'Instant on' computers moved a step closer today, according to IBM and Infineon claimed, thanks to the two companies' joint efforts to develop magnetic memory chips.
Boffins from both companies believe Magnetic Random Access Memory (MRAM) could now come to market as early as 2005. To help bring that about, they will demonstrate a 0.18 micron - the smallest size reported to date for MRAM technology, the companies claim - chip containing 128Kb of MRAM. Integrating MRAM with a digital chip process is a major step forward in bring the technology to market.
MRAM essentially comprises a grid of microscopic magnets rather than the transistor-based memory cells used to create standard SDRAM. But MRAM requires less power to flip the magnetic field of each magnet to change it from a 0 to a 1 than an SDRAM cell, can do so more quickly and doesn't lose its data when the power's cut, just like Flash memory. But, again, MRAM is faster and uses less power than Flash. Non-volatility also means that unused memory can be powered down to conserve energy.
IBM and Infineon suggest that this will allow computer makers to offer machines that boot up almost instantly - the exact state of the processor can be saved to MRAM, ready to be returned to that state as soon as the power is switched on.
MRAM's speed also raises the possibility of further blurring the line between short-term storage (memory) and long-term storage (hard drive). Virtual memory schemes already use the latter to enhance the former, but MRAM may be quick and one day cheap enough to end the distinction between the two. A non-volatile extended memory map would allow computers to retain applications, and permanent and temporary data in memory. Files and applications could remain permanently open, whether they were being used or not, with the precise state retained until next needed.
Actually, the concept of files would probably vanish, as the whole concept of a file system, geared as it is toward long-term storage, would be rendered unnecessary. Just a thought.
Such a world is some way away - for now, the goals of IBM and Infineon are more prosaic: developing the technology and the fabrication processes behind it. ®