mersenneforum.org Force FFT-size to be used
 Register FAQ Search Today's Posts Mark Forums Read

 2017-11-16, 21:39 #1 kruoli     "Oliver" Sep 2017 Porta Westfalica, DE 24478 Posts Force FFT-size to be used Hey, I'm trying to execute non-PrimeNet-work like this (just as example): Code: PRP=N/A,1,2,1000000,51,99,0,3,1 PRP=N/A,1,2,1000000,35181,99,0,3,1 The first line will be executed with 64 K FFT, the second line with 100 K FFT. The bit sizes of the numbers are identical. The result is an extremely small round-off. Am I mistaken when I think that both of them should be able to use the same FFT? If so, how can one enforce a specific FFT size for an assignment or even the whole worktodo-file, if all the numbers have similar size?
 2017-11-16, 23:20 #2 Batalov     "Serge" Mar 2008 Phi(4,2^7658614+1)/2 13·773 Posts You can enforce FFT size. But you shouldn't - in this case. Bitsize is not the only factor for the FFT size choice. The k and/or c value will affect the FFT size too. (In this case c value is large, 35181.)
 2017-11-17, 15:31 #3 kruoli     "Oliver" Sep 2017 Porta Westfalica, DE 101001001112 Posts The round-off is really small. Even then you should not decrease FFT-size? Just out of curiosity: How do you set a specific FFT?
 2017-11-17, 16:31 #4 ATH Einyen     Dec 2003 Denmark 2×17×101 Posts PRP=FFT2=64K,1,2,1000000,51,99,0,3,5 You should use type 5 PRP (the last 5 in the line) unless you are double checking a previous PRP result run with another type. Edit: Nevermind I missed that it was not primenet work. Last fiddled with by ATH on 2017-11-17 at 16:32
2017-11-17, 18:14   #5
kruoli

"Oliver"
Sep 2017
Porta Westfalica, DE

1,319 Posts

Quote:
 Originally Posted by ATH PRP=FFT2=64K,1,2,1000000,51,99,0,3,5
Nice, thanks! :)

 Similar Threads Thread Thread Starter Forum Replies Last Post Uncle Lumpy Software 5 2017-03-11 03:27 davieddy Puzzles 29 2012-08-27 12:26 jasong Math 21 2012-02-13 16:05 joblack PrimeNet 6 2009-04-25 15:16 starrynte PrimeNet 4 2008-10-11 23:04

All times are UTC. The time now is 10:18.

Sat Jan 28 10:18:37 UTC 2023 up 163 days, 7:47, 0 users, load averages: 0.88, 0.89, 1.01