SoylentNews is people
SoylentNews
Submitted via IRC for SoyCow3941
[...] The team of researchers have devised a way to factor large composite integers by harnessing the massive parallelism of novel computer architectures that mimic the functioning of the mammalian brain. So called neuromorphic computers operate under vastly different principles than conventional computers, such as laptops and mobile devices, all based on an architecture described by John von Neumann in 1945.
In the von Neumann architecture, memory is separate from the central processing unit, or CPU, which must read and write to memory over a bus. This bus has a limited bandwidth, and much of the time, the CPU is waiting to access memory, often referred to as the von Neumann bottleneck.
Neuromorphic computers, on the other hand, do not suffer from a von Neumann bottleneck. There is no CPU, memory, or bus. Instead, they incorporate many individual computation units, much like neurons in the brain.
These units are connected by physical or simulated pathways for passing data around, analogous to synaptic connections between neurons. Many neuromorphic devices operate based on the physical response properties of the underlying material, such as graphene lasers or magnetic tunnel junctions. Because of this, these devices consume orders of magnitude less energy than their von Neumann counterparts and can operate on a molecular time scale. As such, any algorithm capable of running on these devices stands to benefit from their capabilities.
Source: Brain-like computers moving closer to cracking codes
(Score: 2) by takyon on Monday March 26 2018, @02:45AM (3 children)
Quantum computing was supposed to be the killer of RSA and other encryption techniques, but I guess neuromorphic can do it too.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 0) by Anonymous Coward on Monday March 26 2018, @03:18AM
You seem to be implying that because quantum computers weren't practical for cracking crypto before a second technology which could do so started to be developed that they will never progress from where they are now?
What?
(Score: 3, Interesting) by JoeMerchant on Monday March 26 2018, @03:29AM
Apparently not yet... Neuromorphic may be outperforming current best in class heuristics, but that's a long way from "cracking."
Also, I wonder, how do they know when they've cracked the code? If there's another layer of encryption that renders the decrypted message as basically white noise, will their algorithm be able to tell that it has succeeded? Maybe RSA, but symmetric methods I doubt.
🌻🌻 [google.com]
(Score: 2) by maxwell demon on Monday March 26 2018, @05:06AM
There's little reason to assume so. It may be faster, but as classical computer, it still has the same complexity constraints. Just like a von-Neumann computer is vastly faster than a Turing machine, but an algorithm that has exponential complexity on a Turing machine also has it on a von-Neumann computer.
What it means is that you'll have to use longer keys than expected to keep RSA safe.
The Tao of math: The numbers you can count are not the real numbers.