tag:blogger.com,1999:blog-593563533834706486.post4660615959314177360..comments2024-03-28T10:35:10.453-04:00Comments on Yet Another Math Programming Consultant: Big Iron vs my laptopErwin Kalvelagenhttp://www.blogger.com/profile/09496091402502236997noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-593563533834706486.post-66069982624938532702014-12-03T09:15:31.352-05:002014-12-03T09:15:31.352-05:00Memory latency and memory throughput are both impo...Memory latency and memory throughput are both important and could be worse on your laptop. Laptops, especially ultrabooks, sometimes reduce the processor speed to prevent overheating.<br /><br />Somewhat related: http://www.quora.com/Which-optimization-algorithms-are-good-candidates-for-parallelization-with-MapReduceAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-593563533834706486.post-18810960915149868712014-12-03T04:18:03.791-05:002014-12-03T04:18:03.791-05:00I currently run my OptaPlanner benchmarks on a 4 y...I currently run my OptaPlanner benchmarks on a 4 year old Xeon desktop (4GB RAM) and surprisingly, it's faster(!) than on my 1 year old laptop (8GB RAM). I still haven't found a good explanation for that, although I suspect CPU cache size is involved.<br /><br />We're doing experiments with big iron too, which lead me to the same conclusion in the past: throwing hardware at a problem doesn't help much. Better algorithms/models indeed do.Geoffrey De Smethttps://www.blogger.com/profile/14668727421421215806noreply@blogger.com