Intel Labs and ITRI Announce Fast Memory Array

By Wolfgang Gruener
published

Intel is starting a memory research project with the Industrial Technology Research Institute (ITRI) of Taiwan to address the performance gap and to reduce the power consumption of memory.

Intel announced the development of a mysterious, super-fast memory array the company expects to deploy in dense environments such as ultrabooks, tablets, smartphones and cloud computing data centers that presumably leverage micro server technologies.

The technology without a name was developed by Intel in a joint-effort with the Industrial Technology Research Institute (ITRI) in Taiwan. The project is expected to be completed within five years and funding of $5 million.

According to media reports, the memory is energy-efficient to achieve longer battery life and faster integration of mobile data, while enhancing performance with improved graphics and overall better user experience on mobile devices.

Intel CTO Justin Rattner said that testing of the devices will begin in 2013 and cover devices ranging from smartphones to supercomputers.

Contact Us for News Tips, Corrections and Feedback

Wolfgang Gruener
10 Comments Comment from the forums
  • azathoth
    "funding of $5 million"
    I would believe a better decision would be to raise the funding to at least $10 million. More efficient RAM in the mobile sector could really take off for them, as long as it is done correctly.
    Reply
  • dozerman
    Good. Faster memory means that caches can be smaller and chip transistors can be more adequately set to better purposes. This is a win for every part of the computer industry assuming intel doesn't horde the findings to themselves.
    Reply
  • milktea
    I bet these memory must be in BGA form which requires them to be soldered onto the main board. Much like how Intel might do away socketed CPU into BGA. :)
    Reply
  • A Bad Day
    dozermanGood. Faster memory means that caches can be smaller and chip transistors can be more adequately set to better purposes. This is a win for every part of the computer industry assuming intel doesn't horde the findings to themselves.
    Or repeats the Rambus fiasco...
    Reply
  • CaedenV
    milkteaI bet these memory must be in BGA form which requires them to be soldered onto the main board. Much like how Intel might do away socketed CPU into BGA.Heck, ram is relatively cheap and easy to make. Just throw 4GB on the chip and call it a day. 5 years from now we are talking about sub 10nm die process, so I am sure that space would not be much of an issue.
    dozermanGood. Faster memory means that caches can be smaller and chip transistors can be more adequately set to better purposes. This is a win for every part of the computer industry assuming intel doesn't horde the findings to themselves.If Intel gets into the memory business it is NOT a win for everybody. It means major job cuts for every memory manufacturer out there. Makes faster systems that can 'race to sleep' more efficiently, so it would be a potential win for consumers, but definitely not a win for everyone.
    Reply
  • serendipiti
    Could be speaking of cache memory ?
    Reply
  • milktea
    serendipitiCould be speaking of cache memory ?Level 4 cache maybe? That runs at level 1 cache speed even?
    Reply
  • Thunderfox
    Five million is pocket change to Intel. Why not just develop it themselves?
    Reply
  • A Bad Day
    ThunderfoxFive million is pocket change to Intel. Why not just develop it themselves?
    Because grinding AMD, ARM, and Nividia into submission (or bankruptcy) is more fun?
    Reply
  • TeraMedia
    $5 million is pocket change indeed. This article just doesn't make a lot o sense. It says, "The technology was developed..." which means that there is already a proof-of-concept. Spending a mere $1m per year to bring a researched, market-changing technology to production just doesn't make any sense when Intel is involved. They probably spend more than that on a trade show. They definitely spend more money than that developing technologies such as 10 GbE. Something doesn't add up. Either the technology isn't as spectacular as the article makes it sound, or Intel is spending $5M jointly, but $25M in-house on this technology.
    Reply