mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Riesel Prime Search

Reply
 
Thread Tools
Old 2007-10-18, 05:44   #1
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

72·11·19 Posts
Default Top-5 prime quadruplet found

Another set of archivable primes for RPS. Found: the 4th highest prime quadruplet of all time! Here it is:

477707955423*2^3802-1, +1, +5, and +7


It is the #1 largest prime quadruplet ever found that contains a Riesel and Proth prime!

(+5 and +7 primes proven with Primo after LLR found them as probable primes.)


Gary
gd_barnes is offline   Reply With Quote
Old 2007-10-18, 07:13   #2
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

27·5·7 Posts
Default

How long does Primo take to prove a number that size prime? What version of Primo do you have?
-Curtis
VBCurtis is offline   Reply With Quote
Old 2007-10-18, 18:09   #3
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

72×11×19 Posts
Default Primo

Quote:
Originally Posted by VBCurtis View Post
How long does Primo take to prove a number that size prime? What version of Primo do you have?
-Curtis
I am using version 3.0.4. If anyone is interested in downloading it, it is at www.ellipsa.net.

It took 1 hour, 3 minutes for the +5 candidate and 1 hour, 21 minutes for the +7 candidate. I'm not sure why the difference. Nothing else was running on my machine either time. It probably has something to do with how well it could factor the number 1 greater and/or 1 less than each number.

Of course LLR found them to be PRP in about 1/20th of a second.

This is a reduced form of 1910831821692*2^3800-1/+1/+5/+7. I had sieved n=3800 for k<=6T including all even candidates. A whole bunch of 'really big' files!

On another note, I got an Email from Prof. Caldwell. I needed to create a new proof code that includes Primo for the +5 and +7 candidates. I've now done that and he is going to 'move them over'. I wanted to credit LLR solely since it was 99.9% of the total prime-finding time but that doesn't work for PRP's at the top-5000 site. I guess since it didn't prove them prime and whatever software proves them prime is what gets credit. So LLR will get credit for -1 and +1 and Primo for +5 and +7.


Gary
gd_barnes is offline   Reply With Quote
Old 2007-10-26, 06:21   #4
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

72×11×19 Posts
Default Another top 5 quad found!

I decided to finish up my sieved range of k=1-6T for n=3800 looking for quads. With barely better than a 1 in 4 chance of finding another with the sieved range left, the prime Gods were on my side and it kicked out another one! Here it is:

2722420456827*2^3800-1, +1, +5, +7


This one is now the 4th largest and the prior one is 5th. Together they are the top 2 quads ever to contain Riesel and Proth primes.

Curtis, for some reason, Primo runs faster on an Athlon than on a core2-duo. This one only took 40 and 44 mins. respectively to prove the +5 and +7 primes. Checking the prior one on the Athlon, it took 42 and 50 mins. respectively vs. over an hour each on the core2-duo. I tested each one at a time on both machines with the other core running something else in both cases. Very strange!


Gary
gd_barnes is offline   Reply With Quote
Old 2007-10-26, 08:45   #5
em99010pepe
 
em99010pepe's Avatar
 
Sep 2004

2×5×283 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
I I tested each one at a time on both machines with the other core running something else in both cases. Very strange!
Gary,

Can you set Primo's affinity instead of letting the OS choose it automatically? Try going to task manager and set core 0 to Primo and core 1 to the other thing you're running, then see what happens....

Carlos

Last fiddled with by em99010pepe on 2007-10-26 at 08:46
em99010pepe is offline   Reply With Quote
Old 2007-10-26, 09:11   #6
Kosmaj
 
Kosmaj's Avatar
 
Nov 2003

70468 Posts
Default

It is well known that Primo is much faster of Athlons and Opterons because it's based on integer computation which is much faster on AMD cpu's. Pentiums are much faster for floating point numbers. I found that my old Athlon-2000, cpu clock 1.66G is just a little bit faster than (2 year old) 3.6G Pentium Xeon.
Kosmaj is offline   Reply With Quote
Old 2007-10-26, 15:31   #7
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

72×11×19 Posts
Default

Quote:
Originally Posted by Kosmaj View Post
It is well known that Primo is much faster of Athlons and Opterons because it's based on integer computation which is much faster on AMD cpu's. Pentiums are much faster for floating point numbers. I found that my old Athlon-2000, cpu clock 1.66G is just a little bit faster than (2 year old) 3.6G Pentium Xeon.
Very interesting. Thanks for the info, Kosmaj. I never would have guessed it with as much faster as P4s are at LLRing. I seem to learn something new several times each week with this effort.

Carlos, I had LLR going on one core (on another effort) and Primo on the other when I tested Primo on both the Core2-duo and dual-core Athlon. Since both are 2-core machines, I'm thinking having a separate affinity would have made little difference. Let me know if you think otherwise.

Hum...I'm not familiar with how to change the core in the task manager. I'll have to look into that. Primo doesn't seem to have a way to change the affinity.



Gary

Last fiddled with by gd_barnes on 2007-10-26 at 15:39
gd_barnes is offline   Reply With Quote
Old 2007-10-26, 16:11   #8
Cruelty
 
Cruelty's Avatar
 
May 2005

2·809 Posts
Default

Quote:
Originally Posted by gd_barnes View Post
Hum...I'm not familiar with how to change the core in the task manager. I'll have to look into that. Primo doesn't seem to have a way to change the affinity.
Hint: right clik on a process name
Cruelty is offline   Reply With Quote
Old 2007-10-26, 16:17   #9
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

72×11×19 Posts
Default

Quote:
Originally Posted by Cruelty View Post
Hint: right clik on a process name
OK, thanks. I see that now. Duh!

Question for anyone here...on a dual-core machine, is there any benefit to having one instance of LLR run on CPU 0 and the other on CPU 1 instead of the default of letting them both run on both CPUs? I always just let both instances default to running on both CPUs.


Thanks,
Gary
gd_barnes is offline   Reply With Quote
Old 2007-10-26, 22:46   #10
amphoria
 
amphoria's Avatar
 
"Dave"
Sep 2005
UK

53268 Posts
Default

Gary,

If you look at Task Manager | Performance you should see that the two instances will fight each other if you do not set the affinity of each. The symptom is that that each instance does not always show 100% CPU usage. The net result is that you lose some processing power for each instance. I always set the affinity either using the program itself or using a program that I wrote that allows you to set the affinity of any given program under Windows. Alternatively you can manually set the affinity as described by Carlos.
amphoria is offline   Reply With Quote
Old 2007-10-27, 01:54   #11
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

72·11·19 Posts
Default

Quote:
Originally Posted by amphoria View Post
Gary,

If you look at Task Manager | Performance you should see that the two instances will fight each other if you do not set the affinity of each. The symptom is that that each instance does not always show 100% CPU usage. The net result is that you lose some processing power for each instance. I always set the affinity either using the program itself or using a program that I wrote that allows you to set the affinity of any given program under Windows. Alternatively you can manually set the affinity as described by Carlos.

Excellent! Thanks a bunch. I'll change all of my dual-core machines now to have separate affinities for each instance of LLR, srsieve, or newpgen.


Gary
gd_barnes is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Prime Quadruplet Emirps Trejack Puzzles 11 2016-04-22 19:27
Base 13 prime found Siemelink Conjectures 'R Us 0 2009-10-05 18:27
Prime found paulunderwood 3*2^n-1 Search 61 2008-04-22 12:19
I think I found a huge prime monst Lounge 16 2008-01-16 21:01
new mersenne prime found unregistered Data 25 2005-01-07 19:26

All times are UTC. The time now is 12:10.

Fri Nov 27 12:10:27 UTC 2020 up 78 days, 9:21, 4 users, load averages: 1.14, 1.25, 1.23

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.