Intel announces new memory, XPoint

A day of XPoint news; Micron says it will initiallly cost 4-5x more than NAND, here at Myce.

Robert17 said:

A day of XPoint news; Micron says it will initiallly cost 4-5x more than NAND, here at Myce.

Click to expand...

If it is really 1000x faster than typical SSD NAND, 4-5x the cost is pocket change for a single drive that can saturate PCI-E x16 3.0 on its own.

petteyg359 said:

If it is really 1000x faster than typical SSD NAND, 4-5x the cost is pocket change for a single drive that can saturate PCI-E x16 3.0 on its own.

Click to expand...

Hear, hear!

Hear here!

As referenced in the article, the 1000x figure is a measurement at the cell level, pre-controller, pre-encryption, etc. The downstream figure I recently read about is about 10x faster than current NVMe SSDs, so still in the ballpark depending on your needs & wallet. Obviously, enterprises will get to play first.

THG has a photo (not very revealing) of Intel's Optane drive installed in a system, plus a benchmark comparing an Intel 750 SSD to Optane. Says client products before the end of the year. Looks like one holdup is firmware at the moment but mostly it's coming upon 'Good-to-Go Day' according to the article.

Robert17 said:

THG has a photo (not very revealing) of Intel's Optane drive installed in a system, plus a benchmark comparing an Intel 750 SSD to Optane. Says client products before the end of the year. Looks like one holdup is firmware at the moment but mostly it's coming upon 'Good-to-Go Day' according to the article.

Click to expand...

PCI-E x4 is somewhat of a bottleneck if it even approaches 10x faster than a regular NAND drive...

Just a bit of new info on 1st article PCM products. A full read of the article is appropriate but basically this isn't nearly what Intel has been teasing in
their pre-marketing efforts:

"The consumer products will initially sport capacities of either 16GB or 32GB, leveraging the NVMe protocol at PCIe Gen 3.0 x2 bandwidth in the M.2 form-factor. Mirroring NAND technology, the greater capacity solution will sport the highest performance: with the 16GB part coming in at 1400 MB/s read and 300 MB/s write speeds, against the 32 GB's 1600 MB/s and 400 MB/s, respectively. We see similar results in regards to IOPS, with the 16GB solution offering up to 285,000 read and 70,000 write operations per second, against the 32 GB's solution respective 300,000 read and 120,000 write. As usual with new technologies, expect all these metrics to only go up in time."


Source

Saw this at Tom's today, highlight is that it's near & fast if the 'leak' has veracity.

It appears that in the next couple of months, just in time for Christmas, fully fledged Optane SSDs are coming up on availability. Pricey, but based on the low latency, endurance, and warranty, they may be worth a look if the need is for speed. Yes, the bottlenecks will now be with software. Good read at PCPer.

the newegg listing says they will release on the 30th?
https://www.newegg.com/Product/Product.aspx?Item=N82E16820167438&Tpk=N82E16820167438

MAN i want one but sheesh 4 bills for 280gb is pretty stiff. how long will it take for prices and competition to level off? im thinking this will be a xmas 2018 present for me unless prices get mucho better before then. but but i want one

What competition? If you want the performance there isn't really anything else like it apart from the enterprise version. Haven't kept up to date if anyone else is doing a similar technology but even if they are, we still need to see product.

Micron shared in the development of this tech, so they likely have some products in the pipeline. I'll bet a ham sammich that they and Intel will have similar pricing once/if Micron's version of Optane hits the streets.

Optane is a Micron/Intel effort. They have been collaborating for years in the R&D to get this memory to work. They must have made a deal that Intel would be first shipper than Micron soon after. Outside of Optane, Xpoint is just not ready for mass production, and probably won't be when GEN Z comes around. Optane and Xpoint memory will most likely disappear within a year.

Dolk said:

Optane is a Micron/Intel effort. They have been collaborating for years in the R&D to get this memory to work. They must have made a deal that Intel would be first shipper than Micron soon after. Outside of Optane, Xpoint is just not ready for mass production, and probably won't be when GEN Z comes around. Optane and Xpoint memory will most likely disappear within a year.

Click to expand...

What's Gen Z? Not heard of that one before, and search comes back with post-millennial talk...

The single place I've found taking pre-orders in UK is asking a more sane amount for it now than they were last week, but still not insignificant.

Gen-Z is a new bus architecture that is very similar to PCI-E and can be used on PCI-E with the correct protocal setup. However, its true intent is to bring true modularity to the x86 architecture. It also allows for non-Von Neumann architectures to be brought to the table.

So basically think of all the parts of your computer having access to everything. Your GPU has full access to the memory, and back plane I/O (USB/SATA/etc). Non-Von Neumann architecture would allow for operations to be done within the memory rather than just storing the memory and waiting to be used for operations. Once you start to decentralize the CPU and allow each part to operate on its own, you can start to create a higher degree computing architecture. That being, one that does not fully rely on all the parts being available in a system, but exploits those parts to create a higher advantage. I'm not saying you will only need a GPU for gaming and will not need anything else, but it does mean that a console type PC could easily compete with today's highest end systems thanks to the higher degree of flexibility the architecture provides.

GEN-Z is currently in development and you'll probably see it in servers within a few generations (most likely). I do not know about client side. Non-Von Neumann type architecture has been around forever but its never really seen the light of day.

For more info:

http://genzconsortium.org/
https://www.anandtech.com/show/10751/gen-z-consortium-formed-developing-a-new-memory-interconnect

https://en.wikipedia.org/wiki/Von_Neumann_architecture

I think I get your argument, and computer architecture in this sense isn't something I've looked at in a long time. Even at consumer level, there are obvious bottlenecks that could be improved on, but anything significantly different from what we have now but faster is going to take time for software to catch up. So on that note, I'm not going to hold my breath for this. It is far enough out it isn't worth thinking about from a user perspective.

I think Xpoint will gain more traction - why would it disappear if there is nothing else to take its place? It wont replace flash SSDs simply due to cost, but will remain there for the high end if wanted. I was kinda wondering based on the earlier post if there were technologies that would render Xpoint obsolete, but I guess not yet.

Well Xpoint is just a huge risk on the enterprise side. Outside of Optane SSD, Optane Memory is an Intel only standard, and is not compatible with AMD/ARM. This is a market risk for now, as AMD may take a big chunk from Intel in the server market in the next year or two. They offer more PCI-E per CPU and more cores per a CPU than Intel. So a 1-Socket AMD ~= 2-Socket Intel. Imagine doing an upgrade and you can half your memory and CPU cost but get the same memory, storage, PCI-E density. It should be a healthy competition.

I was looking at this from a consumer point of view. Also, to clarify if I say Xpoint I mean the underlying technology, not the use it is put to. Optane seems to be the branding for the SSD-like usage. I don't know if there is a commonly used name for the ram-substitute usage case. For my interests, I don't care about ram quantity only ram speed, so the use case of interest here is as a high end SSD. The high pricing will slow adoption, which I assume will be in place until they can ramp up production. Might even be better if they don't get much traction in the cheap ram substitute area so more of it goes to the SSD use case, and bring the price down there.