mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > FermatSearch

Reply
 
Thread Tools
Old 2016-12-02, 12:28   #1
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

25×149 Posts
Default Fermat code performances

placeholder
ET_ is online now   Reply With Quote
Old 2016-12-02, 14:24   #2
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

595310 Posts
Default

On the productivity page, you should remove proth.exe. I think it would also look better if you changed the tables so that the n value would represent the row and the program to represent the column. I look forward to pmfs and fermonant being added to the table(s).

I also suggest merging the productivity and download pages. I think it should be clear which OSes each program can run on and whether or not the software is open source. This could be done in another table, for example:

Code:
╔════════════╦═════╦════════╦═════════╦════════╦════════╦════════╗
║  Program   ║ CPU ║  GPU   ║ Windows ║ Linux  ║  Mac   ║ Source ║
╠════════════╬═════╬════════╬═════════╬════════╬════════╬════════╣
║ Fermat     ║ x86 ║        ║ <link>  ║        ║        ║        ║
║ GMP-Fermat ║     ║        ║ <link>  ║ <link> ║ <link> ║ <link> ║
║ FermFact   ║ x86 ║        ║ <link>  ║        ║        ║        ║
║ mmff       ║ x86 ║ NVidia ║ <link>  ║ <link> ║        ║ <link> ║
║ pfgw       ║ x86 ║        ║ <link>  ║ <link> ║ <link> ║ <link> ║
║ pmfs       ║ x86 ║        ║ <link>  ║ <link> ║        ║        ║
╚════════════╩═════╩════════╩═════════╩════════╩════════╩════════╝
Of course you have to add links to the files and fix any mistakes. You could also add the underlying algorithm (as mentioned on the productivity page). This would make it really easy for someone to zoom in on the programs available to them if they participate in the search.
rogue is offline   Reply With Quote
Old 2016-12-02, 14:57   #3
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

25×149 Posts
Default

Quote:
Originally Posted by rogue View Post
On the productivity page, you should remove proth.exe. I think it would also look better if you changed the tables so that the n value would represent the row and the program to represent the column. I look forward to pmfs and fermonant being added to the table(s).

I also suggest merging the productivity and download pages. I think it should be clear which OSes each program can run on and whether or not the software is open source. This could be done in another table, for example:

Code:
╔════════════╦═════╦════════╦═════════╦════════╦════════╦════════╗
║  Program   ║ CPU ║  GPU   ║ Windows ║ Linux  ║  Mac   ║ Source ║
╠════════════╬═════╬════════╬═════════╬════════╬════════╬════════╣
║ Fermat     ║ x86 ║        ║ <link>  ║        ║        ║        ║
║ GMP-Fermat ║     ║        ║ <link>  ║ <link> ║ <link> ║ <link> ║
║ FermFact   ║ x86 ║        ║ <link>  ║        ║        ║        ║
║ mmff       ║ x86 ║ NVidia ║ <link>  ║ <link> ║        ║ <link> ║
║ pfgw       ║ x86 ║        ║ <link>  ║ <link> ║ <link> ║ <link> ║
║ pmfs       ║ x86 ║        ║ <link>  ║ <link> ║        ║        ║
╚════════════╩═════╩════════╩═════════╩════════╩════════╩════════╝
Of course you have to add links to the files and fix any mistakes. You could also add the underlying algorithm (as mentioned on the productivity page). This would make it really easy for someone to zoom in on the programs available to them if they participate in the search.
The productivity page has been updated yesterday, I suggest you to reload the page

Once there was no productivity page, and everything was placed on the download page; then, someone hinted me: "If you keep things separated, both pages will gain in readability...". I will think back about it, to get some understandable content. The problem here is that we have 5 different algorithms, 3 different OSes (plus GPUs and source code), 3 distinct approaches to the problem (ECM, factoring, gfn) and the chance to use (or not) [multiple different] external sievers. Putting everything on a nice table tends to become hard...

While on this, I will also consider to modify the old and half broken testing interface, and enhance the news page to maintain some historical information while still showing only the last 20 elements.

We could add a php page to test small and easy ranges just clicking an online button.

I should finally add the new pages of Wilfrid Keller's gfn search to the links.

So, lots of new frills and whistles are coming... and if you have ideas I will gladly take them into consideration.

Luigi
---

Last fiddled with by ET_ on 2016-12-02 at 14:59
ET_ is online now   Reply With Quote
Old 2016-12-02, 16:35   #4
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

5,953 Posts
Default

Quote:
Originally Posted by ET_ View Post
The productivity page has been updated yesterday, I suggest you to reload the page
I see that now. Very odd that it didn't refresh on its own when I looked at it earlier today. I suspect that the browser did not detect that it had been updated.

Some of the programs (feromant for example) are not on the download page.

I think that there is still value to some of my suggestions. I can mock something up in html this weekend if you are interested.
rogue is offline   Reply With Quote
Old 2016-12-02, 19:45   #5
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

25·149 Posts
Default

Quote:
Originally Posted by rogue View Post
I see that now. Very odd that it didn't refresh on its own when I looked at it earlier today. I suspect that the browser did not detect that it had been updated.

Some of the programs (feromant for example) are not on the download page.

I think that there is still value to some of my suggestions. I can mock something up in html this weekend if you are interested.
Any help is appreciated! Your suggestions are solid gold for me, and a hint to contrast my native lazyness...
ET_ is online now   Reply With Quote
Old 2016-12-02, 20:13   #6
pinhodecarlos
 
pinhodecarlos's Avatar
 
"Carlos Pinho"
Oct 2011
Milton Keynes, UK

27·37 Posts
Default

Quote:
Originally Posted by ET_ View Post
Any help is appreciated! Your suggestions are solid gold for me, and a hint to contrast my native lazyness...
Latins are lazy by nature that's why Portugal, Spain, Italy are in crisis like Greece.
pinhodecarlos is online now   Reply With Quote
Old 2016-12-02, 22:00   #7
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

5,953 Posts
Default

Try the html in the attached
Attached Files
File Type: txt Programs.txt (8.1 KB, 177 views)
rogue is offline   Reply With Quote
Old 2016-12-03, 07:05   #8
Gary
 
Gary's Avatar
 
"Gary"
Aug 2015
Texas

22·13 Posts
Default

Since you asked for ideas, here are a few more

On the downloads page, move the program descriptions out of the "Small and average Fermat numbers F24 ~ F2000" section, since that implies those programs are only useful in that range. You could then move the performance discussion for the three ranges (very small, small and large) to the productivity page, where readers could more easily compare the performance discussion to the tables.

I like rogue's idea of getting all the download links into a single table. Another column for version would also be good, and programs like FermFact that have two downloadable versions could have two rows in the table.

On the productivity page, coloring pmfs as orange (Montgomery multiplication) is not entirely accurate, since it calls GMP mpz_mul which dynamically selects between several multiplication algorithms, including FFT. Maybe add a new color for GMP/dynamic, or remove the colors entirely and put the algorithm descriptions in the program descriptions on the downloads page (Maybe each author could provide a description of their program's algorithms. I would be happy to provide one for pmfs).

Gary is offline   Reply With Quote
Old 2016-12-03, 14:33   #9
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

10010101000002 Posts
Default

Quote:
Originally Posted by rogue View Post
Try the html in the attached
Nice one, Mark, thanks
I will update the productivity page with your table, and add the comments Gary offered.

Here it is. Comments are appreciated.
Now starting working on the download page.

luigi

Last fiddled with by ET_ on 2016-12-03 at 15:25
ET_ is online now   Reply With Quote
Old 2016-12-03, 15:36   #10
rogue
 
rogue's Avatar
 
"Mark"
Apr 2003
Between here and the

5,953 Posts
Default

Quote:
Originally Posted by Gary View Post
remove the colors entirely and put the algorithm descriptions in the program descriptions on the downloads page (Maybe each author could provide a description of their program's algorithms. I would be happy to provide one for pmfs).
I agree with this statement. I don't think that the algorithm used is that important.

You mentioned that there are two builds for FermFact. I was not aware of that. What is the difference between them?

It should be worth noting that some of these programs will only find Fermat factors and that others will find GFN factors. Also, some of these programs rely solely on trial division or sieving to eliminate candidates and others do a primality test in addition to trial division/sieving. I think that pfgw is the only one that does a primality test, but that means that some of the other programs are unable to test for GFN factors.

Last fiddled with by rogue on 2016-12-03 at 15:39
rogue is offline   Reply With Quote
Old 2016-12-03, 17:10   #11
Gary
 
Gary's Avatar
 
"Gary"
Aug 2015
Texas

22×13 Posts
Default

I like the new productivity table very much! Would it be possible to show the key program parameters used for these benchmark runs? A user would need to know these parameters to achieve these same results. For programs like GMP-Fermat and pmfs that have a sieve, the number of primes used in the sieve would be needed. For pmfs the number of threads is key also. This could be done with a few additional header rows at the top of the table, or footnotes below the bottom of the table.

Quote:
Originally Posted by rogue View Post
It should be worth noting that some of these programs will only find Fermat factors and that others will find GFN factors. Also, some of these programs rely solely on trial division or sieving to eliminate candidates and others do a primality test in addition to trial division/sieving. I think that pfgw is the only one that does a primality test, but that means that some of the other programs are unable to test for GFN factors.
This is exactly the type of information I was thinking could be in each program description on the downloads page. To make it easy to compare program features, the downloads page could also summarize this information in a table similar to the productivity table, with a column for each program and a row for each capability / algorithm. Some rows (with column values shown in ()) might might be: Factors found (Fermat, GFN), Primary algorithm (Trial division, Primality test), Multiplication (GMP powm, GMP mul, Montgomery, etc), Division (GMP powm, GMP div by factor, GMP div by K, etc), Seive (Yes, Yes in FermFact, etc).

The two version of FermFact (0.9 and 2.0) are shown at the bottom of the downloads page. I don't know the differences, but if 2.0 is a superset of the abilities of 0.9 then maybe we could just remove 0.9.

Last fiddled with by Gary on 2016-12-03 at 17:13
Gary is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Fermat code ET_ FermatSearch 44 2019-04-28 16:50
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 15:58.

Wed Oct 28 15:58:11 UTC 2020 up 48 days, 13:09, 3 users, load averages: 1.73, 1.80, 1.76

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.