 as opposed to 3.9 on A=32, but 2.83 threadsec/rel rather than 6.0. A=32 was not a good idea for this job! On the bright side, we got lots of relations in a Q-range that Greg doesn't search, so we helped the factorization quite a lot.
Definitely *don't* set override A=32!

Last fiddled with by VBCurtis on 2020-03-27 at 18:55
VBCurtis is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Poly select and planning for 2,2330M VBCurtis Cunningham Tables 69 2020-05-08 06:39
Poly select and planning for 2,2210M swellman Cunningham Tables 51 2020-03-22 22:09
Poly select and test-sieving for RSA232 VBCurtis Operation Kibibit 25 2020-01-07 01:57
YAFU Poly Select Deadline amphoria YAFU 22 2016-09-17 09:47
Starting NFS skipping poly select jux YAFU 5 2016-01-02 01:01

All times are UTC. The time now is 10:57.

Fri May 29 10:57:07 UTC 2020 up 65 days, 8:30, 0 users, load averages: 0.88, 1.18, 1.31

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.