Subject: Storage | April 3, 2018 - 04:56 AM | Allyn Malventano
Tagged: Optane Memory, Optane, NVMe, Intel, 8th generation core, 800p, 3D XPoint
Remember *way* back just before CES 2017, when we caught that 'Optane Memory Storage Accelerator' entry on some Lenovo laptop release docs? Well, those obviously never happened, and we figured out why a few months later when we reviewed Intel's Optane Memory products and realized that the first iteration of these products had no apparent hardware power management capabilities, meaning they would draw excessive power while idling in a mobile platform.
While the Optane Memory launch was a year ago, just last month we tested the 800P - what was meant to be the true usable standalone M.2 packaging for Optane. This part was nearly physically identical to Optane Memory, but with some tweaks to available capacities, and more importantly, support for hardware lower power idle states. While this opened the door for use in laptops, it still did not completely close the loop on an Optane-based caching solution for mobile platforms. That loop gets closed today:
Along with a round of other new 8th generation CPU announcements (covered by Ken here), Intel has also launched a 'Core Plus' series, which are essentially the same 8th gen Core i3 / i5 / i7 parts, but with the addition of Optane Memory caching. These will be a newer, more power efficient version of the Optane Memory caching parts. While these were previously available in 16GB and 32GB capacities, this new round will add a 64GB tier to the mix.
Another update being made to Optane Memory is that instead of caching the OS drive, Optane Memory will be able to cache a secondary data drive. This would be ideal for a system that was already using a fast NVMe SSD or 800P/900P as the OS drive, where the user also wanted to cache a very large secondary data HDD. The Optane Memory caching is currently limited to caching either the OS drive or a secondary drive - no current possibility to split the higher capacity Optane Memory modules across two separate drives (we asked, and will continue to press this suggestion).
Not sure what all of this 'Optane' / '3D XPoint' stuff is all about? Check out my article detailing how it all works here
Subject: Storage | March 9, 2018 - 05:08 PM | Jeremy Hellstrom
Tagged: ssd, PCIe 3.0 x2, Optane, NVMe, Intel, Brighton Beach, 800p, 58GB, 3D XPoint, 118GB
The price of the 480GB 900P is somewhat prohibitive but the small size of the 32GB gumstick also causes one pause; hence the 800P family with a 58GB and a 118GB model. They bear price tags of $130 and $200, as you may remember from Al's review. The Tech Report also had a chance to test these two Optane sticks out, with some tests not covered in our review, such as their own real world copying benchmark. If you are looking for a second opinion, drop by and take a look.
"Intel's duo of Optane SSD 800P drives promises the same blend of impressively-low latency and performance consistency as its larger Optane devices at a price more builders can afford. We ran these drives through our storage-testing gauntlet to see whether they can make a name for themselves as primary storage."
Here are some more Storage reviews from around the web:
- Intel Optane 800P @ The SSD Review
- Intel SSD 600p Series 512 GB @ TechPowerUp
- Intel 2/8TB DC P4500 NVMe SSDs gets Reviewed - Amazing Capacity and Speed! @ The SSD Review
- he 1TB WD Blue 3D SSD @ TechARP
- Crucial MX500 500GB SSD @ Kitguru
- QNAP TS-431X2-8G 10GbE NAS Server Review @ NikKTech
- SilverStone MS09 m.2 SATA External SSD Enclosure @ Benchmark Reviews
Introduction, Specifications and Packaging
Intel has wanted a 3D XPoint to go 'mainstream' for some time now. Their last big mainstream part, the X25-M, launched 10 years ago. It was available in relatively small capacities of 80GB and 160GB, but it brought about incredible performance at a time where most other early SSDs were mediocre at best. The X25-M brought NAND flash memory to the masses, and now 10 years later we have another vehicle which hopes to bring 3D XPoint to the masses - the Intel Optane SSD 800P:
Originally dubbed 'Brighton Beach', the 800P comes in at capacities smaller than its decade-old counterpart - only 58GB and 118GB. The 'odd' capacities are due to Intel playing it extra safe with additional ECC and some space to hold metadata related to wear leveling. Even though 3D XPoint media has great endurance that runs circles around NAND flash, it can still wear out, and therefore the media must still be managed similarly to NAND. 3D XPoint can be written in place, meaning far less juggling of data while writing, allowing for far greater performance consistency across the board. Consistency and low latency are the strongest traits of Optane, to the point where Intel was bold enough to launch an NVMe part with half of the typical PCIe 3.0 x4 link available in most modern SSDs. For Intel, the 800P is more about being nimble than having straight line speed. Those after higher throughputs will have to opt for the SSD 900P, a device that draws more power and requires a desktop form factor.
- Capacities: 58GB, 118GB
- PCIe 3.0 x2, M.2 2280
- Sequential: Up to 1200/600 MB/s (R/W)
- Random: 250K+ / 140K+ IOPS (R/W) (QD4)
- Latency (average sequential): 6.75us / 18us (R/W) (TYP)
- Power: 3.75W Active, 8mW L1.2 Sleep
Specs are essentially what we would expect from an Optane Memory type device. Capacities of 58GB and 118GB are welcome additions over the prior 16GB and 32GB Optane Memory parts, but the 120GB capacity point is still extremely cramped for those who would typically desire such a high performing / low latency device. We had 120GB SSDs back in 2009, after all, and nowadays we have 20GB Windows installs and 50GB game downloads.
Before moving on, I need to call out Intel on their latency specification here. To put it bluntly, sequential transfer latency is a crap spec. Nobody cares about the latency of a sequential transfer, especially for a product which touts its responsiveness - something based on the *random* access latency, and the 6.75us figure above would translate to 150,000 QD1 IOPS (the 800P is fast, but it's not *that* fast). Most storage devices/media will internally 'read ahead' so that sequential latencies at the interface are as low as possible, increasing sequential throughput. Sequential latency is simply the inverse of throughput, meaning any SSD with a higher sequential throughput than the 800P should beat it on this particular spec. To drive the point home further, consider that a HDD's average sequential latency can beat the random read latency of a top-tier NVMe SSD like the 960 PRO. It's just a bad way to spec a storage device, and it won't do Intel any favors here if competing products start sharing this same method of rating latency in the future.
Our samples came in white/brown box packaging, but I did snag a couple of photos of what should be the retail box this past CES:
Subject: Storage, Shows and Expos | January 9, 2018 - 07:32 PM | Allyn Malventano
Tagged: XPoint, Optane, Intel, CES 2018, CES, 800p, 60GB, 3D XPoint, 120gb
Intel broke news just now that they will be launching a larger version of their 16/32GB Optane Memory modules. The new 800P looks very much the same as its little brother but is designed to operate as a sole boot SSD. Mobile applications are also possible now as the 800P includes power management features that the Optane Memory modules lacked (as they were not intended for mobile).
We are under embargo as far as performance goes, but from what we know about how Optane parts scale, it's a safe bet that performance will be very close to what we've seen out of the Optane Memory parts. Warranty will be 5 years with an endurance of ~200GB per day. No word on cost at this time. Overall these though fit nicely between Optane Memory (16/32GB) and the 900P (280/480+GB) capacity points.
The elephant in the room is the capacity. While these can store more than the 16/32GB variants, 60/120GB may not be enough for most users out there. Fortunately, devices like these are great in Zx70 RAID or even VROC configurations!
Subject: Storage | November 15, 2017 - 09:59 PM | Allyn Malventano
Tagged: NVDIMM, XPoint, 3D XPoint, 32GB, NVDIMM-N, NVDIMM-F, NVDIMM-P, DIMM
We're finally starting to see NVDIMM materialize beyond the unobtanium. Micron recently announced 32GB NVDIMM-N:
These come with 32GB of DRAM plus 64GB of SLC NAND flash.
These are in the NVDIMM-N form factor and can offer some very impressive latency improvements over other non-volatile storage methods.
Next up is Intel, who recently presented at the UBS Global Technology Conference:
We've seen Intel's Optane in many different forms, and now it looks like we finally have a date for 3D XPoint DIMMs - 2nd half of 2018! There are lots of hurdles to overcome as the JEDEC spec is not yet finalized (and might not be by the time this launches). Motherboard and BIOS support also needs to be more widely adopted for this to take off as well.
Don't expect this to be in your desktop machine anytime soon, but one can hope!
Press blast for the Micron 32GB NVDIMM-N appears after the break.
Introduction and Specifications
Back in April, we finally got our mitts on some actual 3D XPoint to test, but there was a catch. We had to do so remotely. The initial round of XPoint testing done (by all review sites) on a set of machines located on the Intel campus. Intel had their reasons for this unorthodox review method, but we were satisfied that everything was done above board. Intel even went as far as walking me over to the very server that we would be remoting into for testing. Despite this, there were still a few skeptics out there, and today we can put all of that to bed.
This is a 750GB Intel Optane SSD DC P4800X - in the flesh and this time on *our* turf. I'll be putting it through the same initial round of tests we conducted remotely back in April. I intend to follow up at a later date with additional testing depth, as well as evaluating kernel response times across Windows and Linux (IRQ, Polling, Hybrid Polling, etc), but for now, we're here to confirm the results on our own testbed as well as evaluate if the higher capacity point takes any sort of hit to performance. We may actually see a performance increase in some areas as Intel has had several months to further tune the P4800X.
This video is for the earlier 375GB model launch, but all points apply here
(except that the 900P has now already launched)
The baseline specs remain the same as they were back in April with a few significant notable exceptions:
The endurance figure for the 375GB capacity has nearly doubled to 20.5 PBW (PetaBytes Written), with the 750GB capacity logically following suit at 41 PBW. These figures are based on a 30 DWPD (Drive Write Per Day) rating spanned across a 5-year period. The original product brief is located here, but do note that it may be out of date.
We now have official sequential throughput ratings: 2.0 GB/s writes and 2.4 GB/s reads.
We also have been provided detailed QoS figures and those will be noted as we cover the results throughout the review.
Introduction, Specifications and Packaging
It’s been two long years since we first heard about 3D XPoint Technology. Intel and Micron serenaded us with tales of ultra-low latency and very high endurance, but when would we have this new media in our hot little hands? We got a taste of things with Optane Memory (caching) back in April, and later that same month we got a much bigger, albeit remotely-tested taste in the form of the P4800X. Since April all was quiet, with all of us storage freaks waiting for a consumer version of Optane with enough capacity to act as a system drive. Sure we’ve played around with Optane Memory parts in various forms of RAID, but as we found in our testing, Optane’s strongest benefits are the very performance traits that do not effectively scale with additional drives added to an array. The preferred route is to just get a larger single SSD with more 3D XPoint memory installed on it, and we have that very thing today (and in two separate capacities)!
You might have seen various rumors centered around the 900P lately. The first is that the 900P was to supposedly support PCIe 4.0. This is not true, and after digging back a bit appears to be a foreign vendor mistaking / confusing PCIe X4 (4 lanes) with the recently drafted PCIe 4.0 specification. Another set of rumors centered around pre-order listings and potential pricing for the 280 and 480 GB variants of the 900P. We are happy to report that those prices (at the time of this writing) are way higher than Intel’s stated MSRP's for these new models. I’ll even go as far as to say that the 480GB model can be had for less than what the 280GB model is currently listed for! More on that later in the review.
Performance specs are one place where the rumors were all true, but since all the folks had to go on was a leaked Intel press deck slide listing figures identical to the P4800X, we’re not really surprised here.
Lots of technical stuff above, but the high points are <10us typical latency (‘regular’ SSDs run between 60-100us), 2.5/2.0 GB/s sequential reads/writes, and 550k/500k random read/write performance. Yes I know, don’t tell me, you’ve seen higher sequentials on smaller form factor devices. I agree, and we’ve even seen higher maximum performance from unreleased 3D XPoint-equipped parts from Micron, but Intel has done what they needed to do in order to make this a viable shipping retail product, which likely means sacrificing the ‘megapixel race’ figures in favor of offering the lowest possible latencies and best possible endurance at this price point.
Packaging is among the nicest we’ve seen from an Intel SSD. It actually reminds me of how the Fusion-io ioDrives used to come.
Also included with the 900P is a Star Citizen ship. The Sabre Raven has been a topic of gossip and speculation for months now, and it appears to be a pretty sweet looking fighter. For those unaware, Star Citizen is a space-based MMO, and with a ‘ship purchase’ also comes a license to play the game. The Sabre Raven counts as such a purchase and apparently comes with lifetime insurance, meaning it will always be tied to your account in case it gets blown up doing data runs. Long story short, you get the game for free with the purchase of a 900P.
Subject: General Tech | June 8, 2017 - 11:22 AM | Alex Lustenberg
Tagged: X399, x370, x299, wwdc, video, shield, podcast, plex, pixel, macbook, Mac Pro, Logitech G413, Lian-Li, gigabyte, computex, asus, asrock, apollo lake, 3D XPoint
PC Perspective Podcast #453 - 06/07/17
Join us for talk about continued Computex 2017 coverage, WWDC '17, and more!
The URL for the podcast is: http://pcper.com/podcast - Share with your friends!
- iTunes - Subscribe to the podcast directly through the iTunes Store (audio only)
- Google Play - Subscribe to our audio podcast directly through Google Play!
- RSS - Subscribe through your regular RSS reader (audio only)
- MP3 - Direct download link to the MP3 file
Hosts: Ryan Shrout, Jeremy Hellstrom, Josh Walrath, Allyn Malventano
Peanut Gallery: Alex Lustenberg, Ken Addison
Week in Review:
News items of interest:
1:10:50 Honey, I shrunk the silicon
Hardware/Software Picks of the Week
Subject: General Tech, Memory, Storage | May 26, 2017 - 10:14 PM | Tim Verry
Tagged: XPoint, Intel, HPC, DIMM, 3D XPoint
Intel recently teased a bit of new information on its 3D XPoint DIMMs and launched its first public demonstration of the technology at the SAP Sapphire conference where SAP’s HANA in-memory data analytics software was shown working with the new “Intel persistent memory.” Slated to arrive in 2018, the new Intel DIMMs based on the 3D XPoint technology developed by Intel and Micron will work in systems alongside traditional DRAM to provide a pool of fast, low latency, and high density nonvolatile storage that is a middle ground between expensive DDR4 and cheaper NVMe SSDs and hard drives. When looking at the storage stack, the storage density increases along with latency as it gets further away from the CPU. The opposite is also true, as storage and memory gets closer to the processor, bandwidth increases, latency decreases, and costs increase per unit of storage. Intel is hoping to bridge the gap between system DRAM and PCI-E and SATA storage.
According to Intel, system RAM offers up 10 GB/s per channel and approximately 100 nanoseconds of latency. 3D XPoint DIMMs will offer 6 GB/s per channel and about 250 nanoseconds of latency. Below that is the 3D XPoint-based NVMe SSDs (e.g. Optane) on a PCI-E x4 bus where they max out the bandwidth of the bus at ~3.2 GB/s and 10 microseconds of latency. Intel claims that non XPoint NVMe NAND solid state drives have around 100 microsecomds of latency, and of course, it gets worse from there when you go to NAND-based SSDs or even hard drives hanging of the SATA bus.
Intel’s new XPoint DIMMs have persistent storage and will offer more capacity that will be possible and/or cost effective with DDR4 DRAM. In giving up some bandwidth and latency, enterprise users will be able to have a large pool of very fast storage for storing their databases and other latency and bandwidth sensitive workloads. Intel does note that there are security concerns with the XPoint DIMMs being nonvolatile in that an attacker with physical access could easily pull the DIMM and walk away with the data (it is at least theoretically possible to grab some data from RAM as well, but it will be much easier to grab the data from the XPoint sticks. Encryption and other security measures will need to be implemented to secure the data, both in use and at rest.
Interestingly, Intel is not positioning the XPoint DIMMs as a replacement for RAM, but instead as a supplement. RAM and XPoint DIMMs will be installed in different slots of the same system and the DDR4 RAM will be used for the OS and system critical applications while the XPoint pool of storage will be used for storing data that applications will work on much like a traditional RAM disk but without needing to load and save the data to a different medium for persistent storage and offering a lot more GBs for the money.
While XPoint is set to arrive next year along with Cascade Lake Xeons, it will likely be a couple of years before the technology takes off. Supporting it is going to require hardware and software support for the workstations and servers as well as developers willing to take advantage of it when writing their specialized applications. Fortunately, Intel started shipping the memory modules to its partners for testing earlier this year. It is an interesting technology and the DIMM solution and direct CPU interface will really let the 3D XPoint memory shine and reach its full potential. It will primarily be useful for the enterprise, scientific, and financial industries where there is a huge need for faster and lower latency storage that can accommodate massive (multiple terabyte+) data sets that continue to get larger and more complex. It is a technology that likely will not trickle down to consumers for a long time, but I will be ready when it does. In the meantime, I am eager to see what kinds of things it will enable the big data companies and researchers to do! Intel claims it will not only be useful at supporting massive in-memory databases and accelerating HPC workloads but for things like virtualization, private clouds, and software defined storage.
What are your thoughts on this new memory tier and the future of XPoint?
- Intel Has Started Shipping Optane Memory Modules
- Intel Optane Memory 32GB Review - Faster Than Lightning
- A Closer Look at Intel's Optane SSD DC P4800X Enterprise SSD Performance
Introduction and Specifications
XPoint. Optane. QuantX. We've been hearing these terms thrown around for two years now. A form of 3D stackable non-volatile memory that promised 10x the density of DRAM and 1000x the speed and endurance of NAND. These were bold statements, and over the following months, we would see them misunderstood and misconstrued by many in the industry. These misconceptions were further amplified by some poor demo choices on the part of Intel (fortunately countered by some better choices made by Micron). Fortunately cooler heads prevailed as Jim Handy and other industry analysts helped explain that a 1000x improvement at the die level does not translate to the same improvement at the device level, especially when the first round of devices must comply with what will soon become a legacy method of connecting a persistent storage device to a PC.
Did I just suggest that PCIe 3.0 and the NVMe protocol - developed just for high-speed storage, is already legacy tech? Well, sorta.
That 'Future NVM' bar at the bottom of that chart there was a 2-year old prototype iteration of what is now Optane. Note that while NVMe was able to shrink down the yellow bar a bit, as you introduce faster and faster storage, the rest of the equation (meaning software, including the OS kernel) starts to have a larger and larger impact on limiting the ultimate speed of the device.
NAND Flash simplified schematic (via Wikipedia)
Before getting into the first retail product to push all of these links in the storage chain to the limit, let's explain how XPoint works and what makes it faster. Taking random writes as an example, NAND Flash (above) must program cells in pages and erase cells in blocks. As modern flash has increased in capacity, the sizes of those pages and blocks have scaled up roughly proportionally. At present day we are at pages >4KB and block sizes in the megabytes. When it comes to randomly writing to an already full section of flash, simply changing the contents of one byte on one page requires the clearing and rewriting of the entire block. The difference between what you wanted to write and what the flash had to rewrite to accomplish that operation is called the write amplification factor. It's something that must be dealt with when it comes to flash memory management, but for XPoint it is a completely different story:
XPoint is bit addressible. The 'cross' structure means you can select very small groups of data via Wordlines, with the ultimate selection resolving down to a single bit.
Since the programmed element effectively acts as a resistor, its output is read directly and quickly. Even better - none of that write amplification nonsense mentioned above applies here at all. There are no pages or blocks. If you want to write a byte, go ahead. Even better is that the bits can be changed regardless of their former state, meaning no erase or clear cycle must take place before writing - you just overwrite directly over what was previously stored. Is that 1000x faster / 1000x more write endurance than NAND thing starting to make more sense now?
Ok, with all of the background out of the way, let's get into the meat of the story. I present the P4800X: