I could see the benefit on a non-phone mobile device. Completely cutting power as a deep sleep without needing a lengthy boot sequence could be nice.
Unless it’s actually “just as fast” as volatile memory (including progress to the latter) and not more expensive, though, it seems like it wouldn’t justify the tradeoffs.
This is where micron and Intel tried the phase change memory, optane, they could never make it cheap and fast enough to direct replace.
pRAM can actually be faster than DRAM in terms of latency. Main problems are cost, density and power consumption (to varying degrees, depending on the concrete technology)
Shit I remember reading about MRAM at least a decade ago. I thought it was going to be the next big thing to advance computers, but it never showed up outside some rare industrial use cases.
The great thing they could do with that is have RAM that keeps its data while powered off. You could possibly turn your PC on and off like a lightswitch, instead of booting up over a number of seconds/minutes. But now that we have NVME SSDs and stuff, we are already getting close to that
So could it be plausible one day to run a RAM-less computer? If persistent memory speed matched DRAM speed, would there still be a benefit to distinguishing RAM from SSD, beyond cost?
MRAM, FeRAM, and ReRAM all have a limited number of write cycles. It’s certainly possible that the write endurance will be high enough that it doesn’t matter by the time it can replace DRAM in a computer though.
Density would be a factor as well for both cost and device dimension/weight.
Cache is way faster than RAM, but it takes up too much die space and power to be the only volatile memory.
The ultimate goal of persistent RAM endeavours is to build all-RAM computers. You have all your storage and RAM in one. Would eliminate most loading and boot times, if just the density was high enough (actually is for some small, embedded devices)
That’s just semantics though. That’s exactly what I meant. No separation of RAM and storage.
I called it RAM-less, rather than all-RAM, because we already have the concept of virtual RAM on storage. So you could have your files in a file structure and your volatile memory in a virtual RAM file. And you wouldn’t necessarily even need to load files or programs into virtual RAM if you were only reading them, you could have a strict file pointer.
Wasn’t SOTRAM the hottest thing around, or is that idea dead again?
This is the best summary I could come up with:
Persistent memories can or will soon match DRAM in terms of speed, which could see it eventually replaced in many applications if one of these technologies can scale up and bring the costs down.
One benefit of replacing DRAM with persistent memory is obvious; it keeps its content even without power, meaning there is less danger of losing data.
However, DRAM has been around for a long time and is cheap to produce and available in high densities – hurdles that persistent memories will also need to surmount.
The webinar pointed out that Optane, which Intel dropped in 2022, proved how difficult it is for rival memory technologies to compete with DRAM (and NAND flash) on price.
Techniques for wear leveling have been developed and incorporated into the software that manages SSDs to handle this, such that the experts contend that “any emerging memory can get by with an equally low endurance as long as it’s put behind the right controller.”
But the SNIA experts see CXL, or compute express link, as the future interface of choice going forward as it can present any type of memory to the host processor.
The original article contains 801 words, the summary contains 191 words. Saved 76%. I’m a bot and I’m open source!
