SoylentNews is people
SoylentNews
Nantero, the company that invented carbon nanotube-based non-volatile memory in 2001 and has been developing it since, has announced that seven chip fabrication plants are now manufacturing its Nano-RAM (NRAM) wafers and test chips in preparation for mass production, which requires the product designs to be completed. The company has announced that aerospace giant Lockheed Martin and Schlumberger Ltd., the world's largest gas and oil exploration and drilling company, will be customers seeking to use its chip technology. The memory, which can withstand 300 °C temperatures for years without losing data, is natively thousands of times faster than NAND flash and has virtually infinite read/write resilience. Nantero plans on licensing its intellectual property to allow others to create gum stick SSDs using DDR4 interfaces. NRAM has the potential to create memory that is vastly more dense that NAND flash, as its transistors can shrink to below 5 nanometers in size, three times more dense than today's densest NAND flash. At the same time, NRAM is up against a robust field of new memory technologies that are expected to challenge NAND flash in speed, endurance and capacity, such as Phase-Change Memory and Ferroelectric RAM (FRAM).
You may want to take a look at Memristors too.
Original Submission
(Score: 2) by maxwell demon on Thursday June 04 2015, @05:31AM
There are, however, cases where you want volatility; for example for temporary storage of security related stuff like entered passwords.
The Tao of math: The numbers you can count are not the real numbers.
(Score: 1, Interesting) by Anonymous Coward on Thursday June 04 2015, @06:42AM
The solution to this is to store the information encrypted and keep the encryption key in volatile SRAM. Add a little hardware to allow the CPU to act on the encrypted NVRAM transparently at a slight latency penalty. Not my idea; heard about this approach from a friend who does research into "The Machine" style operating systems.
(Score: 2) by kaszz on Thursday June 04 2015, @07:57AM
Some actors will make sure such circuitry never gets widespread..
(Score: 0) by Anonymous Coward on Thursday June 04 2015, @01:30PM
(Score: 2) by Gravis on Thursday June 04 2015, @08:05AM
There are, however, cases where you want volatility; for example for temporary storage of security related stuff like entered passwords.
1) Software should always securely erase passwords after they are used.
2) Passwords you should completely avoid keeping it in plain text at any point. With modern AMD64 processors, there is an x86 security extension that can encrypt and decrypt in a flash. before you say anything about intel chips, please go read what AMD64 really is because *gasp* Intel uses it too!
3) Ideally you would want to passwords stored in on processor cache to avoid it showing up in RAM
(Score: 2) by maxwell demon on Thursday June 04 2015, @09:03AM
Which doesn't help if the computer loses power while the password is still used. Remember, we're talking about a hypothetical NVRAM computer here, not the ordinary computer of today where information just vanishes as soon as the power is gone.
How many keyboards are able to encrypt passwords as you type them?
How do you get the password into the processor cache?
The Tao of math: The numbers you can count are not the real numbers.
(Score: 2) by Gravis on Thursday June 04 2015, @10:52AM
Which doesn't help if the computer loses power while the password is still used. Remember, we're talking about a hypothetical NVRAM computer here, not the ordinary computer of today where information just vanishes as soon as the power is gone.
even more reason to keep it in cache. however, with this kind of memory retention, it would be prudent to encrypt the RAM by default.
How many keyboards are able to encrypt passwords as you type them?
how many keyboards hold multiple keystrokes? that aside, this is about software, physical security is an entirely different problem.
How do you get the password into the processor cache?
with current processors, you leverage address translation in your favor so that part of your executable's virtual memory map is directed to cache or use a system service to do it for you.
if you truly are interested, the first step is to understand CPU cache. http://en.wikipedia.org/wiki/CPU_cache [wikipedia.org]