Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.programming.threads > #4394
| From | Sky89 <Sky89@sky68.com> |
|---|---|
| Newsgroups | comp.programming.threads |
| Subject | About Reversible algorithms and Energy-efficient computing |
| Date | 2018-06-07 17:02 -0400 |
| Organization | A noiseless patient Spider |
| Message-ID | <pfc6i2$e5$2@dont-email.me> (permalink) |
Hello, About Reversible algorithms and Energy-efficient computing Reversible algorithms behave the same way. “If you can play everything backwards, then no energy has escaped during your computation,” says Demaine. “That’s good news. It means we can effectively sidestep Landauer’s principle.” While conservative computing may enable programs to run twice as fast, reversible computing could enable them to run millions of times faster. Using specially devised theoretical models, Demaine and Lynch have spent the past six months analyzing basic algorithms to see whether they can be made reversible—or more reversible. (There’s a fundamental limit on how reversible some algorithms can be.) Already they’ve found more-efficient replacements for some algorithms used in everyday computational tasks such as sorting, searching, and finding the shortest path between two points in a network. One example is “binary search trees,” which are procedures for organizing data so the data can be retrieved quickly. According to Demaine, binary search trees are used in nearly every computer ever made, and they involve millions of functions and a lot of energy consumption. “But with a couple of tricks, we got energy use down to zero,” he says. “With the new algorithms, we require only the energy needed to store the data, no additional energy to organize it.” Demaine is pleased with their progress. “It’s like starting over,” he says. “Take all the algorithms you learned in your undergraduate class and throw them out the window, or look at all existing algorithms and say, OK, this is bad, let’s make it better.” And their results so far “just scratch the surface of what’s possible,” he says, noting in particular the huge potential for energy savings in procedures used for processing big data, such as when running network routers or performing web searches. Read more here: http://energy.mit.edu/news/energy-efficient-computing/ Thank you, Amine Moulay Ramdane.
Back to comp.programming.threads | Previous | Next | Find similar | Unroll thread
About Reversible algorithms and Energy-efficient computing Sky89 <Sky89@sky68.com> - 2018-06-07 17:02 -0400
csiph-web