Wake me when this stuff actually sees a mainstream product launch. Until then, it's just another vaporware.
Carbon nanotube memory tech gets great big cash dollop
Future memory hardware startup Nantero has had a big funding round, suggesting its technology is getting closer to becoming a reality. It got $31.5m in an over-subscribed round to continue developing its nanotube-based non-volatile RAM (NRAM) semiconductor technology, which it says has DRAM read/write speed and is ultra-high …
-
Friday 12th June 2015 11:22 GMT chris 17
The day when RAM & NV storage become indistinguishable is almost upon us :).
I've been saying for a while its not far off, especially with the fast 2GB/s drives we see now (MacBook) being an indication of what is to come.
We only use RAM because NV storage is too slow for the CPU to do meaningful work, even RAM can be too slow so we have on cpu caches. Having a 1TB disk that's also RAM would be epic.
-
Friday 12th June 2015 11:40 GMT Anonymous Coward
Expect hibernation and shutdown times to be *loooong* on company laptops.
If this is used in place of DRAM, it will be necessary to zero all the memory prior to sleep or shutdown to prevent embarrassing data loss in the event of a stolen laptop... unless we start encrypting memory contents, that is.
-
Friday 12th June 2015 15:22 GMT Dr. Mouse
Re: Expect hibernation and shutdown times to be *loooong* on company laptops.
If this is used in place of DRAM, it will be necessary to zero all the memory prior to sleep or shutdown to prevent embarrassing data loss in the event of a stolen laptop... unless we start encrypting memory contents, that is.
I think, with this kind of tech, we would need to start thinking in a very different manner.
If it, or one of the alternatives, proves to be as good as claimed, we will no longer have RAM and storage. They will be the same thing. So yes, we would probably need to start putting encryption on RAM as we do on storage devices. There would need to be transparent decryption in hardware.
However, we already have the same vulnerability as you are talking about in current tech, just slightly different. As I understand it, if the contents of a disk is encrypted, it is read through a driver and stored in RAM in an unencrypted form, at least for a short time. Now think of the number of users who just put their laptop to sleep: This will keep the contents of the RAM. Someone can come along, swipe the laptop, "freeze" the SODIMMs, transplant them to another machine and read the data. It is not quite so simple, of course, but it can be done.
So, if transparent encryption is baked in to the specification, there is actually a reduction in the vulnerability to data theft: All data on the non-volatile RAM is encrypted on write, decrypted on read. If they transplant the modules, they can only read encrypted data (assuming the key is stored elsewhere, preferably with a passcode of some kind to access it, and preferably in a secure element with a low probability of being hacked).
Therefore I humbly submit for your consideration that this could very well end up making our systems MORE secure.
-
Friday 12th June 2015 17:21 GMT Anonymous Coward
Re: Expect hibernation and shutdown times to be *loooong* on company laptops.
"Therefore I humbly submit for your consideration that this could very well end up making our systems MORE secure."
But wouldn't this transparent encryption create a lag that may not be desired in performance-intensive applications? Creating a possible conflict of interests if a device is BOTH performance-intensive AND portable? Anyway, this wouldn't do much for the most common form of data theft: taking the entire laptop WITH the cipher chips in it, allowing them to work on the problem in situ.
-
-
-
-
This post has been deleted by its author
-
Friday 12th June 2015 14:12 GMT Charles 9
Not just crap but hard to handle. It only worked at a certain minimum temperature, so it had to be literally warmed up to work, which is why Konami had to come up with their noted "Morning Music" as a warm-up signal for their Bubble System games. Not to mention the reading process was destructive, meaning you had to feed the data back in as soon as it was read, and if something went wrong in between, the whole works got corrupted.
-
-
-
Friday 12th June 2015 19:45 GMT Anonymous Coward
Re: I missed something
very low. Not zero of course, but the C tubes used for circuitry in these type of devices are near superconductive. Latency is down to how slowly the doped atoms change from one molecular bonding to another.
The holdup is probably cost of manufacturing larger storage sizes. Building complex NT structures at scale number without defects is still a bit of a problem. I will be watching the price tag per bit/byte on this one rather than latency.
-
Friday 12th June 2015 13:32 GMT Andy The Hat
In 2012 they reckon they were manufacturing 4Mb devices ... Their press release archive basically says 'we've got some more money (again)' but blatantly does not say "in three years we've been able to develop 8Mb devices" or "we've managed to commercialise a production system for 4Mb devices" or even "This is our demonstration product, look at what it can do, and here are the specs ...".
I think vapourware is about right.
-
Biting the hand that feeds IT
