mersenneforum.org CUDALucas (a.k.a. MaclucasFFTW/CUDA 2.3/CUFFTW)
 Register FAQ Search Today's Posts Mark Forums Read

2010-06-14, 13:39   #188
mdettweiler
A Sunny Moo

Aug 2007
USA (GMT-5)

11000011010012 Posts

Quote:
 Originally Posted by ET_ Thank you, so Issue 2 is cleared As for k*2^n+/-1, they are quite the numbers I'm sieving as possible Fermat factors... Luigi
Could you post one or more of the headers from the ABCD file here? (They're the lines that start with "ABCD".) It's possible that FermFact uses a nonstandard format as compared to the ones generally used by srsieve (which, as I understand it, is what ppsieve is designed to read--though I will disclaim that I have never used ppsieve myself).

2010-06-14, 14:19   #189
ET_
Banned

"Luigi"
Aug 2002
Team Italia

2×29×83 Posts

Quote:
 Originally Posted by mdettweiler Could you post one or more of the headers from the ABCD file here? (They're the lines that start with "ABCD".) It's possible that FermFact uses a nonstandard format as compared to the ones generally used by srsieve (which, as I understand it, is what ppsieve is designed to read--though I will disclaim that I have never used ppsieve myself).
Here they are:

ABC file:
Code:
ABC $a*2^$b+1
1201 100016
1201 100076
1201 100148
1201 100168
1201 100268
1201 100360
1201 100388
1201 100416
ABCD file:
Code:
ABCD $a*2^100000+1 [1255] 18 20 52 8 4 120 44 34 18 62 10 30 42 Thanks Luigi 2010-06-14, 16:25 #190 mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts Quote:  Originally Posted by ET_ Here they are: ABC file: Code: ABC$a*2^$b+1 1201 100016 1201 100076 1201 100148 1201 100168 1201 100268 1201 100360 1201 100388 1201 100416 ABCD file: Code: ABCD$a*2^100000+1 [1255] 18 20 52 8 4 120 44 34 18 62 10 30 42 Thanks Luigi
Aha, I think I see what the problem is. Your first ABC file should be fine, but the ABCD file is for a fixed-n format--ppsieve is designed for fixed-k sieves, so it won't understand that kind of notation. If you can convert it to an $a*2^$b+1 or k*2^$b+1 format, it will likely work. 2010-06-14, 18:18 #191 ET_ Banned "Luigi" Aug 2002 Team Italia 2×29×83 Posts Quote:  Originally Posted by mdettweiler Aha, I think I see what the problem is. Your first ABC file should be fine, but the ABCD file is for a fixed-n format--ppsieve is designed for fixed-k sieves, so it won't understand that kind of notation. If you can convert it to an$a*2^$b+1 or k*2^$b+1 format, it will likely work.
I'm afraid not... I tested the ABC file, and it is not correctly parsed either. :-(

Luigi

2010-06-14, 21:18   #192
mdettweiler
A Sunny Moo

Aug 2007
USA (GMT-5)

186916 Posts

Quote:
 Originally Posted by ET_ I'm afraid not... I tested the ABC file, and it is not correctly parsed either. :-( Luigi
Huh, that's strange. Since it's designed for k*2^n+1 numbers, I would think that it would surely work fine with them.

Here, I've got another idea. Try replacing the "ABC $a*2^$b+1" in the first file with "0:P:0:2:257"--essentially changing the file to NewPGen format. Then, run it through the srfile program (included with the previously-linked srsieve) as:
srfile -a filename
where filename is the name of your now-NewPGen format sieve file. srfile should then convert the file to an "ABCD k*2^\$a+1" format file--which ppsieve surely should take.

2010-06-30, 04:01   #193
msft

Jul 2009
Tokyo

2·5·61 Posts

Hi,
Quote:
 M( 70022021 )C, 0x8fc5e5f795149ff8, n = 4194304, MacLucasFFTW v8.1 Ballester

 2010-06-30, 22:19 #194 lycorn     Sep 2002 Oeiras, Portugal 146110 Posts How long did it take?
 2010-07-01, 02:15 #195 wavelet3000   May 2010 7 Posts Nice, there are only 12 exponents above that available for double-checking, the highest of which is in 150M range. What is the upper limit of 8M fft?
2010-07-01, 06:02   #196
msft

Jul 2009
Tokyo

61010 Posts

Hi, lycorn
Quote:
 Originally Posted by lycorn How long did it take?
18 days.

2010-07-01, 06:16   #197
msft

Jul 2009
Tokyo

26216 Posts

Hi, wavelet3000
Quote:
 Originally Posted by wavelet3000 Nice, there are only 12 exponents above that available for double-checking, the highest of which is in 150M range. What is the upper limit of 8M fft?
This program is memory eater.
I think GTX480 can support 16M fft.

2010-07-04, 08:54   #198
frmky

Jul 2003
So Cal

24×131 Posts

Quote:
 Originally Posted by msft I think GTX480 can support 16M fft.
Nope. On the GTX 480, the 8M FFT runs at 18.53 ms/iteration, but the 16M FFT gives cufftSafeCall() CUFFT error in the cufftPlan1d() call.

 Similar Threads Thread Thread Starter Forum Replies Last Post LaurV Data 131 2017-05-02 18:41 Brain GPU Computing 13 2016-02-19 15:53 Karl M Johnson GPU Computing 15 2015-10-13 04:44 fairsky GPU Computing 11 2013-11-03 02:08 Rodrigo GPU Computing 12 2012-03-07 23:20

All times are UTC. The time now is 15:19.

Sun May 16 15:19:41 UTC 2021 up 38 days, 10 hrs, 0 users, load averages: 4.21, 3.93, 3.77