Micron's first 3D XPoint drive breaks 9 GB/s read and write speeds

mongeese

Posts: 643   +123
What just happened? The X100 claims to be the “world’s fastest SSD,” and with 9 GB/s speeds, 2.5 million IOPs and 8 μs (microsecond) latency, it’s certainly blisteringly quick. But unlike other products claiming similar specs the X100 does it without breaking a sweat: no active cooling, no unstable raid, no massive capacity required, no PCIe 4.0 exclusivity, no gigantic size, no strange drivers.

“Micron’s innovative X100 product brings the disruptive potential of 3D XPoint technology to the data center, driving breakthrough performance improvements for applications and enabling entirely new use cases,” said Micron Executive Vice President and Chief Business Officer Sumit Sadana.

Micron first partnered with Intel to form IM Flash in 2005 to develop breakthrough storage technologies. What they came up with was 3D XPoint, a cross between RAM and NAND that was durable, dense and incredibly low latency. Released in the form of Intel’s Optane SSDs it has also unfortunately been too expensive, but with Micron joining the scene, XPoint could become more attainable. Prices haven’t been announced but let’s be realistic – expensive might soon be an understatement.

Just about the only place where the X100 is lacking seems to be in competition. Current PCIe 4.0 SSDs are the quickest drives on the market and cap off at about 5 GB/s, and though that can be increased under Raid 0 for 15 GB/s, that configuration is too unstable for most uses. Intel’s Optane 905P is a closer contender with 10 μs latency but it's stuck running at a quarter of the speeds.

Perhaps the X100’s closest contender is the Samsung PM1733 and PM1735, which can reach up to 8 GB/s read speeds. Write speeds are only 3.8 GB/s, however, and they’re limited to PCIe 4.0. Intel is yet to support PCIe 4.0 which locks out most of the market. The X100, meanwhile, supports PCIe 3.0 when in an x16 slot, making it a plug and play solution.

Micron hasn’t announced a timeframe for the X100’s release, and Micron is not a company in a hurry. Though nor do they need to be: if the X100 is as good in real life as it is on paper, then it could be a while before other companies catch up.

Permalink to story.

 
News about an SSD, without ever mentioning available capacities? Seriously?

And with 2TB M.2 drives out there that can perform @5GB/s, all for $400, makes me think that a 9GB/s product shouldn't be positioned for the server market. It is not that much faster to justify huge price increase.
 
Last edited:
So does it support PCIe 4.0 and 3.0 or only 3.0 ?
PCIe is a backwards compatible standard, so anything designed for 3.0 will work in 4.0. Saying that, a PCIe 3.0 device might require more PCIe 4.0 lanes than it would if it had been designed for PCIe 4.0. All we know about the X100 is that the card can be used in an x16 PCIe 3.0 slot.
 
News about an SSD, without ever mentioning available capacities? Seriously?

And with 2TB M.2 drives out there that can perform @5GB/s, all for $400, makes me think that a 9GB/s product shouldn't be positioned for the server market. It is not that much faster to justify huge price increase.

You know it has very low latency? The normal SSD has high latency, which is why this is more suitable for server market despite the "minimal" read/write speed difference.
 
News about an SSD, without ever mentioning available capacities? Seriously?

And with 2TB M.2 drives out there that can perform @5GB/s, all for $400, makes me think that a 9GB/s product shouldn't be positioned for the server market. It is not that much faster to justify huge price increase.

You know it has very low latency? The normal SSD has high latency, which is why this is more suitable for server market despite the "minimal" read/write speed difference.
 
PCIe is a backwards compatible standard, so anything designed for 3.0 will work in 4.0. Saying that, a PCIe 3.0 device might require more PCIe 4.0 lanes than it would if it had been designed for PCIe 4.0. All we know about the X100 is that the card can be used in an x16 PCIe 3.0 slot.
Thanks, but I guess I should have been more precise. I meant to ask if it would take advantage of PCIe 4's additional bandwidth or if it's only a PCIe 3 card. If it's the latter, I think that Micron are missing out on an opportunity wrt Rome servers.
 
Back