XPoint /Optane

>>58200142
Well considering that the entire 200-series chipsets are literally 100-series chipsets with optane support, I would fucking hope that optane comes along during the Kaby Lake cycle

>>58200142
CES next week we'll know. But honestly I am more interested in Samsung's Z-SSD tech.

>>58200499
>Z-SSD
That's just regular a SSD.

>>58200142
A fast but pricy line of SSDs and the second coming of rambus. Daily reminder that 3d xpoint RAM will require special xeon chips.

>>58200592
It's called optane now and it will be on Z270 motherboards with LGA 1151

>>58200621
Source? I find that highly questionable because the memory controller is on the chip, meaning that if it's chipset specific then it's just more intel jewery.

>>58200142
While I'm excited I don't really see a point. I already have a raid0 ssd as well as a ramdisk and I can't notice a difference between the two when I put programs in them.

>>58200824
It's probably a proprietary form factor of some kind
I mean even pcie x4 can't match ddr2 bandwidth, which is what optane is supposed to do
Look up the Z270 specs yourself, exact same as Z170 but with optane support

Remember sata express?
Yeah, me neither.

>>58201209
>optane support
And what, exactly, does that mean? Again, I seriously doubt that they're releasing it in ram form yet. Even if they did release it, it'll still be a failure. They'll have made cheaper ram with higher latency. Woohoo!

>>58201343
>not buying a TB of ram for 2TB SSD prices

>>58201343
>implying 100% non-volatile ramdisk/diskram isn't useful

If this the track of SSD speeds nowadays, how long until ram itself is obselete?

>>58201446
You've completely misunderstood, it's half the price of ram, and six times more expensive than an SSD.
>>58201459
Yes, let's all buy slow ram because it would let the OS hibernate without taking the second or so it takes to dump things onto an SSD.

>>58200142
Why are they asking there customers what they're own product is?

Shouldn't it be the other way round?

>>58201210
Why do they keep putting more than one SATA Express ports on motherboards, or any at all? I've never used one, but there's been plenty of times where I've needed more SATA ports.

I have read somewhere that the mobo for zen will support it. We should have more news about it in 2017.

>>58201681
>zen will support it.
It'll support the SSD version, meaning SATA, PCIe or nvme. There is no way in he'll they're supporting the ram version, that means changing the memory controller. It wouldn't surprise anyone if Intel are making the ram version Intel exclusive anyway, because that's just how they roll.

>>58201656
hahahahaahha

>>58200562
That difference will be one is expensive and hard to get(Optane) the other is not really expensive and performs almost the same as the former.

>>58201793
The RAM version won't be much if not of any use for the avarage/geyman user.

I believe that we're in for somewhat of a disappointment regarding it.
NVMe drives have gotten so damn fast, that I doubt they're going to be showing a product with 5x faster speeds or anything.
We're probably going to get something that's twice as fast at best and 10x more expensive.

I'm more interested in NRAM
http://www.tomshardware.com/news/fujitsu-carbon-nanotube-memory-nram,32603.html
This is going to be aimed for a totally different price range, yet supposedly another ultra fast solution.

>>58200142

You're in luck.

http://www.anandtech.com/show/10932/kaby-lake-systems-with-intel-optane-ssds-coming-soon

Seems like low capacities only right now and only used for caching.

So essentially, they will be shipping in a relatively useless state.

>>58204378

>ThinkPads will offer 16GB Optane SSDs in M.2 2242 form factor
>16GB
Ouch.
You might as well buy the cheapest 16GB of RAM you can find and use that as a RAMdisk to get faster and most likely cheaper storage.

I get a vibe that Optane is anything but in a ready state to enter the market.
It's like they've been delaying for so long, that they just decided to throw these insignificant sized drives on the market to show that they're actually working with it..
Not to mention they have to be bringing some god tier performance to the market, in order to give people reason to choose vastly more expensve Optane over Samsung's NVMe drives.

>>58204541

The problem is twofold. On top of what you just mentioned, the jump to X-Point from NAND is not the same as magnetic spinning disks to NAND.

I'm still hopeful someday there is a universal memory that can scale from processor cache to hard drive and the only factor influencing latency is distance from processor. But the thing is there are only two realistic candidates right now that can do that: MRAM and ReRAM. The latter is slower and much more experimental while MRAM has been in research for a long time.

Really going to take a while, sadly, before we can have our cake and eat it too.

>>58201459
Got a question, weren't there 5.25 converters that you could load ram into, and it would work as a ram disk?

do they still make these things, and do they have a battery option to make it non volatile?

http://www.anandtech.com/show/10932/kaby-lake-systems-with-intel-optane-ssds-coming-soon

AYYMD IS FINISHED & BANKRUPT

>>58200142
It's only worth it if your dataset can't fit into RAM but even the heaviest consumer workloads won't use more than 32GB RAM. If you need more RAM then just buy more RAM.

>>58201793
They did change the memory controller from skylake and up, dumbfuck