mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > FermatSearch

Reply
 
Thread Tools
Old 2016-12-02, 20:12   #23
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22·1,481 Posts
Default

Quote:
Originally Posted by ET_ View Post
The number of Ns is close to the number of ks N=80000 to 90000, k=80000 to 90000 is a centered example.
Unfortunately that doesn't help me understand the algorithm being used.
rogue is offline   Reply With Quote
Old 2017-07-13, 09:59   #24
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

129E16 Posts
Default

July 13th, 2017
Roman Maznichenko released a new version (v2.017) of his Feromant_CUDA program, now ready to work with Pascal generation GPUs.
He also announced that a Linux version of Feromant is under testing.
You will find it in the download page.
Thank you very much Roman!

Luigi
ET_ is offline   Reply With Quote
Old 2017-09-04, 16:10   #25
feromant
 
"Roman"
Dec 2016
Everywhere

2×13 Posts
Default Released Linux build Feromant_CUDA

Pleased to announce the release Linux build my program Feromant_CUDA (v2.017).
The software has been tested on the GPU GTX580 (Fermi) and GTX1050
(Pascal) and is working correctly. Performance linux build slightly
inferior to the Windows program. Version of Nvidia's driver is 375.66, CUDA Toolkit 8.0.61.
You can try it on the download page.


Roman
feromant is offline   Reply With Quote
Old 2017-09-04, 16:21   #26
feromant
 
"Roman"
Dec 2016
Everywhere

2·13 Posts
Default Released Linux build Feromant_CUDA

Pleased to announce the release Linux build my program Feromant_CUDA (v2.017).
The software has been tested on the GPU GTX580 (Fermi) and GTX1050
(Pascal) and is working correctly. Performance linux build slightly
inferior to the Windows program. Version of Nvidia's driver is 375.66, CUDA Toolkit 8.0.61.
You can try it on the download page.


Roman
feromant is offline   Reply With Quote
Old 2018-02-25, 23:41   #27
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22×1,481 Posts
Default

Does Feromant test GFN divisibility only? If so, can it be modified to test for xGFN divisibility as well? The same questions apply to pmfs and mmff. I know that GMP-Fermat does not.
rogue is offline   Reply With Quote
Old 2018-02-26, 15:28   #28
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

2×2,383 Posts
Default

Quote:
Originally Posted by rogue View Post
Does Feromant test GFN divisibility only? If so, can it be modified to test for xGFN divisibility as well? The same questions apply to pmfs and mmff. I know that GMP-Fermat does not.
I am fairly sure pmfs does not actually test GFN divisibility, but it is on the to-do list of Gary.
Serge developed a version of mmff with gfn capabilities.
I will ask Roman when he contacts me for a new range.
ET_ is offline   Reply With Quote
Old 2018-02-27, 16:11   #29
feromant
 
"Roman"
Dec 2016
Everywhere

2·13 Posts
Default

Quote:
Originally Posted by rogue View Post
Does Feromant test GFN divisibility only? If so, can it be modified to test for xGFN divisibility as well? The same questions apply to pmfs and mmff. I know that GMP-Fermat does not.
Feromant is applicable for the Fermat number divisibility test, for GFN and xGFN there is no. Above support for divisibility tests for GFN and xGFN can be thought of.
feromant is offline   Reply With Quote
Old 2018-02-28, 03:54   #30
Gary
 
Gary's Avatar
 
"Gary"
Aug 2015
Texas

72 Posts
Default

Luigi is correct: pmfs does not currently support searching for factors of GFNs.

As for adding the ability to pmfs, let me ask a few "rookie" GFN questions:
  1. I have seen GFN defined as both a^2^n + 1 and as a^2^n + b^2^n. See http://mathworld.wolfram.com/General...matNumber.html. Which is the currently accepted definition of a GFN?
  2. What is a xGFN? I assume "x" means Extended, but what is the equation?
  3. Are all factors of a GFN or xGFN of the form k*2^m + 1 with k odd and m>n?
Thanks!
Gary is offline   Reply With Quote
Old 2018-12-17, 19:48   #31
feromant
 
"Roman"
Dec 2016
Everywhere

1A16 Posts
Default New versions of program feromant and feromant_CUDA (v2.018) for Windows is released!

I am pleased to announce new versions of feromant and feromant_CUDA programs. Feromant_CUDA for Linux will be released soon. You can try it on the download page on Fermatsearch.org project's website.


New in version 2.018

Additions and changes

1) Rewrote code for the siever.
2) Extended output to the file results.txt.
This is done thanks to the request of Peter Strasser,
an active participant in the project Fermatsearch.org.
3) Additional information is now saved to the checkpoint file feromant.cpt.
This is to ensure that data is written to the results file for the entire range
calculation period, not just for the last session.
(IMPORTANT!!! Because of the file format change, complete the current range
before you upgrade to a new version of the program).
4) Added two kernels for processing candidates up to 640 bits
(previously the program was limited to 608 bits).
5) Introduced GPU performance classes (low, medium and high). For each of them
selected profiles containing the corresponding source data (PrimeCount and SieveRange).
This led to a significant increase in the performance of the program for
low-end and middle-end GPU.
6) Changed the behavior of the program when reading an wrong line from the file worktodo.txt.
If earlier the program terminated with the corresponding error code,
now it deletes the wrong line and proceeds to the next one.
Also on request Peter Strasser.
7) Introduced support for GPU generation Volta and Turing.
8) And some minor changes.

Correction of errors and bug fixes

1) Rewrote code for the parser for the file worktodo.txt.
Now the parser correctly splits one line of the task into several lines,
each of which will be executed on the corresponding kernel.
2) Fixed a bug that caused the program to crash when the calculation speed
is approximately 10-15 million candidates per second. It is fixed by
resizing the grid (Threads Per Grid parameter) for a low-end and
middle-end GPU.
3) Fixed the value of tested candidates written to the file results.txt.
Previously, it was a multiple of the value of Threads per Grid max (1048576),
now the exact value is recorded.
4) And some minor errors.
feromant is offline   Reply With Quote
Old 2018-12-17, 21:08   #32
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

22·1,481 Posts
Default

Has this table been updated? Should the table show relative speed based upon the GPU? I assume that some GPUs have much higher throughput than others.
rogue is offline   Reply With Quote
Old 2018-12-18, 12:26   #33
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

2×2,383 Posts
Default

Quote:
Originally Posted by rogue View Post
Has this table been updated? Should the table show relative speed based upon the GPU? I assume that some GPUs have much higher throughput than others.
The table has been updated. The Feromant_CUDA values are referred to a somewhat faster than average computer.
The CPU/GPU values are reported to give an idea of the order of performance speedup using a fast (but not state-of-the-art) GPU.

k per second vary with CPU and GPU type, with the algorithm, with the use of GPU and/or OpenCL functions, with the number of running threads on both CPU and GPU, with the deep of the siever used (if any) and its own threads... it is not easy to mesh all these numbers into a single table, but in my opinion having moretables would only confuse the standard user.

Still, I'm here to consider new ideas.
ET_ is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Fermat code performances ET_ FermatSearch 29 2018-03-15 19:06
Rho code Happy5214 YAFU 3 2015-11-01 21:54
Please help me with my code daxmick Programming 15 2014-02-14 11:57
Code Help Andrew Programming 12 2013-02-16 20:53
New Code JohnFullspeed Programming 20 2011-09-04 04:28

All times are UTC. The time now is 01:44.

Thu Oct 1 01:44:11 UTC 2020 up 20 days, 22:55, 1 user, load averages: 1.66, 1.45, 1.53

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.