HP is thinking of the long term as evidenced by their estimate of 2016 as the release date for the first viable DIMMs using memristors. Their plans are much larger than a new type of memory, they are planning a scalable architecture dubbed The Machine which will take advantage of the high speed and lower power needs of memristors to develop a new type of system which will need to use photonic interconnects to keep up with the memristors. They see this scaling from tiny devices and mobile phones with 100TB of storage to supercomputers whose speeds will make a mockery of the current record holder, the Fujitsu K. Of course many of the claims The Register heard HP make should be taken with a grain of salt, after all the memristor was originally predicted to hit the market a year ago. It is something to look forward to, who doesn't want faster, denser and more power efficient storage?
"The beleaguered IT giant plans to rejuvenate itself with a set of advanced technologies that, when combined, make a device called "The Machine" that can be as small as a smartphone and as large as a 160-rack supercomputer, the company announced at its HP Discover event in Las Vegas on Wednesday."
Here is some more Tech News from around the web:
- Mozilla To Sell '$25' Firefox OS Smartphones In India @ Slashdot
- Console OS will let you run Android on a Windows PC or tablet @ The Inquirer
- Gigabyte sees over 50% growth in 2013 profits @ DigiTimes
- Evernote taken out by DDoS attack @ The Register
- New Version of 3DMark Basic Edition (2013) Now Available on NGOHQ
- Net Neutrality: FCC Hack is a Speed Bump on the Internet Fast Lane @ Hack a Day
HP Labs are probably spending
HP Labs are probably spending most their time turning memristors into cloud-enabled, non user serviceable, hardware-as-a-service consumables with tens of thousands of synergistic redundancies.
A prototype 3D memory cube
A prototype 3D memory cube made of memristors, linked to a processor by some photonic fiber that, he said, should be capable of 6 terabytes per second of data transfer. Latency times of, in a cluster example ” where any single byte within that whopping 160PB cluster can be addressed in under 250 nanoseconds.” not bad for that much data. Getting closer to a star trek isolinear chip, and those backup times will be fast. But how many cat videos can you store in a kiloquad, and what are Google’s rules of acquisition.
That is some serous bandwidth, so tablets will be able to play today’s crisis on Ultra settings.