Chinese company with $24 billion funding claims its will bring DDR4 speeds to NAND

Y though? And why would we need DRAM then?

 

Better wait to actually see results from third party evaluators than believe a marketing speech from a new manufacturer. In fact even old manufacturers shouldn't be trusted before results are verified in practice. Intel's Lolptane didn't transform the world either. Intel even had to blackmail mobo manufacturers to make the merchandise move at all.

 

Intel made a lot of mistakes with Optane but the underlying technology is pretty awesome.

The price per GB, obscure form factors, low capacity and hoops you had to jump through all got in the way of Optane's success.

Had Intel created an all in one SSD with Optane cache and software built in they would have had a killer product.

Intel does not even support the best Optane setups officially. That stupid 58GB 800P SSD is actually 100% compatible with Optane cache and is far superior to their 16/32GB offerings but Intel does not support this use case. This is baffling as the 800P SSDs are useless as they are designed (OS + apps drive) but as cache for a SATA drive they let you create a compelling solution. A 2TB 860 EVO + 58GB Optane cache is both faster and cheaper than a 2TB 970 EVO but Intel wont acknowledge that this works and if you ask they will tell you flat out that the 800P drives are stand alone devices, not cache drives.

If this new Xtacking technology can be implemented in a far more plug and play way and at a price point that people can justify it could actually change the industry.

 

I would L-O-V-E to see SSDs getting the speeds of RAM, really. And not only because of MB/s but in terms of >>latency<< as well. SSDs tend to have a latency in the ms-range (milliseconds), while RAM is in the ns-range (nanosecond).

If you are able to drive 340 miles per hour your car is fast without doubt. But when you need about 15 minutes to change from one gear to another, you won't win a race, do you agree?

 

While I agree with everything you said, remember that what makes RAM (and optane) so superior to SSDs currently is not simply raw transfer rates, it is also the latency to fetch a specific piece of data.

Let's hope this Xtacking technology will be more than just huge sequential speeds. A persistent memory for all usage would be game changing

 

Not only that but if you have something like a RAID card attached to a bunch of huge HDDs having persistent memory would allow you have a huge cache without the need for battery backup on the cache. I am kind of surprised that we don't already see Intel RAID cards with Optane cache. Imagine a 16X RAID card with all the typical bells and whistles but instead of 8GB of RAM and a BBU you have 380GB of Optane cache with no need for a BBU. I am sure they could make those Optane DIMMs pop right into a RAID card making the cards upgradable.

 

Keep in mind, pretty much the reason why RAM was used since the dawn of computers is because disk drives have been too slow. If storage is fast enough to keep up with or even exceed the speed of RAM, imagine how much better workflows will be. You could effectively remove the process of loading data; the CPU can basically just read data directly from the disk drive. Not only does this save a lot of time in not having to wait for the data to be loaded into RAM, but the CPU would only read from the disk what it needs to read and nothing more, which would improve performance even further. This of course implies the software is built to take advantage of this; no existing software is built to just feed the CPU directly from the disk, at least not without loading in excess data.

Meanwhile, if something like HSA were to be implemented properly, there would be no need to load data over a PCIe bus, either - the GPU could read data directly from the disk and bypass the CPU entirely (when gathering resources - obviously it needs some data from the CPU to know what specifically to render).

The 16GB and 32GB Optanes were a joke and a real failure. However, the 900P is, in my opinion, the best thing Intel has made since Sandy Bridge. It is expensive, but it has the performance to back it up.

 

at what price ? and how many cycles can take before start failing ?will be the deciding factor even if they deliver even half what they are promising

 

They really should have released the 800P as the first Optane product and offered it as a stand alone SSD or Optane cache.

I agree about the 16/32GB drives, even RST supported 64GB. They also should have waited to release the Optane software as it is today instead of the 0 features POS it debuted as.

Optane today lets you pin folders and executables as well as shows you the breakdown of what Optane is doing.

The 900P were OK but the 905P is where the capacities really became something useful in more cases.

I have not seen it reported on anywhere but it also seems that Optane is now pulling an AMD and combining both RAM and Optane to offer up pretty crazy sequential read speed.

This is from my travel laptop and no, I do not not have RAM cache installed. This is the performance from activating the current build of Optane software on a 800P:



This is what Optane could be doing for people but Intel does not support the 800P as Optane cash and I have no idea why.

 

Im discounting them until they have an actual product in the market. I know i'm cynical. Getting new technology like this stood up and working in the real world is hard, and it's even harder to make a profit from it. I hope they are successful but suspect they won't be.

 

I remember the last time some chinese company made amazing claims like this, supposedly 100% screen-to-body ratio on a phone and then showed of yet another notched phone

this just sounds like they are raid-ing a bunch of nand to get x7 sequential speeds, at best, there's no mention of random perf or anything else

I would love to see this and so far I see sketchy brands on amazon selling TLC drives that are just barely twice the cost of decent HDDs so it seems they are producing something and it isnt half bad, but this would be quite the jump from that

 

Unfortunately the technology is either too expensive to mass produce or Intel isn't even interested in using it that much. If something is too expensive, it will stay as a niche market, thus not really playing a huge role in the bigger picture. You don't really see Optane drives competing with NAND drives in the SSD arena. If I had a single M.2 slot and had to choose between a 500GB NAND SSD and a 64GB Optane drive, the choice wouldn't be difficult, even if the Optane drive offered better random read and write.

 

I look at Optane as more of a tech preview product than anything else. People are buying them and doing some really cool stuff but I don't think anyone can argue that the 900P is worth the price premium over the 970. The 1TB 905P in a 'no budget' build makes sense and caching a 6TB+ HDD to a 800P might also make sense (this works but isn't supported) but these are not common scenarios.

There is even a guy that did an 8 way VROC 0 Optane setup.

https://rog.asus.com/forum/showthre...when-will-this-be-supported-in-the-BIOS/page8

 

It's not so black and white - Intel knows they have something good and people are willing to pay for it. You aren't their target demographic, and neither am I (just because I acknowledge Optane is good, that doesn't mean I'm willing to buy it).

 

Sure. I have seen the numbers and can admit it's good technology, as much as I dislike Intel as a company. However, my first post talked about changing the PC world. 3D Xpoint has remained so small market wise that it's merely serving the few who are willing to spend the considerable money. They are happy with it, but it doesn't mean anything to the vast majority relying on NAND SSDs with more reasonable prices. My initial post might have been too ambiguous in this sense, perhaps.

 

But that's how just about every single new technology works.... Just 10 years ago, a 32GB SATA2 drive cost $500. 5 years ago, SATA3 SSDs cost around $1-2 per GB for budget drives. Back in 1997, a DVD player cost $600 (which in today's money is around $950).
And like I said, you aren't their target demographic. Optane is currently focused on workstations and servers, where their prices are actually very reasonable. Look up the cost of enterprise SSDs and you'll find they're all crazy expensive. Remember - there is a lot of competition in the SSD market and before Optane, Intel hasn't really stood out much from the crowd. And yet, they crippled Optane by limiting it to specific chipsets. They wouldn't do this if they felt Optane wouldn't succeed.

It seems like you're trying pretty hard to hate on this. Keep in mind, I'm not a fan of Intel either; I usually avoid them whenever I can just out of principle. But despite my disdain for them, I can't help but admit that they did a good job with Optane, and its price is acceptable considering the performance and longevity of it.

 

Nah, I'm not hating the technology. If I hate something, it's Intel's pricing. If it's enterprise market pricing, then it's obviously expensive, but why can't I hate enterprise prising? Just make a consumer version of it, then. And I don't mean those ridiculous 16/32GB models.

But once again, I only meant it's not going to change the PC scene if you only find it in the offices, server caves, and in the PCs of the rare HC enthusiasts. As long as Optane SSDs aren't a realistic choice for most, they will stay relatively small. Although that being said, I'm also too old-fashioned since I don't care about all the cloud based stuff and such, but if that's where the world is going, naturally only server farms will need good drives and people's personal terminals could survive with some 32GB drive just fine. Although I won't be one of those people.

 

I'm not sure I understand. That's like getting irritated about the price of an 16-wheeler truck. They're very expensive, but they're not meant to be a daily driver and grocery-getter.
There will be consumer-level releases of Optane, but for the time being, Intel needs large companies and wealthy people to pay for their research and engineering, just as every new technology must do. Once that happens, it'll start to trickle down to consumer-grade hardware, and then Intel (or Micron or Samsung or whoever, doesn't matter) will release the next crazy-fast and expensive technology. The cycle continues indefinitely.

You're right that Optane isn't really going to change the PC scene, but like I said, it wasn't supposed to.

 

Hey I have a question about Optane, so I have a 480GB 900p, I never used it because im waiting to move to new case (Defien R6) that i got two months ago and build a semi new system, but too lazy to do it, maybe ill finally do when intel comes out with i9.
ANYWAY Do you need to allocate the whole optane drive for cache or you can allocate a Partition from Optane drive? Can i cut say 100Gb partition out of my optane drive and allocate it as Optane cache to one of my HDDs?