mersenneforum.org  

Go Back   mersenneforum.org > Prime Search Projects > Conjectures 'R Us

Reply
 
Thread Tools
Old 2008-12-30, 15:20   #188
Flatlander
I quite division it
 
Flatlander's Avatar
 
"Chris"
Feb 2005
England

31×67 Posts
Default

I have finished 660-700M. (I got restless.)

I ran it on two cores and got a surprising result.
660-680 resulted in 87 primes remaining.
680-700 resulted in 124 primes remaining.
(Redundant ks were not tested; as per your new testing idea.)

I have manually checked my script (k from 1 to 200) to make sure that everything is being tested as it should. Ks were tested/not tested exactly as expected.

"Primes are very strange" but perhaps you could check the distribution of the remaining ks to see if I need to double-check?
Attached Files
File Type: zip R-base3-660-700Mremaining.zip (1.2 KB, 51 views)
Flatlander is offline   Reply With Quote
Old 2008-12-30, 17:09   #189
michaf
 
michaf's Avatar
 
Jan 2005

7378 Posts
Default

Sent two minutes ago.

Quote:
Originally Posted by gd_barnes View Post
Micha,

Thanks for the update.

Can you send me your primes and k's remaining up to k=550M right now? Like I was saying to Kenneth in the last post, it helps me keep from getting too far behind on verification and udpating of the pages for efforts with a lot of primes and k's.


Thanks,
Gary
michaf is offline   Reply With Quote
Old 2008-12-31, 04:27   #190
jasong
 
jasong's Avatar
 
"Jason Goatcher"
Mar 2005

DB116 Posts
Default

(Stuff that was important when it was written has been deleted because it doesn't apply to my reply :) )
Quote:
Originally Posted by gd_barnes View Post
Unfortunately there isn't really much that you wrote here that convinces me that you can give an accurate accounting of all of the k's in 30 days.

Are you really going to use PFGW to test every k to n=25000? If so, you're in trouble. The effort will take you months! You have to sieve and then run LLR or Phrot for the higher n-ranges. Trial factoring with the -f switch in PFGW will be far too slow for such a large k-range. I admit that I did just that for a k=10M range because I wanted to start it and forget it but I wouldn't dream of doing it for a k=100M range.
I'm not really sure whether that's true or not, I've figured the total amount of data that will result by noticing that all the ks that don't yield primes are are comprised of a line 15 bytes long, like so...

7xxxxxxxx*3^n-1

So, that means that the minimum size of all the files at the end will be 15 bytes times 50,000,000 individual ks, which amounts to 750,000,000, which, while a definite minimum, is probably only about 10% less(if that) than what will actually result. The last I looked, I have two files so far, one is about 82,000kB and the other is about 89,000kB(this is how Windows displays things, so I'm going to keep things simple by going by that)

(random edit: I have recently eaten a huge meal and my sleeping patterns have been bad in the past few days, so I'm feeling decidely unsteady. I hope that doesn't show in my writing)

The scripts I'm using, which weren't written by me because I found them on the CRU forums, stop processing a k when a prime is found, and...While I don't think I'm capable of writing a program on my own in any decent amount of time, I am capable of understanding the very well written script enough that I can use a combination of my Linux laptop's tail program and a small change to the script to continue from where I left off. Unfortunately, my Linux laptop kicked the bucket a couple days ago, and my friend who was going to help me fix it(no offense meant if he's reading this) has gone temporarily AWOL.

Quote:
One thing I should warn you about PFGW: If you are telling it to stop processing a k when it finds a prime for the k (which you should), if you stop PFGW in the middle, it will NOT remember the k's that it has found primes for when you restart it. This can mess you up big! For that reason, I recommend doing one k to n=25K, the next k to n=25K, and so on instead of progressing upwards by n-value on all k's at once. If you have to restart, it's a quick change to the PFGW script to make it start at the k where it left off.
I'm fading fast, so I'll go ahead and leave this out of order. Sorry.

Quote:
I need one more thing from you:

What is your current status? Let me give more detail:

I think you said you were running 2 cores so...
Please look in your 2 PFGW GUI windows and tell me what they say. By that I mean:
1. What k is it processing and what n-value is it at?
-or-
2. If it is scrolling thru the ranges too fast to see, please tell me what the last 3-PRP that it found is.

One of those two things will tell me just about all that I need to know. With that info., I can likely calculate a nearly exact estimated completion date for you.
I'm not sure how long it will take for one of these files to open, so I'll click on both and hope they're open by the time I'm done with my bedtime routine.

Here we go:

714,193,524*3^3-1 and 740,618,462*3^2-1 (commas added by me) are the last primes found in the file. I'm doing them in ranges of 25 million, so both ranges have about 5 million numbers to go. Then, hopefully with a better strategy, I'll crunch 750 million to 800 million.

If I may make a request: NewPGen works command-line, very much like it's Linux counterpart. The file of ks that haven't yielded a prime is basically lines of the form 7xxxxxxxx*3^n-1. If someone could make a script that could take these k-values, whatever they are, listed this way, plus a range of n, and tell it to sieve that range to 1G, then someone could simply use srfile to patch them all together and sieve them higher. Not sure how difficult that would be, and I'm by no means demanding the script.
Quote:
...Also, I may ask for that last bit of info. 1-2 times a week to make sure that you are progressing at a rate that would allow completion in the amount of time that you think it should.
I'll watch it as well. I've got more free time than I deserve, so after my laptop gets fixed(hopefully) I intend to dig around on the Internet to try to make that script I mentioned.


Quote:
Gary
Remember the Bon Jovi song,"Sleep When I'm Dead?" Well, if there's an earthquake or any other horrible emergency in the next 5 hours or so, I'm a goner. Good night.
jasong is offline   Reply With Quote
Old 2008-12-31, 04:38   #191
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

22·13·197 Posts
Default

Thanks to all for sending the primes and k's remaining as needed. That will help a lot. I'll be on a business trip from Jan. 1st to 7th. Being able to partially update primes and k's remaining for your ranges will save me a lot of time when I get back.


Gary
gd_barnes is offline   Reply With Quote
Old 2008-12-31, 05:04   #192
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

22×13×197 Posts
Default

Jason,

I can tell quite a bit by the latest primes that you posted:

1. You are searching by k-value. (That's good!)

2. You have progressed far more quickly than I expected. (That's even better!)

3. I was correct in saying you could not finish in 30 days by only using PFGW but wrong in saying you could not finish by Jan. 31st. (That's good too! lol)

So all good news.

Here's an estimate for you:
Your starting date/time: Dec. 18th @ 11:40 PM GMT
(as per a post by you in this thread)

Current date/time of your recent status post: Dec. 31st @ 4:25 AM GMT

Amount of time elapsed so far: 12 days, 4 hour, 45 mins or 12.2 days. <-----

Ranges completed so far:
k=700M-714.19M or 14.19M
-and-
k=725M-740.62M or 15.62M

Total completed: 14.19M+15.62M = 29.81M

Total range to test k=700M-800M or 100M.

Percentage complete: 29.81M / 100M = 29.81% <-----

Total estimated time to complete entire range:

12.2 days / 29.81% = ~41 days

Starting date: Dec. 18th

Estimated completion date: Dec. 18th + 41 days = Jan. 28th <-----

Note: This assumes that you have set PFGW to search all k's up to n=25K. If so, you're in good shape. If not, you're in very bad shape.

Please use this as a guide for estimating completion dates in the future. It's quite simple once you've done it a couple of times. If you have any questions, just ask. Note: This only applies when you are searching by k-value. If searching by n-value, it is much trickier.

How about that? The estimate comes right in near the end of Jan! OK, you have your chance. Don't mess it up! lol I'll officially reserve the range for you. In your situation, using PFGW for the entire thing is probably best even if it's not very efficient. It's simple and you don't have to do much other than making sure that it is continuing to run. Please make sure that all k's are accounted for when you are done.

One more request: When your k=700M-725M and 725M-750M ranges are done, please sort the primes by n-value and send me all primes for n>500. That will save a lot of space going back and forth. It is no problem for me to rerun all of the primes for n<=500 in < 1 day for verification purposes. If you can't find a way to sort the file by n-value, then you should probably send me all of the primes. If you need to send 10 Emails to do so, that is fine.

One more instruction: Be sure and check for primes in both the prime-pfgw.log and pfgw.log files. You don't want to miss a bunch of them.

One final thing: Please make sure your computer is on at all times and crunching on 2 cores. Please think ahead here. Several problems that can affect continuous 24x7 efforts:
1. Make sure that any automatic Windows updates that restart your computer are turned off.
2. If it is an always connected machine, make sure that your virus protection is up to date.
3. If online gaming on the machine, make sure that your memory is sufficient so that the machine does not lock up while playing and running 2 cores on PFGW.
4. If others will use the machine, make sure that they know the machine is always turned on and not to mess with anything at the bottom of the screen. I've had my kids shut my machine off accidently when they were trying to turn it on because the monitor was turned off. They didn't know the machine was already on.


Gary

Last fiddled with by gd_barnes on 2008-12-31 at 05:20
gd_barnes is offline   Reply With Quote
Old 2008-12-31, 13:36   #193
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

240048 Posts
Default

Quote:
Originally Posted by Flatlander View Post
I have finished 660-700M. (I got restless.)

I ran it on two cores and got a surprising result.
660-680 resulted in 87 primes remaining.
680-700 resulted in 124 primes remaining.
(Redundant ks were not tested; as per your new testing idea.)

I have manually checked my script (k from 1 to 200) to make sure that everything is being tested as it should. Ks were tested/not tested exactly as expected.

"Primes are very strange" but perhaps you could check the distribution of the remaining ks to see if I need to double-check?

Chris,

You'll need to send me the primes first for a reasonable verification.

I can only say two things at this point:
1. It appears that you have tested appropriate MOB. All 9 k's remaining that are divisible by 3 are prime for k-1.
2. Yes, it is very possible to have 87 k's remaining in one range and 124 k's remaining in the next range of the same size.

One final thing: The total of 211 k's remaining for a k=40M range compares very reasonably to the 144 k's remaining for a k=30M range (after removing appropriate MOB) that Micha just sent me.

After getting the primes for n>1000 or something close to that, I can do final verification and list the range as complete and show the k's remaining on the pages.


Thanks,
Gary
gd_barnes is offline   Reply With Quote
Old 2008-12-31, 14:09   #194
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

22·13·197 Posts
Default

Quote:
Originally Posted by KEP View Post
Status for Riesel base 3 k>100M to k<200M:

Core 1: n=61078 (67 PRPs found)
Core 2: n=68871 (79 PRPs found)
Core 3: n=54007 (51 PRPs found)
Core 4: n=57834 (54 PRPs found)

+ 16 additional PRPs for the n=25001 to n=26000 range.

A total of PRPs out of 490 possible is found. Less than 250000 tests remain. ETA is 4-6 weeks, maybe less since now a total of 4 commited cores on the Quad has been reinstated for this particular challenge.

Regards

KEP

EDIT: k=3677878 is still at n=298000 and testings is not scheduled to begin for earliest 1 week from now! (~13550 remain)

Very good. Your total in this post is 267 primes but you sent me 269 primes. I assume that you found 2 more primes between the time you posted this and when you sent me the primes.

On the web pages, I will only eliminate k's and show primes for n<=54K since that is the highest range you have fully completed. If I knew what cores were on what k-range, it would still be too messy and time-consuming to divide up your range on the pages. The n-ranges and primes found have to be in sync or we can lose track of where we are at.

I won't lose the primes for n=54K-68K. They'll just be waiting to post later on when all of your k-ranges are at n>=68K.


Gary
gd_barnes is offline   Reply With Quote
Old 2008-12-31, 14:27   #195
Flatlander
I quite division it
 
Flatlander's Avatar
 
"Chris"
Feb 2005
England

31·67 Posts
Default

Primes for 660-700M, n from 1k-25k.
Attached Files
File Type: zip R-base3-660-700Mprimes.zip (34.6 KB, 50 views)
Flatlander is offline   Reply With Quote
Old 2009-01-02, 04:16   #196
jasong
 
jasong's Avatar
 
"Jason Goatcher"
Mar 2005

5·701 Posts
Default

I'm rather unhappy with how long things are going to take for me. I was wondering if anyone was using scripts that wouldn't be too hard for me to adapt to my needs.

Basically, I'd like to stop things, note where they're at, then do the rest of my range only up to n=2000 in pfgw. I know for a fact that sieving will speed up the process tremendously, as far as continuing to n=25,000 is concerned. The real problem is dealing with found primes, as well as removing them from the sieve files.

Any help would be greatly appreciated. :)

Edit: Oops, forgot to mention that I'm in Windows. I have a Linux laptop that could probably do what I want, but it's busted, and my online "tech guy" has been nowhere to be found for the last few days. If he doesn't come back online by the 10th, I'm going to hand it over to my dad so that he can confirm to my mother that he can't fix it. Then I'll bust out the cash and find a repairman. But that's assuming the 10th shows up before my friend.

(I already know he can't fix it, but I need to convince him so that HE can convince my MOTHER, so that she doesn't give me THE LOOK when I spend the money on a repairman. We need to hope that the US military never figures out how to generate THE LOOK on command, because that would truly make us invincible.)

Last fiddled with by jasong on 2009-01-02 at 04:24
jasong is offline   Reply With Quote
Old 2009-01-04, 15:36   #197
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

22·13·197 Posts
Default

Quote:
Originally Posted by jasong View Post
I'm rather unhappy with how long things are going to take for me. I was wondering if anyone was using scripts that wouldn't be too hard for me to adapt to my needs.

Basically, I'd like to stop things, note where they're at, then do the rest of my range only up to n=2000 in pfgw. I know for a fact that sieving will speed up the process tremendously, as far as continuing to n=25,000 is concerned. The real problem is dealing with found primes, as well as removing them from the sieve files.

Any help would be greatly appreciated. :)

Edit: Oops, forgot to mention that I'm in Windows. I have a Linux laptop that could probably do what I want, but it's busted, and my online "tech guy" has been nowhere to be found for the last few days. If he doesn't come back online by the 10th, I'm going to hand it over to my dad so that he can confirm to my mother that he can't fix it. Then I'll bust out the cash and find a repairman. But that's assuming the 10th shows up before my friend.

(I already know he can't fix it, but I need to convince him so that HE can convince my MOTHER, so that she doesn't give me THE LOOK when I spend the money on a repairman. We need to hope that the US military never figures out how to generate THE LOOK on command, because that would truly make us invincible.)

Jason,

How about finally finishing something that you reserve on one of the projects that I run completely? Come on...I know you can do it. Just let it run. Otherwise you're batting .000...that is 0 for 3 on completing reserved efforts at NPLB and CRUS.

In the future, please listen to me when I say...reserve something smaller. It never hurts. What is the point in reserving such large efforts? Reserving something smaller does not mean that you cannot do a larger effort. It just shows you're learning how to do it in the first place.

Regardless, if you can't do it, we need specifics:
1. Are you going to finish k=700M-750M to n=25K? Please don't stop in the middle of that.
2. Is it your intent to run k=750M-800M to only n=2000?

If you stop in the middle of #1, I will most likely throw your work out with one exception. You would need to answer:
1. What k's are left remaining?
2. What n-range have you searched to?
3. Have all k's been searched to the same n-range?

If we can't get clear answers on those, then I don't want to deal with trying to figure what is left in the middle of a range like I did with your sieve file at NPLB. It's much easier for me to just let others start over with smaller ranges.

Please answer these questions so that we know what to unreserve for you.


Gary

Last fiddled with by gd_barnes on 2009-01-04 at 15:38
gd_barnes is offline   Reply With Quote
Old 2009-01-06, 13:26   #198
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

22×13×197 Posts
Default

I have recieved an unsolicited offer to do the k=700M-800M range.

Jason, don't worry about sending what you have done. We will get it taken care of.

Consider something smaller and less complex in the future.


Thanks,
Gary
gd_barnes is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Bases 251-500 reservations/statuses/primes gd_barnes Conjectures 'R Us 2206 2020-11-25 20:28
Bases 101-250 reservations/statuses/primes gd_barnes Conjectures 'R Us 862 2020-11-25 17:01
Bases 4-32 reservations/statuses/primes gd_barnes Conjectures 'R Us 1418 2020-11-24 19:45
Bases 33-100 reservations/statuses/primes Siemelink Conjectures 'R Us 1673 2020-11-18 12:14
Sierp base 3 reservations/statuses/primes gd_barnes Conjectures 'R Us 388 2020-10-21 19:42

All times are UTC. The time now is 03:14.

Tue Dec 1 03:14:52 UTC 2020 up 82 days, 25 mins, 1 user, load averages: 1.75, 1.99, 1.87

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.