<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:atom="http://www.w3.org/2005/Atom">
	<channel>
		<title>Compile Time hashing 2.0</title>
		<description>Discuss Compile Time hashing 2.0</description>
		<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2</link>
		<lastBuildDate>Sun, 20 May 2012 07:58:41 +0000</lastBuildDate>
		<generator>JComments</generator>
		<atom:link href="http://www.chrissavoie.com/component/jcomments/feed/com_content/65/10" rel="self" type="application/rss+xml" />
		<item>
			<title>RE: Compile Time hashing 2.0</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-28</link>
			<description><![CDATA[you need compiler optimization turned on to have it generate compile time constant for CONSTHASH("deba") gcc -O3 generates movl $2084699690, %eax w/o optimization it will generate 2700 lines of computations]]></description>
			<dc:creator>borislav</dc:creator>
			<pubDate>Fri, 10 Dec 2010 08:23:45 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-28</guid>
		</item>
		<item>
			<title>Tommyknocker says:</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-26</link>
			<description><![CDATA[OK, whats wrong with this comment input thing??? It just cuts my comments! Another try: look at the Hash functions from your test "Full Hash Comparison Code" and from your GenConstHash.pl script. They are different, not only slightly different. So which one is the better one, which gave you the best results? Thanks in advance for your answer :-)]]></description>
			<dc:creator>Tommyknocker</dc:creator>
			<pubDate>Wed, 11 Feb 2009 02:33:21 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-26</guid>
		</item>
		<item>
			<title>Tommyknocker says:</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-24</link>
			<description><![CDATA[Hello Chris, I am confused by your different hash functions. When I download "Full Hash Comparison Code" ans look into your consthashmacro.h, I see the following code: #define CONSTHASH_FUNCTION(next, value) ((value]]></description>
			<dc:creator>Tommyknocker</dc:creator>
			<pubDate>Wed, 11 Feb 2009 02:29:26 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-24</guid>
		</item>
		<item>
			<title>Which one is your best hash fu</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-23</link>
			<description><![CDATA[Hello Chris, I am confused by your different hash functions. When I download "Full Hash Comparison Code" ans look into your consthashmacro.h, I see the following code: #define CONSTHASH_FUNCTION(next, value) ((value]]></description>
			<dc:creator>Tommyknocker</dc:creator>
			<pubDate>Wed, 11 Feb 2009 02:27:53 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-23</guid>
		</item>
		<item>
			<title>danger says:</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-18</link>
			<description><![CDATA[Yeah, crunch time definitely takes priority :-) Quickly, though -- I just ran the same tests with Visual C++ 2008 Express Edition, and ended up with even worse results than I had with GCC: now, neither the downloaded header nor the freshly-generated header generate fully compile-time hashes. Are you using the full Professional version of MSVC, perhaps?]]></description>
			<dc:creator>danger</dc:creator>
			<pubDate>Wed, 09 Jul 2008 00:14:46 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-18</guid>
		</item>
		<item>
			<title>csavoie says:</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-17</link>
			<description><![CDATA[:confused: It's definitely possibly that the versions could be different. Though I do need to say that with the latest changes I was only testing in visual studio 2008 so it's entirely possible that certain macro configurations might not work. I'll check them as soon as I get a chance, heading into alpha now so I don't have much mental energy. Thanks for all the responses.]]></description>
			<dc:creator>csavoie</dc:creator>
			<pubDate>Tue, 08 Jul 2008 18:54:12 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-17</guid>
		</item>
		<item>
			<title>danger says:</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-16</link>
			<description><![CDATA[Okay, I give up; it's eating my comments :-) Suffice to say, the two macros are slightly different. Other differences: the header I generated still caused a compile error when used in a case statement (see st4lk3r's comment above). However, it did successfully reduce CONSTHASH() calls to a compile-time constant, adding just a few bytes of code bloat per hash. Unfortunately, running the new version through your test suite generated a very large number of collisions. Was the version of the header in your source archive generated using a different version of GenConstHash.pl? Do you have any ideas why these problems might be occurring? (My tests were performed with GCC 3.4.2 under CygWin, for what it's worth). One last thing: thanks a ton; this is an excellent article and an extremely useful piece of code, even if it does have a few kinks remaining. I look forward to future revisions!]]></description>
			<dc:creator>danger</dc:creator>
			<pubDate>Tue, 01 Jul 2008 17:55:19 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-16</guid>
		</item>
		<item>
			<title>danger says:</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-15</link>
			<description><![CDATA[Yours: #define CONSTHASH_FUNCTION(next, value) ((value]]></description>
			<dc:creator>danger</dc:creator>
			<pubDate>Tue, 01 Jul 2008 17:54:06 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-15</guid>
		</item>
		<item>
			<title>danger says:</title>
			<link>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-14</link>
			<description><![CDATA[Yes, this is weird. The version of consthashmacro.h included in the source archive definitely doesn't seem to be compile-time. When I use it in a simple test program, it causes about 200 bytes of code bloat per call to CONSTHASH(). I tried regenerating the file with: GenConstHash.pl -l 64 -r CONSTHASH -t consthashmacro2.h and the resulting file contained a slightly different definition of the CONSTHASH_FUNCTION() macro. Yours: #define CONSTHASH_FUNCTION(next, value) ((value]]></description>
			<dc:creator>danger</dc:creator>
			<pubDate>Tue, 01 Jul 2008 17:53:15 +0000</pubDate>
			<guid>http://www.chrissavoie.com/articles-mainmenu-29/15-research/65-hash2#comment-14</guid>
		</item>
	</channel>
</rss>

