FMS 2014: HGST Claims 3 Million IOPS and 1.5us Access Time SSD - updated with pics

Subject: General Tech, Storage, Shows and Expos | August 7, 2014 - 11:17 AM |
Tagged: ssd, phase change memory, PCM, hgst, FMS 2014, FMS

According to an HGST press release, the company will bring an SSD based on phase change memory to the 2014 Flash Memory Summit in Santa Clara, California. They claim that it will actually be at their booth, on the show floor, for two days (August 6th and 7th).

The device, which is not branded, connects via PCIe 2.0 x4. It is designed for speed. It is allegedly capable of 3 million IOPS, with just 1.5 microseconds required for a single access. For comparison, the 800GB Intel SSD DC P3700, recently reviewed by Allyn, had a dominating lead over the competitors that he tested. It was just shy of 250 thousand IOPS. This is, supposedly, about twelve times faster.

View Full Size

While it is based on a different technology than NAND, and thus not directly comparable, the PCM chips are apparently manufactured at 45nm. Regardless, that is significantly larger lithography than competing products. Intel is manufacturing their flash at 20nm, while Samsung managed to use a 30nm process for their recent V-NAND launch.

What does concern me is the capacity per chip. According to the press release, it is 1Gb per chip. That is about two orders of magnitude smaller than what NAND is pushing. That is, also, the only reference to capacity in the entire press release. It makes me wonder how small the total drive capacity will be, especially compared to RAM drives.

Of course, because it does not seem to be a marketed product yet, nothing about pricing or availability. It will almost definitely be aimed at the enterprise market, though (especially given HGST's track record).

*** Update from Allyn ***

I'm hijacking Scott's news post with photos of the actual PCM SSD, from the FMS show floor:

View Full Size

View Full Size

In case you all are wondering, yes, it does in fact work:

View Full Size

View Full Size

View Full Size

One of the advantages of PCM is that it is addressed at smaller sections as compared to typical flash memory. This means you can see ~700k *single sector* random IOPS at QD=1. You can only pull off that sort of figure with extremely low IO latency. They only showed this output at their display, but ramping up QD > 1 should reasonably lead to the 3 million figure claimed in their release.

Source: HGST
August 5, 2014 | 05:58 AM - Posted by Anonymous (not verified)

How many laptop makers use PCI based Flash drives, PCI appears to be the way to go for the main system drive on laptops/PCs, and if any thing could improve the performance of a laptop it is a PCI based flash drive, I'd love to see one of these devices tested in a gaming laptop form factor! I am not as interested in the "ultrabook" types of systems, though this will probably be used in ultrabooks/macbooks, But for gaming laptops, and the much ignored, By Intel, desktop replacement laptops that were going to be designed, before Intel got itself sidetracked with its not so successful ultrabook initiative. If they really want to bring life back into the Laptop market, the OEMs should drop the ultrabook only craziness, and begin to manufacture regular form factor laptops, as true desktop replacements, and build more midrange laptops with discrete graphics, and faster PCI based main system drives, and use USB 3.1 10GB ports, Intel has really dropped the ball in getting thunderbolt out there, as a replacement for anything other than firewire. It looks like HP is the only few OEMs with midrange laptop models, that come with discrete GPUs, with any ability to function as a near desktop replacement for under a thousand dollars.

August 8, 2014 | 01:23 AM - Posted by Anonymous (not verified)

"I'd love to see one of these devices tested in a gaming laptop form factor!"

It is somewhat unclear to me what people expect from these faster SSDs. How would this change gaming performance? If you have sufficient ram, then the hard disk/SSD is only hit when you load levels and such. I doubt you would see much difference between a sata 6 ssd and an x4 pci-e ssd for most consumer level use. I have a really slow ssd and my OS boots in a few seconds and programs load almost instantly. For gaming, these SSDs should only make a difference when the game is first loaded and when levels or any other additional data is loaded. Unless the game is designed really badly (or you do not have enough ram), then this should not happen often.

The point is, these are still being used as a solid state disk, which the whole rest of the system architecture and software has assumed to be really slow for a long time. This means that everything is designed to access it as little as possible and cache all of the important stuff in ram. This also means that increasing the speed of the ssd will have no effect most of the time since the data will already be in system ram. Devices like this PCM memory only come into play once the system architecture is changed to expect this storage to actually be fast. I expect the system architecture changes to be several years out at the moment.

Note that PCM memory has been talked about for a very long time, but has not actually been delivered. Hopefully this will actually be a real product. I would assume a current product based on this would be aimed at enterprise applications if they can actually mass produce it.

August 5, 2014 | 05:59 AM - Posted by Anonymous (not verified)

PCM... Phase Change Memory? We heard about this about 10 years ago, but I thought it had been abandoned. IIRC, Intel was leading the charge back then.

August 5, 2014 | 09:02 AM - Posted by mAxius

i know Allyn is looking forward to this

August 6, 2014 | 10:19 AM - Posted by Berto Ray

Set Phase Changers to stun.

August 7, 2014 | 02:12 PM - Posted by Anonymous (not verified)

Aren't those "screenshots" of incredibly UNimpressive benchmarks? When did 3MM IOPS become 400k for random reads? Why is bandwidth in the SATA3 range?

Am I misreading these?

August 8, 2014 | 06:34 AM - Posted by jgstew

The screenshots are showing speed at (QD=1) Queue Depth 1, which is the worst case scenario for total peak performance of high speed storage like this. This is still many times faster than most SSDs at QD=1.

Also, with storage this fast, most workloads will not cause enough IO to get the Queue Depth much above 1.

August 13, 2014 | 12:59 PM - Posted by Anonymous (not verified)

The 400k is for 4k reads, that's where their PCIe link bandwidth tops out. Not really related to low latency at small random reads.

August 7, 2014 | 07:38 PM - Posted by Anonymous (not verified)

Protected Sub GameChanger(string isThisReal)
{
If (isThisReal == "yes")
{
GameChanger("yes")
}
}

Let's hope itsNotReal. Wouldn't want to crash the universe with that infinite loop.

August 8, 2014 | 01:37 AM - Posted by Anonymous (not verified)

Looks like it is using an fpga as the controller, so performance of this development board may not be as high as the final product, if it actually does come to market.

August 8, 2014 | 01:38 AM - Posted by Anonymous (not verified)

Looks like it is using an fpga as the controller, so performance of this development board may not be as high as the final product, if it actually does come to market.

August 8, 2014 | 01:38 AM - Posted by Anonymous (not verified)

Looks like it is using an fpga as the controller, so performance of this development board may not be as high as the final product, if it actually does come to market.

August 9, 2014 | 03:57 PM - Posted by Anonymous coward (not verified)

They demoed 3M with 512B IOs at high QD, I think 8, and 400K at low QD, but that was for 4KB IOs. It is quite a lot faster than 8000 that I get on my SATA SSD.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.