Intel’s Optane memory modules launched with a lot of fanfare in 2015, and were recently discontinued, in 2022, with similar fanfare. It was a sad day for me, a lover of abstraction-breaking technologies, but it was forseeable and understandable.
At the time of Optane’s launch, a lot of us were excited about the idea of having a new storage tier, sitting between DRAM and flash. It was announced as having DRAM endurance and speed with the persistence and size of flash. It was a futuristic memory technology, but the technology of the future met the full force of Wright’s Law.
Nima Badizadegan
I definitely remember Optane being presented as a huge deal, but it seems it fell between the cracks of other technologies developing around it and their prices dropping fast.
Despite still outperforming even current generation drives for latency… it just couldn’t compete on cost at the middle and low end. A segment you have to sell in in order to sell enough volume to get cost down to make profits on the HPC stuff.
Also many of their lower end devices kind of sucked. The they wasted production capacity on the 16 and 32GB sticks that largely have ended up in large lot sales on ebay. Hybrid drive solutions generally suck and should be avoided should be a rule of thumb by now (unless someone develops one that is provably good).
Apparently there was also wasted production capacity on Optane RAM modules… which never gained any sort of wide adoption. They probably should have narrowed their focus to only Optane DIMM production, and designed a PCIe adapter to create the other products… and ignored the U.2 and M2. drives were flash could dominate anyway. This would have allowed them to sell into the PC market at the enthusiast and heavy duty crerator segments, while Optane DIMMS could sell for $ to the HPC market. And the components would be naturally segmented because Desktop PC Memory controllers dont’ support Optane.
That’s a decent description. Optane has amazing properties but Intel couldn’t find a niche to make the Optane factory profitable. SSDs made enough profit that manufacturers could keep investing in ‘Next Gen’ speeds and technology. Not so with Optane.
I’ve worked in pc service at those times and never build pc with that. Sad nie, never saw how it worked
Marshal Jim Raynor,
I think the idea behind optane had merit. Using a flash cache in front of much cheaper/larger/slower storage is extremely beneficial and would be like bringing “flashcache” and “bcache” to the masses. Making this technology accessible is great, but intel were too greedy and optane.vendor locking and implementation restrictions made it both expensive and cumbersome. IMHO this would have fared a lot better as a ubiquitous open standard.
The problem with Optane was that it promised to be a non-volatile RAM, but ended up having speeds just above an SSD (at a much higher cost).
Raw /throughput/ wasnt massively higher. /latency/ was absurdly better than NAND and still is.
Lots of other pros to it. It isnt just about speed. Its durability. Performance across lots of scenarios. Stability of performance. Etc etc.
It is a much more elegant tech vs current NAND offerings. NAND is just “good enough” and cheap.
helf,
That’s a good point about latency. It’s trivial to scale up throughput by adding more chips. Even spinning disks can support arbitrary throughput. But low latency requires the underlying technology to actually be fast.
IMHO optane-like caches and multi-tier solutions are genuinely useful at speeding up commonly accessed resources backed by giant disks but it was never affordable enough to make market inroads with consumers.
I love optane. I bulk bought a ton of 58 and 118gb P1600X drives for cheap. I use them as caching drives everywhere. I dont have to worry about wear. Latency is absurdly low. And r/w is more than adequate.
I have a P4800X I use in my main workstation. It is fantastic. <3
The TDW of modern NAND is a sad joke in comparison. Anything approaching optane level wearing is insanely over provisioned. "But it's fast!"…
I also run a highpoint nvme raid card with 8x118GB in a stripe. *chef's kiss*
helf,
Indeed. TDW has been going downhill, which is to be expected when we shove more bits into shrinking cells. Granted some consumers may not require that much writing, but when it comes to TLC & QLC some of the write cycles are low enough to cause concern. The stress testing & benchmarking that I used to perform gratuitously on new components now give me pause with SSD since it uses up drive lifecycle. The news gets worse for computers that have SSD soldered in place. Now a dead SSD means the entire computer is unusable and non-user-serviceable, often without warning to the user.
“These MacBooks DIED from SSD Failure & How To Prevent It (Not for RICH dudes!!) ”
https://www.youtube.com/watch?v=MZuv4TIjk-I
I’m really disgusted by this practice, it’s a prime example of technology becoming worse, t’s bad for consumers, it’s bad for the environment, but we end up here because planned obsolescence and shorter lifecycle promotes more sales. So although it is wasteful, manufactures do it because it’s more profitable for them when products don’t last as long.
I guess I better name now is Unobtainium