mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Software

Reply
 
Thread Tools
Old 2011-01-12, 18:21   #1
tichy
 
Nov 2010

7610 Posts
Default Unfixing FFT length for LL test

Hello,

some time ago I started an LL test on an exponent which apparently falls into some boundary region, because at the start of the test mprime (25.11) first checked if the 2560K FFt length is sufficient wrt calculation errors. Later in the worktodo file it appeared with a fixed FFT length attached.

Then I switched to mprime 26.4. I used the same worktodo entry and checkpoint left by the 25.11 version. However, now I've noticed that it takes mprime 26.4 less time per iteration with a slightly larger exponent for which it has selected 2688K FFT length, e.g.:

Code:
[Worker #1 Jan 12 21:08] Resuming primality test of M50493523 using Core2 type-3 FFT length 2688K, Pass1=896, Pass2=3K, 7 threads
[Worker #2 Jan 12 21:08] Resuming primality test of M49044143 using Core2 type-3 FFT length 2560K, Pass1=640, Pass2=4K, 7 threads
[Worker #2 Jan 12 21:13] Iteration: 19380000 / 49044143 [39.51%].  Per iteration time: 0.022 sec.
[Worker #1 Jan 12 21:15] Iteration: 17090000 / 50493523 [33.84%].  Per iteration time: 0.019 sec.
So I'm considering removing the 2560K FFT length fix and verifying both the new selected FFT length and iteration timing for that exponent, however, I'd like to make sure I won't break anything (e.g. internal checkpoint file structure is independent of the FFT length).

Thank you in advance,
tichy is offline   Reply With Quote
Old 2011-01-12, 19:16   #2
Prime95
P90 years forever!
 
Prime95's Avatar
 
Aug 2002
Yeehaw, FL

19·397 Posts
Default

You can safely delete the "FFT2=" string from worktodo.txt. The save files are compatible.
Prime95 is offline   Reply With Quote
Old 2011-01-12, 21:18   #3
tichy
 
Nov 2010

22×19 Posts
Default

Thank you. Mprime again tried 2560K FFT, so this time I enforced 2688K length - and to my surprise it seems to works faster:
Code:
[Worker #1 Jan 13 00:02] Iteration: 19850000 / 49044143 [40.47%].  Per  iteration time: 0.020 sec.
[Worker #2 Jan 13 00:02] Iteration: 17590000 / 50493523 [34.83%].  Per  iteration time: 0.022 sec.
[Worker #1 Jan 13 00:05] Iteration: 19860000 / 49044143 [40.49%].  Per  iteration time: 0.019 sec.
[Worker #2 Jan 13 00:05] Iteration: 17600000 / 50493523 [34.85%].  Per  iteration time: 0.020 sec.
[Worker #1 Jan 13 00:08] Iteration: 19870000 / 49044143 [40.51%].  Per  iteration time: 0.019 sec.
[Worker #2 Jan 13 00:08] Iteration: 17610000 / 50493523 [34.87%].  Per  iteration time: 0.020 sec.
[Worker #1 Jan 13 00:11] Iteration: 19880000 / 49044143 [40.53%].  Per  iteration time: 0.019 sec.
[Worker #2 Jan 13 00:12] Iteration: 17620000 / 50493523 [34.89%].  Per  iteration time: 0.020 sec.
[Worker #1 Jan 13 00:14] Iteration: 19890000 / 49044143 [40.55%].  Per  iteration time: 0.019 sec.
[Worker #2 Jan 13 00:15] Iteration: 17630000 / 50493523 [34.91%].  Per  iteration time: 0.020 sec.
tichy is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Help with FFT length Gradient Software 3 2013-12-16 01:53
Factorization length test? siegert81 Math 17 2011-12-20 15:33
Blend test is failing on FFT length 1120k every time Unregistered Information & Answers 0 2011-12-05 02:07
FFT Length Samoflan Information & Answers 8 2010-02-16 22:05
FFt length mack Information & Answers 1 2009-09-06 03:24

All times are UTC. The time now is 23:24.


Fri Aug 6 23:24:25 UTC 2021 up 14 days, 17:53, 1 user, load averages: 4.38, 4.15, 4.07

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.