Intel SSDs Based On 3D XPoint Technology Next Year

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
Aside from all the talk of personalization, BMX stunts and spiderbots, one of the biggest announcements to come out of the Intel Developer Forum keynote today was news that Intel's Optane SSDs, based on 3D XPoint memory, will ship in 2016. If Intel's performance claims are true, things could get pretty damn interesting once Optane solid state drives ship next year. Your thoughts?

For those of you that missed it, you can see our recent 3D XPoint write-up here.
 
Last edited:
So all of our (flash based) SSDs will be obsolete and (because we are all [H] tech freaks) need to be replaced long before anyone could have any SSD fail due to limited write/erase cycles.

Pretty funny.
 
This tech comes next year but it will take a good while before it makes it into consumer hardware. Micron, Samsung etc have put a lot of resources into existing memory for SSDs and that will play out before we all jump on the new fancy.

Save for workstations where every performance increase is welcome, it's hard to see SSD of current tech yielding quickly. It isn't the same as with hard drives -> SSD transition where HDDs are painfully slower in access time, and that transition took years.

So all in all very cool of course but SSD wise, I see little immediate movement.
 
Steve said:
I believe I will Optane one of these when they come out. ;)
Click to expand...

93RDJks.jpg
 
dr.kevin said:
does this mean instant game loading?
Click to expand...

Unfortunately most games see little to no benefit in load times from the slowest ssd to the fastest. Even ram drives make very little difference on most games. This is an older article on it but I believe it's still relevant http://www.tomshardware.com/reviews/ssd-gaming-performance,2991.html

It's why for desktop loads something like the intel 750 SSD doesn't excite me at all, hardly anything hits very high queue depths where that drive actually delivers its performance.
 
Am I the only one who saw the IOPs numbers and instantly thought "database iops mmmmmmmmmmmmm" like Homer?
 
Elf_Boy said:
So all of our (flash based) SSDs will be obsolete and (because we are all [H] tech freaks) need to be replaced long before anyone could have any SSD fail due to limited write/erase cycles.

Pretty funny.
Click to expand...

Doubtful, this will surely carry a premium. Small SSD might be obsolete but if they're <256GB they're either a few year old by now or were relatively cheap, larger ones remain useful for storage while 3D XPaper will slot in for system drives.

Pricing and interface is the thing IMO.

PCI-E/M.2 NAND based SSD might be obsolete if this shares a common interface (PCI-E is confirmed, hopefully they do 2.5"/U.2 too for those without slots to spare tho) AND doesn't carry much more of a premium than those drives already do over SATA ones.

Lots of ifs in there for tech that's still a year away.
 
lollerwaffle said:
This tech comes next year but it will take a good while before it makes it into consumer hardware. Micron, Samsung etc have put a lot of resources into existing memory for SSDs and that will play out before we all jump on the new fancy.

Save for workstations where every performance increase is welcome, it's hard to see SSD of current tech yielding quickly. It isn't the same as with hard drives -> SSD transition where HDDs are painfully slower in access time, and that transition took years.

So all in all very cool of course but SSD wise, I see little immediate movement.
Click to expand...

That's what I would've thought but they're talking about how Ultrabooks will see it next year...

Frankly I don't buy it, if it's as fantastic as they claim why wouldn't they charge a premium? (limiting it's adoption in price sensitive laptops)

They're certainly not gonna cannibalize their own joint NAND venture either and they've said they're going ahead with 3D NAND roll out etc.

Intel already charges a 100%+ premium for their pro-sumer PCI-E SSD (not saying it's not justified), hard to see this coming in at the same $1/GB.
 
I am so [H]ard for this technology.

And if they think they can ramp up density to get price parity with NAND?

This is a tremendous breakthrough.
 
Steve said:
I believe I will Optane one of these when they come out. ;)
Click to expand...

i agree. I will sell my right nut. Not my left but the precise right one!

bleomycin said:
Unfortunately most games see little to no benefit in load times from the slowest ssd to the fastest. Even ram drives make very little difference on most games. This is an older article on it but I believe it's still relevant http://www.tomshardware.com/reviews/ssd-gaming-performance,2991.html

It's why for desktop loads something like the intel 750 SSD doesn't excite me at all, hardly anything hits very high queue depths where that drive actually delivers its performance.
Click to expand...

thats why this is awesome. The 750 was like 2x better in low que depths than regular SSDs and this thing beats it by 7x!
 
Crucial's flavor should have a good price point and unless Samsung comes out with something similar by then this could be a huge win for Micron in the market.
It would be nice to see Crucial as the go to SSD.
 
Impulse said:
That's what I would've thought but they're talking about how Ultrabooks will see it next year...

Frankly I don't buy it, if it's as fantastic as they claim why wouldn't they charge a premium? (limiting it's adoption in price sensitive laptops)

They're certainly not gonna cannibalize their own joint NAND venture either and they've said they're going ahead with 3D NAND roll out etc.

Intel already charges a 100%+ premium for their pro-sumer PCI-E SSD (not saying it's not justified), hard to see this coming in at the same $1/GB.
Click to expand...

Intel needs to drive people to buy new intel technology somehow. The same performance with less power consumption is nice, but not a driving factor. Upping the performance of storage is nice, but not great.

Ultimately the technology has two big use cases: high-performance workloads (i.e. servers), and replacing both RAM and NAND in a single device (i.e. low-cost mobile, NUCs, chromebooks, compute sticks, streaming devices) where you don't have huge performance needs on the RAM side and you can replace two chips for the price (and circuit board space) of one.

3D NAND is still useful for larger dedicated storage. It's a mature, tested technology. Now that Optane has two brand names though, it's going to be way bigger in the mainstream sector.
 
Just need one 8TB model for no more than $400 and I'm set for the next 10-15 years, until I need a second one.
 
toooskies said:
Intel needs to drive people to buy new intel technology somehow. The same performance with less power consumption is nice, but not a driving factor. Upping the performance of storage is nice, but not great.

Ultimately the technology has two big use cases: high-performance workloads (i.e. servers), and replacing both RAM and NAND in a single device (i.e. low-cost mobile, NUCs, chromebooks, compute sticks, streaming devices) where you don't have huge performance needs on the RAM side and you can replace two chips for the price (and circuit board space) of one.

3D NAND is still useful for larger dedicated storage. It's a mature, tested technology. Now that Optane has two brand names though, it's going to be way bigger in the mainstream sector.
Click to expand...

replace RAM....simply no. That is a terrible idea on so many levels.

1. too slow
2. to low of endurance

just to give you 2 small reasons why that is terrible
 
They need to up the SSD storage capacity at an affordable price.

1TB SSD (not a hybrid) and I'm sold
 
SomeGuy133 said:
replace RAM....simply no. That is a terrible idea on so many levels.

1. too slow
2. to low of endurance

just to give you 2 small reasons why that is terrible
Click to expand...

But you are thinking of direct replacement. Why not have a memory module with an SSD stacked on top of it using the same/similar pathways. The CPU/memory controller could address it, so you could have fast RAM and slower non-volatile storage in the same memory pathway, removing it from the bridge chip access. Or different slots but still attached to the memory controller but not replace RAM, just augment it.
 
ob1 said:
But you are thinking of direct replacement. Why not have a memory module with an SSD stacked on top of it using the same/similar pathways. The CPU/memory controller could address it, so you could have fast RAM and slower non-volatile storage in the same memory pathway, removing it from the bridge chip access. Or different slots but still attached to the memory controller but not replace RAM, just augment it.
Click to expand...

in his example he was referring to a complete replacement
^
|
and replacing both RAM and NAND in a single device
Click to expand...


What you said is a key point and use for NAND and XPoint. Sandisk has a product just like this but embargoed due to legal reasons. Not sure if the embargo dropped though
 
You must log in or register to reply here.
Back
Top