P-complete and the limits of parallelizationhttp://blogs.msdn.com/devdev/archive/2007/09/07/p-complete-and-the-limits-of-parallelization.aspxWe're entering an era where CPU clock speeds will soon cease to scale upwards and instead CPU manufacturers are planning to put more and more independent cores on a chip. Intel plans to release an 80-core chip within 5 years . Consequently the researchen-USCommunityServer 2.1 SP1 (Build: 61025.2)re: P-complete and the limits of parallelizationhttp://blogs.msdn.com/devdev/archive/2007/09/07/p-complete-and-the-limits-of-parallelization.aspx#4836294Sun, 09 Sep 2007 02:37:16 GMT91d46819-8472-40ad-a661-2c78acb4018c:4836294Isaac<p>Nice work. &nbsp;Although I wish you can post more often.</p>A Multicore Language Timetable (Waste More Hardware!) &laquo; SmoothSpan Bloghttp://blogs.msdn.com/devdev/archive/2007/09/07/p-complete-and-the-limits-of-parallelization.aspx#4846189Sun, 09 Sep 2007 21:19:01 GMT91d46819-8472-40ad-a661-2c78acb4018c:4846189A Multicore Language Timetable (Waste More Hardware!) « SmoothSpan Blog<p>PingBack from <a rel="nofollow" target="_new" href="http://smoothspan.wordpress.com/2007/09/08/a-multicore-language-timetable-waste-more-hardware/">http://smoothspan.wordpress.com/2007/09/08/a-multicore-language-timetable-waste-more-hardware/</a></p> re: P-complete and the limits of parallelizationhttp://blogs.msdn.com/devdev/archive/2007/09/07/p-complete-and-the-limits-of-parallelization.aspx#4854231Mon, 10 Sep 2007 19:13:59 GMT91d46819-8472-40ad-a661-2c78acb4018c:4854231Gabe Moothart<p>This is really neat stuff. Have you read any of the Google infrastructure papers (particularly the ones on &lt;a href=&quot;<a rel="nofollow" target="_new" href="http://research.google.com/archive/mapreduce.html&quot;&gt;MapReduce&lt;/a&gt;">http://research.google.com/archive/mapreduce.html&quot;&gt;MapReduce&lt;/a&gt;</a> and &lt;a href=&quot;<a rel="nofollow" target="_new" href="http://research.google.com/archive/sawzall.html&quot;&gt;Sawzall&lt;/a&gt;">http://research.google.com/archive/sawzall.html&quot;&gt;Sawzall&lt;/a&gt;</a>)? They use a highly restricted programming model which allows them to automatically parallelize operations across hundreds or thousands of machines. The Sawzall paper includes a couple of examples in which the exact solution could not be easily distributed, so they used a parallelizable approximation algorithm.</p>