Intel Optane Memory – 3D XPoint for Client Workloads Launched

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Owen

New Member
Feb 15, 2016
11
2
3
41
I thought the point of Optane was extremely low latency and increased write endurance (edit: I missed it - 3 DWPD (32GB) / 6 DWPD (16GB))? I find it strange that there were no mention of either of those...
 

T_Minus

Build. Break. Fix. Repeat
Feb 15, 2015
7,625
2,043
113
Disappointing to the extent that Intel's selling point is pointing out how it is used for launching applications faster, booting faster, etc... I feel like I've been hit with the salesmen stick looking at those graphics Intel provided.

I guess the same could have been said about NVME vs. SSD for all but the most power users / demanding/specific work loads... we'll see :) I have a feeling this is handicapped to prevent crossover as we'd expect. Although, you'd think in m2 format the heat would be handicap enough :) LOL
 
Last edited:

amalurk

Active Member
Dec 16, 2016
311
116
43
102
Disappointing, clearly Intel just wants to milk margins. Obviously I would rather they go all out and disrupt and try to take over the SSD space with mass production of a superior technology.
 

ATS

Member
Mar 9, 2015
96
32
18
48
Disappointing, clearly Intel just wants to milk margins. Obviously I would rather they go all out and disrupt and try to take over the SSD space with mass production of a superior technology.
3d xpoint has lower density than NAND. Also, small non-EOL SSDs aren't exactly cheap.
 

capn_pineapple

Active Member
Aug 28, 2013
356
80
28
had a huge rant, but deleted it. Basically this doesn't bring anything new to the table and the capacity point doesn't really make the applications they've touted make sense.

I do love the price point though, if the disks were double the capacity...

I suppose this would be a good ZFS ZIL/SLOG/L2ARC type solution as a competitor to the ZEUS drives. I wonder if they're supported for that kind of implementation.
 

cactus

Moderator
Jan 25, 2011
830
75
28
CA
That's what I am thinking, but I am not seeing any caps, so I'm thinking it doesn't have Emergency Power Off protection. Maybe they run cacheless so it wouldn't matter.

I suppose this would be a good ZFS ZIL/SLOG/L2ARC type solution as a competitor to the ZEUS drives. I wonder if they're supported for that kind of implementation.
 

S-F

Member
Feb 9, 2011
148
5
18
If it's true that there's no cache on these drives it means that for writes smaller than the size of the cache on a regular SSD these will be slower.

From what I'm seeing this doesn't look interesting at all for a home user and only marginally useful for enterprise. I imagine that by the time of adoption in the enterprise something a lot better will come along.

This was supposed to be near RAM performance but non-volatile.

Maybe this product can suit the needs of a super niche market?


Am I wrong?
 

ATS

Member
Mar 9, 2015
96
32
18
48
If it's true that there's no cache on these drives it means that for writes smaller than the size of the cache on a regular SSD these will be slower.

From what I'm seeing this doesn't look interesting at all for a home user and only marginally useful for enterprise. I imagine that by the time of adoption in the enterprise something a lot better will come along.

This was supposed to be near RAM performance but non-volatile.

Maybe this product can suit the needs of a super niche market?


Am I wrong?
Yes, they are already shipping these to the enterprise market and by all reports there is nothing available and nothing on the horizon that is competitive with what they are shipping.

And why would smaller writes be slower? 3D xpoint isn't block based, it doesn't need to erase sectors to write them.
 

BackupProphet

Well-Known Member
Jul 2, 2014
1,083
640
113
Stavanger, Norway
olavgg.com
The only numbers that I find interesting for a SSD is latency and random write iops at 4kb and 8kb. Anything else is marginally interesting, this is because of how operating systems works and how cache data very efficiently.

Writes on the other hand cannot be cached, and I'm sitting here with a Samsung 750 EVO that is only 2-3 times faster than spinning sas drives.
These Intel drives is a revolution for storage performance, it it just that the majority of people will read the numbers wrong and fool them self into believing these drives are slow.
 

capn_pineapple

Active Member
Aug 28, 2013
356
80
28
I mean, I'm probably going to get one or two of these, they'll of course be limited to the PCIe x2 (if I read that right, it's limited to x2 on stick so there's your bandwidth bottleneck) lanes they can be plugged into, but with the 280MB/s write speed quoted in the OP, the optane is equivalent to SATA 6G and quite a lot slower than my intel 750 NVMe drive (real world around 850MB/s from my tests). Even the 4k IOPS isn't that fantastic.

The only standout feature is that low latency, again comparing it to my 750, the optane is 125% faster on write and 4x faster on reads. Which is fantastic, or would be if it weren't slapped with the PCIe x2 bottleneck (not that it will saturate it in its current form).

Sure it'll be a decent cache drive for spinning rust, but if you're on anything slightly higher speed than the average SSD you're already outperforming this cache. Not to mention that as soon as you fill this cache (moving large files to-from a NAS) you'll be at spindle speeds anyway, as is the case with all caches.

What I really want to get my hands on is that p4800x. Now that drive makes me feel tingly in places I haven't for a long time. That to me is the real story to be told from optane until such a time as their DIMM implementation is brought to market. But I have a feeling that neither of those two solutions will be consumer friendly in the pricing.