IBM scientists say radical new ‘in-memory’ architecture will speed up computers
1–10 of 138 posts
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#2This is the promise of memristors. Despite innumerable articles being written about neuromorphic architectures like they'll be something miraculous, this ability to change from functioning as a bit of memory to being a bunch of functional logic on the fly at the speed of a memory read? That's going to be crazy. It will open up possibilities that we probably can't even imagine right now.
I've never understood why people don't get more excited about memristors. They could replace basically everything. Assuming someone can master their manufacture, they should be more successful than transistors. Of course, I'm still waiting to be able to buy a 2000 ppi display like IBM's R&D announced creating back in the late 1990s or so... so I guess I'd best not hold my breath.
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#3"The result of the computation is also stored in the memory devices, and in this sense the concept is loosely inspired by how the brain computes."
For anyone who is interested in a simple model of how the brain does this, check out "associative memories". The basic idea is that networks of neurons both store memory (in their synapses) and perform the computations to retrieve or recall those memories.
A simple example is the Hopfield network, a single layer of neurons that are recurrently connected with a specific update function: https://en.wikipedia.org/wiki/Hopfield_network
Another is two layers of neurons that are reciprocally connected called a Bidirectional Associative Memory (BAM): https://en.wikipedia.org/wiki/Bidirectional_associative_memo...
edit: grammar
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#4This is the promise of memristors. Despite innumerable articles being written about neuromorphic architectures like they'll be something miraculous, this ability to change from functioning as a bit of memory to being a bunch of functional logic on the fly at the speed of a memory read? That's going to be crazy. It will open up possibilities that we probably can't even imagine right now. I've never understood why peop…
That's why. Memristor articles were all over the news for a bit, but have died down as either companies realized they're too hard to manufacturer or companies have quietly plodded toward making them competitive.
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#5This is the promise of memristors. Despite innumerable articles being written about neuromorphic architectures like they'll be something miraculous, this ability to change from functioning as a bit of memory to being a bunch of functional logic on the fly at the speed of a memory read? That's going to be crazy. It will open up possibilities that we probably can't even imagine right now. I've never understood why peop…
Well with most claims that generalize computing power, such as the title of this, it's mostly aimed at something more specialized.
"The researchers believe this new prototype technology will enable ultra-dense, low-power, and massively parallel computing systems that are especially useful for AI applications."
So it seems that would benefit AI/Machine Learning more than GP processing, from that statement.
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#6What they describe sounds like a pipeline architecture or a systolic array, or a network of interconnected computers. None of these are new concepts from an architectural point of view, but the actual dimensioning could be new.
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#7This is the promise of memristors. Despite innumerable articles being written about neuromorphic architectures like they'll be something miraculous, this ability to change from functioning as a bit of memory to being a bunch of functional logic on the fly at the speed of a memory read? That's going to be crazy. It will open up possibilities that we probably can't even imagine right now. I've never understood why peop…
I can't talk for other people, but I personally don't try to get over-excited over anything unless I see a working proof of concept where the advantages are clear. It's kinda unfortunate, because there's great work being done in research that I should probably feel much more excited about, but I'm continuously bombarded with new discoveries that I never hear about again.
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#8This is the promise of memristors. Despite innumerable articles being written about neuromorphic architectures like they'll be something miraculous, this ability to change from functioning as a bit of memory to being a bunch of functional logic on the fly at the speed of a memory read? That's going to be crazy. It will open up possibilities that we probably can't even imagine right now. I've never understood why peop…
Well with most claims that generalize computing power, such as the title of this, it's mostly aimed at something more specialized. "The researchers believe this new prototype technology will enable ultra-dense, low-power, and massively parallel computing systems that are especially useful for AI applications." So it seems that would benefit AI/Machine Learning more than GP processing, from that statement.
The researchers are those focused in AI applications and as such I expect them to focus advertising that. This technology has far more potential than just that, right down into simple stuff like realtime waveform or spectrographic analysis. Being able to store and process on the exact same 'silicon' without having to run out a few inches to retrieve data will boost latency-sensitive applications.
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#9This is the promise of memristors. Despite innumerable articles being written about neuromorphic architectures like they'll be something miraculous, this ability to change from functioning as a bit of memory to being a bunch of functional logic on the fly at the speed of a memory read? That's going to be crazy. It will open up possibilities that we probably can't even imagine right now. I've never understood why peop…
Well with most claims that generalize computing power, such as the title of this, it's mostly aimed at something more specialized. "The researchers believe this new prototype technology will enable ultra-dense, low-power, and massively parallel computing systems that are especially useful for AI applications." So it seems that would benefit AI/Machine Learning more than GP processing, from that statement.
I'm not so sure. CPUs are stupidly fast, now - the bottleneck is virtually always the time it takes to move data in and out of memory. What if, instead of trying to bring the data to the calculation, you brought the calculation to the data? You could speed up everything.
Fun to think about: in this scenario, the main job of an "optimizing compiler" would be to organize things spatially so that, as a calculation moved around in memory, it always found itself right next to the data it needed, when it needed it.
Re: IBM scientists say radical new ‘in-memory’ architecture will speed up computers
#10Earlier quoted context omitted.
Well with most claims that generalize computing power, such as the title of this, it's mostly aimed at something more specialized. "The researchers believe this new prototype technology will enable ultra-dense, low-power, and massively parallel computing systems that are especially useful for AI applications." So it seems that would benefit AI/Machine Learning more than GP processing, from that statement.
I'm not so sure. CPUs are stupidly fast, now - the bottleneck is virtually always the time it takes to move data in and out of memory. What if, instead of trying to bring the data to the calculation, you brought the calculation to the data? You could speed up everything . Fun to think about: in this scenario, the main job of an "optimizing compiler" would be to organize things spatially so that, as a calculation move…
It makes perfect sense, because this is already how good programmers look at the task, especially considering the penalties for memory access on secondary devices.