The unification of memories is a kind of “Holy Grail” in computing. Despite the available advances, we are still forced to separate memory into two categories: Volatile and fast like RAM, or persistent and slow like storage on SSDs and hard drives. Any project that manages to combine the best of both worlds at a reasonable price will spark a true revolution in performance, and one of the most promising is UltraRAM, developed by a group of scientists at Lancaster University.
Why do we need something like this?
Why do we have to load all programs over and over again when restarting a computer? Why is our work lost every time there is a power cut? The key word is volatility. RAM is a fundamental resource in our computers, but all the information it stores is temporary. In other words, RAM lacks persistence, and when the power is interrupted… *poof*.
This weakness is “solved” with storage memory: Hard drives, SSDs, pen drives and SD cards, magnetic tape, even optical discs. These media are stable and persistent, but they pay the price with much lower speed. For years, the industry has tried to develop a great wildcard, a universal memory that can merge the extreme speed and access times of RAM with the persistence of classic storage. The latest attempt to achieve this is UltraRAM.
UltraRAM, or the search for “persistent RAM”
The idea of UltraRAM is not entirely new, and the first data dates back to July 2019, when it was presented as a solution for IoT devices. But the big news is its implementation in silicon, which could lead to its mass production. The great promise of UltraRAM is to merge the speed, energy efficiency and durability of RAM with a persistence of at least a thousand years for data storage.
How exactly does it achieve this? The study published by a group of scientists at Lancaster University is not exactly “light” so to speak, but we can extract details such as quantum wells of indium arsenide, combined with aluminum antimonide barriers. The device design is complex, and personally I admit I have doubts about how simple or economical its manufacture will be, but at the same time it is not less than an initial version, and nothing prevents us from imagining optimized variants in the future... as long as it leaves the laboratory.
Access the study: Click here
