at the Beckman Institute for Advanced Science and Technology at
Illinois have improved upon a novel, faster form of memory to make it use power
The new memory falls under a class of devices called power phase-change materials (PCM), which store
data as resistance, rather than charge. Previous attempts at PCM memory
had been fast, but less power efficient due to large contacts.
The Beckman team, led by computer engineering professor Eric Pop [profile], explored the prospects of much
smaller phase-change memory using nanoscale contacts and a
gapped carbon nanotube (a tube with a missing
segment in the middle -- essentially two extremely almost touching nanotubes).
The results were that power consumption was cut by a factor of 100 from current
generation phase change memory (the team did not put this in context with
traditional memory, rather than to say it was much more efficient).
Professor Pop optimistically describes, "We're not just talking about lightening
our pockets or purses. This is also important for anything that has to operate
on a battery, such as satellites, telecommunications equipment in remote
locations, or any number of scientific and military applications."
While semiconductor-based phase change memory is arguably commercially viable
due to its speed, this carbon nanotube phase change memory may be one case
where the actual device fails to live up to the hype.
How big an impact will this really have on mobile battery
life, if it makes it to the market? It’s likely that it won’t improve
battery life that much in the long run. Currently, the biggest power
wasters on your mobile phone are your screen/GPU, your CPU, and the wireless
Furthermore, there's no established process to create chips with gapped carbon
nanotubes affordably in a fab. Creating carbon nanotube PCM chips would
be extremely expensive with today's technology.
Ultimately, this could bump memory access speeds slightly, and bump the battery
life 2 or 3 percent in the best-case scenario. While it's true every bit
counts, the cost may more that negate the minimal gains.
In other words you're spending a tremendous premium to go from semiconductor to
carbon nanotubes and the only reward for that switch (reduced power
consumption) is extremely minimal.
And the real question here is "why memory?" There are so many
more attractive power efficiency targets -- like transistors (for CPUs), the
battery itself, wireless modems, and displays.
When and if carbon nanotube production becomes affordable, this idea may become
marginally useful. But for now chalk this one to a lot of bark but not
The team's study is published [abstract] in the journal Science.
quote: Not bitter, just annoyed that he uses this as a forum for his opinions that don't appear to be very well thought out.