Discovery of a 'holy grail' with the invention of universal computer memory

[sman]

Discovery of a 'holy grail' with the invention of universal computer memory

"https://soylentnews.org/article.pl?sid=19/06/20/1821219"

[An] electronic memory device -- described in research published in Scientific Reports -- promises to transform daily life with its ultra-low energy consumption. [...] The device is the realisation of the search for a "Universal Memory" which has preoccupied scientists and engineers for decades.

Physics Professor Manus Hayne of Lancaster University said: "Universal Memory, which has robustly stored data that is easily changed, is widely considered to be unfeasible, or even impossible, but this device demonstrates its contradictory properties."

A US patent has been awarded for the electronic memory device with another patent pending, while several companies have expressed an interest or are actively involved in the research.

The inventors of the device used quantum mechanics to solve the dilemma of choosing between stable, long-term data storage and low-energy writing and erasing. The device could replace the $100bn market for Dynamic Random Access Memory (DRAM), which is the 'working memory' of computers, as well as the long-term memory in flash drives.

[...] Professor Hayne said: "The ideal is to combine the advantages of both without their drawbacks, and this is what we have demonstrated. Our device has an intrinsic data storage time that is predicted to exceed the age of the Universe, yet it can record or delete data using 100 times less energy than DRAM."

[exile360]

I guess the only remaining question is speed.  Energy consumption is not the only consideration with regards to RAM, especially modern classifications like DDR4 which is in most modern systems, the upcoming DDR5 JDEC standard, GDDR5/GDDR5X, GDDR6 and HBM/HBM2 which are used on modern GPUs, and upcoming standards like HBM2E, all of which must operate at extremely high speeds with ever growing amounts of bandwidth in order to keep the cores of CPUs and GPUs fed with data.  A single solution would be ideal that replaces both, is as fast as the fastest DRAM, and is capable of retaining data for at least as long as a modern NVMe SSD as this would eliminate multiple bottlenecks for PCs and other computing devices (including game consoles; the next generation of which are boasting new storage technology that should dramatically reduce or even eliminate load times, even in large games with high resolution assets, which in effect would be the same level of performance delivered by a hypothetical 'all-in-one' type of storage medium).

Intel's Optane technology has been on the market for a while, providing a relatively affordable means of increasing speed and throughput from storage to DRAM, however this is more of a half-measure more suited to servers where the scale of things is increased massively over standard desktop computing devices.

HBM/HBM2/HBM2E are pretty much the fastest forms of memory on the market right now, with bandwidth and speeds massively exceeding those of more traditional VRAM solutions like GDDR5 and GDDR6, however their high cost and increased cooling requirements (particularly since they must be situated pretty much right next to or even within the GPU die to eliminate trace latency) often makes them prohibitively expensive, and in real-world benchmarks they don't show the kinds of increases in performance over GDDR5 and GDDR6 based solutions in more everyday applications like games; they do however provide significant increases in many professional workloads such as machine learning/AI and some professional 3D and 2D rendering applications.

A solution like what they are proposing in the article you posted is definitely intriguing and could be part of a great shift in future technology, however it all depends on how long it takes it to get to market and what other more standard creators of volatile memory have accomplished in the meantime, because they certainly aren't standing still either.  In fact, one of the big benefits of HBM2E is that it requires less power, produces less heat, and is much cheaper to produce, with overall higher yields than previous versions of HBM, so the current generation of GPUs may be the last cards we see anything other than HBM on at least mid-range to high-end gaming and professional graphics cards.

It is definitely an interesting time in computing hardware.

[sman]

More info in "https://www.sciencedaily.com/releases/2019/06/190620100015.htm"

In the home, energy savings from efficient lighting and appliances have been completely wiped out by increased use of computers and gadgets, and by 2025 a 'tsunami of data' is expected to consume a fifth of global electricity.

But this new device would immediately reduce peak power consumption in data centres by a fifth.

It would also allow, for example, computers which do not need to boot up and could instantaneously and imperceptibly go into an energy-saving sleep mode -- even between key stokes.

The device is the realisation of the search for a "Universal Memory" which has preoccupied scientists and engineers for decades.

[exile360]

I'm not so sure about that.  Modern PC's, tablets and smart phones and other devices are pretty efficient these days.  Every time they go through a node shrink and/or architectural optimization or re-architecture they increase their overall efficiency so that they can do more work faster with less power and generating less heat.

For example, my little 15.6" laptop is by far the fastest, most powerful system I have ever used.  It makes all of may former massive desktops look like ancient calculators in comparison.  It's like comparing a modern super car to an old Ford Pinto.  It's just a completely different class of performance on a level I never dreamed of, and it is cooled by 2 tiny copper heatsinks and 2 mini laptop fans.  It's incredible.

That said, they could always do better, and efficiency becomes very important as form factors shrink and batteries get smaller as portability becomes increasingly important.  This is why ARM has been so successful of late with their extremely low power chips.  As for instant-on, we pretty much have that now with modern NVMe PCIe based SSD's.  It only takes seconds to boot my system from a full shutdown, and most of that time is spent going through the slow BIOS splash screen where the system is initializing the hardware before it starts loading data from the drive.  Once Windows actually starts to load it only takes maybe 5~7 seconds to get to the desktop, and only about 3~5 more seconds for all of my startups to finish loading and things to settle down to full idle (and I have quite a few startups as I'm quite excentric with the overclocking tools, system monitors, numerous security applications and sidebar gadgets so I could make it much faster if I stripped it down to just the bare essentials), and I don't even have the latest fastest drive they offer.  It's 1.5 generations behind at this point (there is the 970 Pro and 970 EVO Plus which are both faster than my 960 Pro).

Edited December 1, 2019 by exile360

[sman]

Nothing is ultimate.. How 100m has evolved over time, pushing human limits.. Technology will keep evolving and come up with surprises, not thought of.. let's welcome breakthrough's which can make life easier and more meaningful..

[exile360]

Yeah, that's for sure.  I mean I never expected that they would ever implement a storage interface based on PCI-Express, but now we have NVMe, and with the newest generation hardware supporting even faster standards like PCIe gen 4, things are only going to accelerate.  Over time technology will evolve and speed up as it always has, becoming ever more efficient as it does.  Years from now we'll look back at my statement about how fast my system is now and laugh at the thought that what I had was fast compared to the current tech of those years to come.  The only thing that never changes in life is change itself.  It always comes whether we're ready for it or not, it is inevitable.