FMS 2016: Micron Keynote Teases XPoint (QuantX) Real-World Performance

Subject: Storage | August 9, 2016 - 03:33 PM |
Tagged: XPoint, QuantX, nand, micron

Micron just completed their keynote address at Flash Memory Summit, and as part of the presentation, we saw our first look at some raw scaled Queue Depth IOPS performance figures from devices utilizing XPoint memory:

View Full Size

These are the performance figures from an U.2 device with a PCIe 3.0 x4 link. Note the outstanding ramp up to full saturation of the bus at a QD of only 4. Slower flash devices require much more parallelism and a deeper queue to achieve sufficient IOPS throughput to saturate that same bus. That 'slow' device on the bottom there, I'm pretty certain, is Micron's own 9100 MAX, which was the fastest thing we had tested to date, and it's being just walked all over by this new XPoint prototype!

Ok, so that's damn fast, but what if you had an add in card with PCIe 3.0 x8?

View Full Size

Ok, now that's just insane! While the queue had to climb to ~8 to reach these figures, that's 1.8 MILLION IOPS from a single HHHL add in card. That's greater than 7 GB/s worth of 4KB random performance!

View Full Size

In addition to the crazy throughput and IOPS figures, we also see latencies running at 1/10th that of flash-based NVMe devices.

View Full Size

..so it appears that while the cell-level performance of XPoint boasts 1000x improvements over flash, once you implement it into an actual solution that must operate within the bounds of current systems (NVMe and PCIe 3.0), we currently get only a 10x improvement over NAND flash. Given how fast NAND already is, 10x is no small improvement, and XPoint still opens the door for further improvement as the technology and implementations mature over time.

More to follow as FMS continues!


August 9, 2016 | 04:09 PM - Posted by Buyers

Exciting technology. How long before it's in reasonably priced consumer products?

August 9, 2016 | 05:29 PM - Posted by Allyn Malventano

Intel is shooting for XPoint devices (that people can buy) by the end of the year. Micron next year. Those devices might be enterprise focused, but they should be obtainable.

August 9, 2016 | 04:11 PM - Posted by Anonymous (not verified)

PCIe 4.0 can not get here soon enough for any extra bandwidth provided to feed these hungry beasts! What about any durability figures with QuantX?

August 9, 2016 | 05:27 PM - Posted by Allyn Malventano

It is durable enough to potentially avoid the need for wear leveling if that tells you anything :). The figure we've seen thrown around is 1000x the bit-level endurance compared to flash.

August 9, 2016 | 05:18 PM - Posted by Anonymous (not verified)

I've avoided NAND due to the cost, and xpoint will cost even more, but if they manage DRAM levels of durability and reliability, then it might be worth shelling out for this. I don't care about performance, but storage that just never dies is definitely worth paying for

August 9, 2016 | 09:56 PM - Posted by Anonymous (not verified)

The cost has come down a lot, I wouldn't worry too much about the durability of the new VNAND based drives. Durability really wasn't that much of a worry even before that either. Unless you are doing something unusual, they should last a very long time. Although, buggy software can be an issue. A few years ago, I had an SSD in my old Mac Book Pro. I started getting some wierd lag. After investigating, I found that Chrome was re-writing a large history file every 10 to 15 seconds. It wasn't that noticeable until it caused a slowdown since the SSD does not make any noise. That probably waisted a lot of my cycles.

While x-point is more durable than flash, it is still nowhere close to DRAM. It will still need to be one step out from DRAM in the memory hierarchy. SSDs are already so fast that I don't know how much of a difference an x-point SSD will make in a consumer level system. For most consumers, I don't think going from SATA3 to PCIe connected flash drives is really worth the added cost. X-point could allow very quick resume from sleep and such for mobile, especially if it is implemented as a memory module. You would not want to use it like DRAM though; it would wear out quickly still.

August 9, 2016 | 06:04 PM - Posted by Anonymous (not verified)

Many thanks, Allyn

So glad you're attending FMS!!

Don't forget to meet May Hwang
at Highpoint's booth,
because she did invite you
(see your email INBOX).

How about four of those U.2 Optane SSDs
using Highpoint's new 3840A NVMe RAID controller?

http://highpoint-tech.com/USA_new/CS-product_nvme.htm

YEEEEOOOWWWW!!!

August 9, 2016 | 06:06 PM - Posted by Allyn Malventano

I'm on it. Exhibit hall isn't open yet, and I forgot my cloaking device at home, so I can't sneak in :).

August 9, 2016 | 09:59 PM - Posted by John H (not verified)

Damnit Jim, I'm a storage guru not a magician!

August 10, 2016 | 01:25 PM - Posted by BlackDove (not verified)

Will be interesting to see this integrated with Knights Hill and Skylake Purley.

August 10, 2016 | 03:45 PM - Posted by Dan Lim (not verified)

Perhaps, that drive is designer for enterprise users only, not consumer level side...

August 10, 2016 | 03:45 PM - Posted by Dan Lim (not verified)

designed*

August 10, 2016 | 03:57 PM - Posted by Anonymous (not verified)

that radeon ssg says hello :)