mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > Factoring

Reply
 
Thread Tools
Old 2004-10-24, 14:08   #1
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

2×3×7×179 Posts
Default HP49(119)...

Background:

http://mathworld.wolfram.com/HomePrime.html
http://www.worldofnumbers.com/topic1.htm

The attached text file contains the number to be factored... I start it like this:

[pre]cat hp.txt | nohup nice -19 ecm -c 0 -k 10 11e7 >> hp.log &[/pre]

With the default k, it needs 398MB for stage 2... Set k to 10 and it uses 283MB... I think setting k to 20 would make it use 199MB...

With k set to 10, stage 2 is 8% slower... I have 512MB of memory so I am able to run it with the default k, but only if I am not doing anything else... Setting k to 10 seems like a good compromise... Too bad there isn't a way to run the stage 2 part at night when I am asleep...

According to the web page linked above, all the curves below 55 digits have been done... We need 22,000 curves at this depth...

On my 2.4GHz K8, it takes about 53 minutes per curve with k=10...
Attached Files
File Type: txt hp.txt (205 Bytes, 272 views)
Xyzzy is offline   Reply With Quote
Old 2004-10-24, 14:42   #2
akruppa
 
akruppa's Avatar
 
"Nancy"
Aug 2002
Alexandria

2,467 Posts
Default

Quote:
Too bad there isn't a way to run the stage 2 part at night when I am asleep...
There is: run stage 1 during the day by setting B2 to 1 and saving the residues, i.e.

cat hp.txt | nohup nice -19 ecm -c 0 -save hp49.110M.ecm 11e7 1 >> hp.log &

You may want to set the curve number (-c) to a value so that the stage 1 runs finish before you go to bed. You can also kill the process, but the curve currently in progress will be lost.

Then run stage 2 over night:

nohup nice -19 ecm -resume hp49.110M.ecm 11e7 >> hp.log &

Stage 1 uses almost no memory so it shouldn't interfere with your regular use of the computer much. When stage 2 is done on all residues in the hp49.110M.ecm file, you should delete it before doing more stage 1 the next day. Gmp-ecm 5.0 does not overwrite or append to existing files.

Alex
akruppa is offline   Reply With Quote
Old 2004-10-24, 20:02   #3
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

5,717 Posts
Default

I've been using Home Primes (up to 1000) to set the new ECMNet Client/Server. Unforuntately, my server is behind a firewall. If you want to set up the ECMNet Server to manage the curves run I can send you the source for 2.5. You can either use it privately or you can let others know the IP and port to grab work from. BTW, I have run 1 P-1 and 3 P+1 curves at 11e7 and one P-1 at 26e7.
rogue is online now   Reply With Quote
Old 2004-10-25, 04:03   #4
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

2·3·7·179 Posts
Default

25 curves using B1=110000000 & B2=680270182898...
Xyzzy is offline   Reply With Quote
Old 2004-10-25, 05:04   #5
jinydu
 
jinydu's Avatar
 
Dec 2003
Hopefully Near M48

6DE16 Posts
Default

Whoa, looks like Akruppa has already factored several large factors in the HP49 sequence. How long have you been working on step 100? How does that compare with how long it took to factor large composites on previous steps?
jinydu is offline   Reply With Quote
Old 2004-10-25, 05:09   #6
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

11101010111102 Posts
Default

Yeah, I found that page after reading akruppa's list of work he has done... Very impressive stuff!
Xyzzy is offline   Reply With Quote
Old 2004-10-25, 07:39   #7
xilman
Bamboozled!
 
xilman's Avatar
 
May 2003
Down not across

22×41×61 Posts
Default

Quote:
Originally Posted by Xyzzy
Yeah, I found that page after reading akruppa's list of work he has done... Very impressive stuff!
<aol>Me too!</aol>

Alex and I worked on that sequence for several years before giving up for a while.

Paul
xilman is offline   Reply With Quote
Old 2004-10-25, 08:41   #8
akruppa
 
akruppa's Avatar
 
"Nancy"
Aug 2002
Alexandria

2,467 Posts
Default

HP49 is how Paul and I learned the ropes of GNFS. I guess we invested a total of about 5-6 Athlon GHz years in the various factorizations. The most expenstive factorization was HP49(95), a c153 we did with GNFS. That GNFS took about 1.5 GHz years, but we did a lot of ECM on the composite first. For the c204 of HP49(100), we've done the 5500 curves at B1=11M and 9000 curves at B1=44M listed on Patrick's page, which also took over a GHz year.

Alex
akruppa is offline   Reply With Quote
Old 2004-10-25, 14:41   #9
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

2·3·7·179 Posts
Default

Quote:
Originally Posted by xilman
<aol>Me too!</aol>


I saw you there too! I just forgot to mention it... Sorry!
Xyzzy is offline   Reply With Quote
Old 2004-11-30, 02:20   #10
geoff
 
geoff's Avatar
 
Mar 2003
New Zealand

13·89 Posts
Default

On my 2.9GHz P4 one curve takes 103 minutes (3821 sec. stage 1, 2362 sec. stage 2), nearly twice as long as Xyzzy's K8 3400+

I also did stage 1 on a P3 600MHz which took 19484 sec., about the same per clock as the P4. I don't have enough memory in the P3 to do stage 2.

Both times are with gmp-ecm 5.0.3 and libgmp 4.1.4 optimised for each machine.
geoff is offline   Reply With Quote
Old 2005-02-03, 16:30   #11
Xyzzy
 
Xyzzy's Avatar
 
"Mike"
Aug 2002

2×3×7×179 Posts
Default

139 curves using B1=110000000 and B2=680270182898...
Xyzzy is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
P-1 on M1061 and HP49.99 ATH Factoring 21 2009-10-13 13:16

All times are UTC. The time now is 02:27.

Sat Jun 6 02:27:48 UTC 2020 up 73 days, 52 secs, 0 users, load averages: 1.05, 1.22, 1.24

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.