4 ms·

HP outlines memristor-based computer memory of the future

11 points by andygeers 17y ago

ableal 17y ago
I'm confused. The "April 8th HP announcement" seems to be related to the publication in Nature of a paper from May 2009. A bit of search turned this up: http://www.physorg.com/news190016024.html http://www.physorg.com/news190016024.html (paywalled) http://www.nature.com/nature/journal/v464/n7290/full/nature08940.html http://www.nature.com/nature/journal/v464/n7290/full/nature0... (nice pic, but just an echo of yesterday's NYT pre-news) http://www.mnn.com/green-tech/computers/blogs/computer-chip-breakthrough-mimics-brain-synapse http://www.mnn.com/green-tech/computers/blogs/computer-chip-... Anyone has better ? (meaning more than was available in the news of six months or so ago) P.S. Good short critical comment here: http://blogs.nature.com/news/thegreatbeyond/2010/04/memristance_is_not_futile.html http://blogs.nature.com/news/thegreatbeyond/2010/04/memrista... And finally, the HP press release that search didn't show - had to go manually to the hp.com news page: http://www.hp.com/hpinfo/newsroom/press/2010/100408xa.html http://www.hp.com/hpinfo/newsroom/press/2010/100408xa.html
- glymor 17y ago
  The new work demonstrates that it's possible to build a NAND gate using a combination of three memristors, but only if you use a frequently overlooked logical operation called "material implication." As the authors describe it, for Boolean states p and q, a material implication is "p implies q"—if p is true, then q must also be. http://arstechnica.com/science/news/2010/04/complete-logic-set-performed-using-memristors.ars http://arstechnica.com/science/news/2010/04/complete-logic-s...
  - ableal 17y ago
    Thanks, that's one of my interests. Just the other day, I was looking at the table here: http://en.wikipedia.org/wiki/Logic_gate#Logic_gates http://en.wikipedia.org/wiki/Logic_gate#Logic_gates (implication is #14 of 16 possible boolean functions). Sometimes, going back to relay logic helps. For instance, Clause Shannon's BSTJ Jan.1948 paper, "The Synthesis of Two-Terminal Switching Circuits". Or his 1940 master's thesis, A Symbolic Analysis of Relay and Switching Circuits (http://dspace.mit.edu/handle/1721.1/11173 http://dspace.mit.edu/handle/1721.1/11173)
wheaties 17y ago
The NY Times article was more informative: http://www.nytimes.com/2008/05/01/technology/01hp-Web.html http://www.nytimes.com/2008/05/01/technology/01hp-Web.html Looks like it's 10x slower than DRAM. It might make it to market in 3 years but if it's speed is as slow, who will buy? Regardless, I'm still encouraged science marches onward.
- khafra 17y ago
  If it's RAM that's nonvolatile and one order of magnitude slower than DRAM, shouldn't we just call it flash memory? A bit disappointing that a huge leap in applying theory to physical objects duplicated what we already have with NAND memory.

11 points by andygeers 17y ago

I'm confused. The "April 8th HP announcement" seems to be related to the publication in Nature of a paper from May 2009. A bit of search turned this up: http://www.physorg.com/news190016024.html http://www.physorg.com/news190016024.html (paywalled) http://www.nature.com/nature/journal/v464/n7290/full/nature08940.html http://www.nature.com/nature/journal/v464/n7290/full/nature0... (nice pic, but just an echo of yesterday's NYT pre-news) http://www.mnn.com/green-tech/computers/blogs/computer-chip-breakthrough-mimics-brain-synapse http://www.mnn.com/green-tech/computers/blogs/computer-chip-... Anyone has better ? (meaning more than was available in the news of six months or so ago) P.S. Good short critical comment here: http://blogs.nature.com/news/thegreatbeyond/2010/04/memristance_is_not_futile.html http://blogs.nature.com/news/thegreatbeyond/2010/04/memrista... And finally, the HP press release that search didn't show - had to go manually to the hp.com news page: http://www.hp.com/hpinfo/newsroom/press/2010/100408xa.html http://www.hp.com/hpinfo/newsroom/press/2010/100408xa.html

glymor 17y ago
The new work demonstrates that it's possible to build a NAND gate using a combination of three memristors, but only if you use a frequently overlooked logical operation called "material implication." As the authors describe it, for Boolean states p and q, a material implication is "p implies q"—if p is true, then q must also be. http://arstechnica.com/science/news/2010/04/complete-logic-set-performed-using-memristors.ars http://arstechnica.com/science/news/2010/04/complete-logic-s...
- ableal 17y ago
  Thanks, that's one of my interests. Just the other day, I was looking at the table here: http://en.wikipedia.org/wiki/Logic_gate#Logic_gates http://en.wikipedia.org/wiki/Logic_gate#Logic_gates (implication is #14 of 16 possible boolean functions). Sometimes, going back to relay logic helps. For instance, Clause Shannon's BSTJ Jan.1948 paper, "The Synthesis of Two-Terminal Switching Circuits". Or his 1940 master's thesis, A Symbolic Analysis of Relay and Switching Circuits (http://dspace.mit.edu/handle/1721.1/11173 http://dspace.mit.edu/handle/1721.1/11173)

wheaties 17y ago

The NY Times article was more informative: http://www.nytimes.com/2008/05/01/technology/01hp-Web.html http://www.nytimes.com/2008/05/01/technology/01hp-Web.html Looks like it's 10x slower than DRAM. It might make it to market in 3 years but if it's speed is as slow, who will buy? Regardless, I'm still encouraged science marches onward.

khafra 17y ago
If it's RAM that's nonvolatile and one order of magnitude slower than DRAM, shouldn't we just call it flash memory? A bit disappointing that a huge leap in applying theory to physical objects duplicated what we already have with NAND memory.