mersenneforum.org  

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

Reply
 
Thread Tools
Old 2016-12-30, 21:34   #1
carpetpool
 
carpetpool's Avatar
 
"Sam"
Nov 2016

23·41 Posts
Post Timing for large candidate

I didn't want to bump the reservation thread and make it off topic, so I am posting the time for the largest candidate reserve here. Norton will delete the LLR executable file for some reason so I can't use it (not the main point here). I tested the largest candidate with PFGW:

C:\Users\Sam\PFGW>pfgw64.exe -q"1826*123^250000+1"
PFGW Version 3.8.3.64BIT.20161203.Win_Dev [GWNUM 28.6]

1826*123^250000+1 is composite: RES64: [836643028A83634B] (7031.6835s+0.0699s)

on Windows i5 core.
carpetpool is offline   Reply With Quote
Old 2016-12-30, 21:48   #2
carpetpool
 
carpetpool's Avatar
 
"Sam"
Nov 2016

23×41 Posts
Post

This is for

http://www.mersenneforum.org/showthr...=15830&page=47
carpetpool is offline   Reply With Quote
Old 2016-12-30, 23:18   #3
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

1,487 Posts
Default

Quote:
Originally Posted by carpetpool View Post
I didn't want to bump the reservation thread and make it off topic, so I am posting the time for the largest candidate reserve here. Norton will delete the LLR executable file for some reason so I can't use it (not the main point here). I tested the largest candidate with PFGW:

C:\Users\Sam\PFGW>pfgw64.exe -q"1826*123^250000+1"
PFGW Version 3.8.3.64BIT.20161203.Win_Dev [GWNUM 28.6]

1826*123^250000+1 is composite: RES64: [836643028A83634B] (7031.6835s+0.0699s)

on Windows i5 core.
First of all if your result here is correct , then you reserved about 1.5 years of work doing 24/7/365 on your I5
Second my I7 done same for only 1594 seconds, so it looks like PFGW is so much slower, or your I5 doesnot have AVX.
Third, Before start do more sieving ( at least to double from now) because it is faster to sieve then to do LLR
Currently I found factors every 164 second, so... it has much sieve to do before even start LLR

Expecting to find factors for about 1191.31 terms in this range.
sr2sieve 1.9.3 started: 100007 <= n <= 400000, 5001618137077 <= p <= 10000000000000
5002744409483 | 122*123^375886+1
5004515732747 | 1826*123^383911+1
p=5006650515749, 11917025 p/sec, 2 factors, 0.1% done, 164 sec/factor, ETA 04 Jan 21:24

So as you can see if you double sieve depth, you will remove about 1190 candidates, runing on one core it will be done in on 4.1 so if you run on four cores it will be done in one day.

Consider sieving before start.

5007891939133 | 1952*123^302585+1 is also removed, so you now have three candidate you dont need to test :)

Last fiddled with by pepi37 on 2016-12-30 at 23:20 Reason: add more info
pepi37 is offline   Reply With Quote
Old 2016-12-30, 23:38   #4
carpetpool
 
carpetpool's Avatar
 
"Sam"
Nov 2016

14816 Posts
Post

I see. I terminated PFGW from testing the sieve file, now the bigger issue is trying to execute sr1sieve (on my computer). (I'll ask for help on some other forum, it's not important here.)

Last fiddled with by carpetpool on 2016-12-30 at 23:39
carpetpool is offline   Reply With Quote
Old 2016-12-30, 23:41   #5
pepi37
 
pepi37's Avatar
 
Dec 2011
After milion nines:)

1,487 Posts
Default

Quote:
Originally Posted by carpetpool View Post
I see. I terminated PFGW from testing the sieve file, now the bigger issue is trying to execute sr1sieve (on my computer). (I'll ask for help on some other forum, it's not important here.)
Not sr1sieve, you must use sr2sieve

first you convert your text sieve file to abcd file using srsieve -a nameoffile
Then you do sr2sieve -i sr_123.abcd -p 5000000000000 -P 10000000000000

Last fiddled with by pepi37 on 2016-12-30 at 23:43
pepi37 is offline   Reply With Quote
Old 2016-12-31, 01:31   #6
VBCurtis
 
VBCurtis's Avatar
 
"Curtis"
Feb 2005
Riverside, CA

11×461 Posts
Default

This thread is as good a spot as any for any assistance you may need with sr1 and sr2sieve. sr1sieve is, like newpgen, for one k-value at a time, while sr2 is for multiple k's on a single base. sr2 scales very well for more than 3 k's; only the case where 2 k's remain is it worth testing two instances of sr1 against one instance of sr2 for speed.

To use sr2 to continue the sieve you have already for S123:

srfile -a {name of current sieve file}

This will convert the ready-for-prp-testing sieve file to sr2 format, with output name something like sr_123.abcd

If your original sieve file has the current sieve depth in the header line, you won't need to tell sr2 where to start; it'll read the header.

sr2sieve -i {name of abcd file} -p {start value for sieving, possibly redundant and unnecessary if header is up to date} -P {end value of the sieve effort, I'd start with double the start value}

This will create a file factors.txt which contains, ahem, the factors. To remove these from the sieve, run: srfile -a -k factors.txt {name of abcd file}

Once you're done sieving, you want a prp file to test: srfile -G {name of abcd file}

Each program can be invoked with "-h" to get a list of flags. This is particularly helpful for srfile, as it can do quite a bit of manipulation.

Good luck!

Last fiddled with by VBCurtis on 2016-12-31 at 01:33 Reason: forgot "2" in srsieve invocation
VBCurtis is offline   Reply With Quote
Old 2016-12-31, 06:02   #7
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

33·17·23 Posts
Default

To determine total test time pick a candidate at ~60% of the n-range as an average test time. So for n=100K-250K, that would be at n=190K. If you are using a candidate at n=250K your estimate will be too long.
gd_barnes is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Timing for different B1 values? CRGreathouse GMP-ECM 8 2018-05-12 05:57
mprime #7 timing summary request aurashift Software 5 2017-11-26 21:28
Strange timing for GMP-ECM 6.2.3 jyb GMP-ECM 5 2010-02-10 14:01
Very strange mem timing behavior, Asus M3A67-EM jwh Information & Answers 1 2009-01-30 18:04
Timing Options Kevin Software 3 2002-09-12 14:03

All times are UTC. The time now is 08:45.


Mon Dec 6 08:45:38 UTC 2021 up 136 days, 3:14, 0 users, load averages: 2.08, 1.74, 1.57

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