Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Rest in Peace, Optane

specbranch.com

Post reply on HN
Print view View original

101–110 of 122 posts

Re: Rest in Peace, Optane

#101

Post by CobaltFire »

buildbot wrote: post #4

Optane SSDs are really special and still have a place. SLC is 30 microseconds at best, FIO puts my 905p optane drive at 5-11 microseconds. That means for a single threaded latency bound task, a 905p may finish it 6 times faster than an SLC ssd. 20 times faster than MLC! Edit - wrong prefix sorry!

I use an Optane drive for boot/pagefile and for my VR simulators (two drives) for exactly this reason. It doesn't help peak FPS, but it has a noticeable effect on the 1% and 0.1% cases, to the point where I notice it and can tune around the difference.

This is a 905p vs SK Hynix Platinum for reference.

Re: Rest in Peace, Optane

#102

Post by numpad0 »

eternityforest wrote: post #27

Why was everyone so obsessed with the idea of Optane as universal memory? From a consumer perspective, it was a fast small SSD that doesn't wear out. That was enough to justify it's value at the prices it went for(Or at least it seems like it, I never personally used it). It didn't need any radical rethink of computing to be worth it. I'm not even a fan of the nonvolatile RAM idea. Rebooting is often the first thing…

RAM in 2010s were more expensive - or rather, more capped than it is today. Consumer desktop platforms always had 4 Unbuffered DIMM slots, servers had multiples of four RDIMM slots, and typical UDIMM were up to 8GB per slot although there were some 16GB modules towards the end of DDR3 era. SSDs were less advanced but were already exceeding 128GB in size when Optane appeared.

So there was significant hurdle going beyond 16GB per machine, and sales pitch for Optane was DRAM substitute at near-NAND density and cost. Which, if taken at face value, implied >64GB RAM without committing to server/workstation platforms. That was appealing, at least to me back then.

Nowadays you can buy 4 sticks of 32GB DDR4 or DDR5 modules for ~$350, which makes a hypothetical 64GB Optane DDR3 module a rather moot point.

Re: Rest in Peace, Optane

#103

Post by afdsganionio »

ot wrote: post #82
mrb wrote: post #35

Earlier quoted context omitted.

" The theory that this market exists in any significant way never pans out in practice. " The transition from HDD to SSD technology is a counter example to your claim. It was a drop-in replacement (same SATA interface) and the tech significantly improved performance without any other software modification needed.

SSDs are orders of magnitude faster than HDD, enough to justify the cost differential and investments to rethink the software stack. Octane was an incremental improvement in performance when compared to SSD, and in cost when compared to RAM. An awkward spot to be in.

>Octane was an incremental improvement in performance when compared to SSD

Optane PM was byte-addressable and had latency of ~300 ns. It rendered the entire block storage abstraction obsolete. It was so radical that using it effectively would require throwing out the assumptions all OSes have made about IO for the past fifty years.

The fact that it offered an incremental improvement even when its unique capabilities were completely ignored shows how phenomenally capable the technology was.

Re: Rest in Peace, Optane

#104

Post by theandrewbailey »

avidphantasm wrote: post #77
mike_hock wrote: post #62

Earlier quoted context omitted.

Like Itanium, it was one of those Intel projects that was simultaneously too ambitious and not ambitious enough. If Intel really wanted to redesign the Von Neumann architecture, they would have had to be prepared to absorb losses for much longer, way north of a decade. The alternative might have been to focus exclusively on providing SSDs using the new technology and maybe try to segue into this new memory architectu…

I think it will be basically impossible to move away from von Neumann unless you control the entire stack, including OS and software. I don’t even think Apple could do it with Mac, because they are general purpose machines (for now at least). Maybe Apple could do it with iOS devices. Nintendo might be able to pull it off, though people trying port titles from other platforms may no longer try to do so. Because so man…

I'm not sure if Sony would want to make a system with a very unique architecture again. Devs complained about how hard it was to program PS2 games, and again with PS3. PS4 and 5 are practically PCs by comparison.

Re: Rest in Peace, Optane

#105

Post by acdha »

mike_hock wrote: post #62
hardware2win wrote: post #6

What was interesting about Optane was that it was kinda an attempt to get rid of disk and ram distinction. You had single device (optane stick) which served as both - ram and disk. After decades of current approach we finally could physically get rid of disks - hdds, ssds or even nvme sticks. https://www.theregister.com/2022/08/01/optane_intel_cancella... >Optane presented a radical, transformative technology but bec…

Like Itanium, it was one of those Intel projects that was simultaneously too ambitious and not ambitious enough. If Intel really wanted to redesign the Von Neumann architecture, they would have had to be prepared to absorb losses for much longer, way north of a decade. The alternative might have been to focus exclusively on providing SSDs using the new technology and maybe try to segue into this new memory architectu…

The other similarity was that they needed to treat developers like VIPs: Itanium failed in part because almost nobody was interested in paying a premium for a slow chip, licensed compiler, and then spend their time optimizing code to match competing chip’s out of the box performance. In both cases, they really needed to flood developers with free hardware & help - especially open source developers working on things like databases where you could see the most wins.

People might have bought Optane if the pitch was “Postgres/MySQL runs twice as fast” rather than hoping someone else would make your purchase cost-effective later.

Re: Rest in Peace, Optane

#106

Post by busterarm »

Ugh. Good riddance. As someone who buys millions in servers from Dell year over year, I cannot begin to describe how much for ~6 years Optane was pushed and pushed and pushed on us and it was all they wanted to talk about every other month and we had absolutely _zero_ desire for it.

I was so sick of empty suit sales reps looking at us funny saying "why for you not want optane!?" every other week (read that like an Idiocracy character, please).

I don't give a shit how much Intel is incentivizing you to sell it to me. I. Don't. Want. It.

Re: Rest in Peace, Optane

#107

Post by pbalcer »

adastra22 wrote: post #99
RyanHamilton wrote: post #91

Earlier quoted context omitted.

No you don't have to fsync. Think of it like RAM. You don't fsync RAM.

You do, in fact. It’s called a memory write barrier. Ensures consistency of data structures as needed. And it call stall the cpu pipeline, so there’s a nontrivial cost involved.

The point is that on PMem that is simply "sfence", and not a potentially super-expensive "fsync" syscall... Fsync is an fsync, not a memory barrier...

Re: Rest in Peace, Optane

#108

Post by pbalcer »

choppaface wrote: post #9
buildbot wrote: post #4

Optane SSDs are really special and still have a place. SLC is 30 microseconds at best, FIO puts my 905p optane drive at 5-11 microseconds. That means for a single threaded latency bound task, a 905p may finish it 6 times faster than an SLC ssd. 20 times faster than MLC! Edit - wrong prefix sorry!

Maybe your fio is hitting RAM? fio for 905p / p5800x is usually low micros. Order of magnitude faster than other nvme but not as fast as RAM.

it's possible to get <1us with Optane PMem (the dimms) in fio, you just need to use the memory-specific engines so that it avoids the fs/block layer.

Re: Rest in Peace, Optane

#109

Post by pbalcer »

rcxdude wrote: post #70
tanelpoder wrote: post #52

Earlier quoted context omitted.

As it's part of the memory domain, not some external block device, the CPU caches can cache it (64B cache line sizes) and maintain cache coherency across CPUs, NUMA nodes (and soon with CXL 3.x ?) across memory pools shared by many servers. So your programming model will be just using CPU loads & stores to access the individual bytes you need directly from the Optane storage, instead of having to do some 512B or 4kB…

There's not much here that requires optane as the underlying storage tech, though. CPUs can already treat PCIe devices as memory, and do the same caching work underneath it. Most modern SSDs are alreadying using PCIe, it's just the protocol on top of it doesn't support mapping the contents of the disk to memory like this, but there's no real reason they couldn't.

The reason is physics. NAND just isn't fast enough (and requires firmware wear-leveling, which can mess up latency) for this to be practical.

Re: Rest in Peace, Optane

#110

Post by bobince »

waxali9815 wrote: post #66

I'd like to offer my experience with the consumer-targeted Optane devices. I had a laptop that I replaced about 2 or 3 years ago, that had a hybrid 512GB HDD + 32GB Optane storage. It was a massive headache for me. One fine day my laptop ran into the common Windows issue of 100% disk utilization. I tried all the common fixes to no avail, and at some point I remembered my disk had some funky new tech called Optane. I…

Ugh, a friend bought some similar device due to a misleading description of what the HD was. He'd tried to change the partitioning without knowing it was there; turns out if you had anything but a single Windows partition (with custom drivers injected in the right undocumented way at install-time) you ended up with a sans-optane uncached 5400rpm HDD, which even at that time was completely unfeasible as a system disc.

Absolutely horrible product. And not even cheap before having to turf it out in favour of a proper SSD.

Post reply on HN