<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Recent posts to Discussion</title><link>https://sourceforge.net/p/isprime64/discussion/</link><description>Recent posts to Discussion</description><atom:link href="https://sourceforge.net/p/isprime64/discussion/feed.rss" rel="self"/><language>en</language><lastBuildDate>Thu, 28 Jan 2021 21:09:45 -0000</lastBuildDate><atom:link href="https://sourceforge.net/p/isprime64/discussion/feed.rss" rel="self" type="application/rss+xml"/><item><title>Files for pseudo primes up to 2^32</title><link>https://sourceforge.net/p/isprime64/discussion/general/thread/e63bb4a518/?limit=100#e25d</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;I could now download 32.dat, and this is what I searched (sadly division filtered up to 7, but usable)&lt;br/&gt;
I got a "download forbidden" message first.&lt;br/&gt;
thanks&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Tester7</dc:creator><pubDate>Thu, 28 Jan 2021 21:09:45 -0000</pubDate><guid>https://sourceforge.net57001a977bc4518979273d9fa60489b54b927e0e</guid></item><item><title>Welcome!</title><link>https://sourceforge.net/p/isprime64/discussion/general/thread/d2bb3829/?limit=25#473d</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;This project has some code to check if 64-bit integers are primes. It uses One, two or three rounds of Miller Rabin with the last one picking a base via a hashing scheme so that the resulys should be fully reliable. I have emphasised SMALL hash tables here, and that means I do not get all the way to 64-bits using Miller Rabin: beyond that I used BPSW (ie one round of Miller Rabin combined with a Lucas test). &lt;/p&gt;
&lt;p&gt;The code is intended to behave properly on a range of platforms, and specifically this should include both 32 and&lt;br/&gt;
64-bit systems. The difference that is that on (many) 64-bit systems if you use g++ as your compiler you can&lt;br/&gt;
make use of a non-standard 128-bit integer type, while elsewhere it is neceesarty to code up the operation&lt;br/&gt;
(a*b)%c on 64-bit inputs using just 64-bit internal working.&lt;/p&gt;
&lt;p&gt;I have used generously licensed code from the Hacker's Delight and for a Mersenne Twister pseudo-random generator here: thanks are due to the authors of those ocmponents and I hope that my incorporating their code as part of my project gives them extra recognition. &lt;/p&gt;
&lt;p&gt;I use the "isprime" function from here in the CSL system that supports the Reduce algebra system. Before I put this material together I had looked for other freely licensed and tidy prime checkers and I hope that what is here will be of interest to those who do similar searches in the future. I believe that what I have here differs from most of the other code bodies that I found in the following ways:&lt;/p&gt;
&lt;p&gt;(1) Respectably portable C++ code that includes support for the case where you need to do (a*b)%c without an extra precision type to use for the intermediate result.&lt;br/&gt;
(2) A deterministic hashing scheme for Miller-Rabin style testing that uses rather compact tables.&lt;br/&gt;
(3) An implementation of BPSW which incoludes an option (which only makes sense for small inputs!) to give a trace that shows its internal workings. The main C++ version here is only usable as is up to 64-bits, but the Reduce-coded version (using the Reduce big-number system) will work for much larger cases, and given any particular bignum package this version would provide a straightforward starting point for adaptation.&lt;br/&gt;
(4) Code and tables involved with the search for and creation of the hash tables.&lt;br/&gt;
(5) A CUDA variant of code that lists false witnesses for 32-bit composites. The output from this was not in fact used, because by the time I wrote it I had run a much more laborious ordinary multi-threaded search - but it may be of interest as a small sample of CUDAode to people very early in their exposure to that.&lt;/p&gt;
&lt;p&gt;So good luck and enjoy - and I would appreciate any comments.&lt;/p&gt;
&lt;div class="codehilite"&gt;&lt;pre&gt;        Arthur Norman
&lt;/pre&gt;&lt;/div&gt;

&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Arthur Norman</dc:creator><pubDate>Mon, 17 Jul 2017 20:02:50 -0000</pubDate><guid>https://sourceforge.netc1eb129545ff30d6ab2f23f02e52cebf62b3bc68</guid></item></channel></rss>