Intel’s 3D Xpoint memory technology has been on the market for almost a year now, but mainstream builders have yet to see an Optane product they can really sink their teeth into. Intel’s 16-GB and 32-GB Optane Memory accelerators carved out a new niche by offering SSD-like (or even greater) speeds to systems otherwise hobbled by mechanical storage. Jeff’s analysis concluded that the little drives aptly handled that use case, but Intel’s decision to restrict their use to Kaby Lake and newer Core CPUs hamstrung their market appeal.
Intel’s next 3D Xpoint client drive followed several months later in the form of the data-center-derived Optane SSD 900P. By all accounts, the 900P is one beast of a drive, but it carries a price tag of $380 for 280 GB or $600 for 480 GB of storage. That’s simply too high for most builders to stomach. Who would spend $600 on storage when that much scratch could buy a couple of sticks of DDR4? (We kid, of course.)
Clearly, there’s some space left to occupy between the teeny-tiny Optane Memory drives and the exorbitant 900P. Today, Intel is launching a product intended to slide smoothly into that gap: the Optane SSD 800P.
|Optane SSD 800P|
|Capacity||Max sequential (MB/s)||Max random (IOps)||Price|
The 800P is only available in 58 GB and 118 GB capacities to start with. While it’s still much smaller than the meaty 250-GB and 500-GB class SSDs the mainstream market has grown accustomed to, these drives can easily handle a Windows installation and a few applications. This sets them apart from the Optane Memory line, which Intel hopes will still entice Kaby Lake and Coffee Lake builders on a budget. Gamers looking to keep more than one or two recent AAA titles on ultra-fast storage will still need to consider an Optane SSD 900P, though.
The 800P drives are NVMe M.2 gumsticks, but unlike most expensive drives that fit that description, they only take advantage of two lanes of PCIe 3.0 bandwidth. Intel says that decision comes down to these drives’ focus on low-latency and low-queue-depth performance versus raw bandwidth at the high queue depths that would be necessary to saturate traditional NAND SSDs. More on that in a second.
With the sticker peeled away, we can see the drives’ two 3D Xpoint packages and controller chip. Intel wasn’t ready to share details about packaging or controller implementation. All we know is that it’s an Intel controller with Intel firmware whipping Intel 3D Xpoint to extreme speeds. The company is keeping its mouth closed for the moment when it comes to technical specifics of its Optane products.
Intel isn’t at all shy about trumpeting Optane’s unique advantages versus NAND, though. The blue team promises 38% better response times than competing PCIe 3.0 x4 drives. It’s particularly bullish on the 800P’s low-queue-depth performance, where the company correctly claims that the vast majority of client workloads live.
Additionally, Intel assures us that the 800P’s sustained performance remains truly good-as-new regardless of how full the drive is. This is in stark opposition to NAND’s behavior. The performance of NAND SSDs can decline precipitously as drives are pushed closer to their total capacity. Finally, Intel touts these drives’ endurance, rating each version of the 800P for 365 terabytes written over the drives’ five-year warranties. That’s absurdly high for the size of these drives, and even higher than Intel’s more conservative initial spec of 200 TBW when the 800P was unveiled at CES.
These improvements over NAND are undoubtedly due to the storage media’s “cross point structure,” as Intel calls it. 3D Xpoint is addressable at the cell level, entirely circumventing the flash-translation-layer rigamarole that NAND drives must deal with as they juggle writing pages and erasing blocks without wearing out too quickly.
3D Xpoint’s superiority over NAND better assert itself spectacularly in these drives, because they don’t come cheap. While neither Optane SSD 800P will set you back as much as a chunk of change as the 900P will, the 58 GB and 118 GB drives still carry suggested prices of $129 and $199, respectively, working out to a whopping $2.19 and $1.68 per gigabyte.
Before we dive into testing, there’s one bit of bad news. Our storage test rigs are beginning to show their age. While the Optane 800P drives are perfectly usable as secondary storage in older systems, our venerable Asus Z97 Pro refused to acknowledge the existence of the 800P as a boot device in its BIOS, rendering it impossible to conduct boot and load tests against the other drives in our test suite. We nonetheless managed to collect our usual complement of results from IOMeter and RoboBench, and can happily throw the Optane duo alongside our usual SSD lineup. Let’s get to it.
IOMeter — Sequential and random performance
IOMeter fuels much of our latest storage test suite, including our sequential and random I/O tests. These tests are run across the full capacity of the drive at two queue depths. The QD1 tests simulate a single thread, while the QD4 results emulate a more demanding desktop workload. For perspective, 87% of the requests in our old DriveBench 2.0 trace of real-world desktop activity have a queue depth of four or less. Clicking the buttons below the graphs switches between results charted at the different queue depths.
Our sequential tests use a relatively large 128 KB block size.
In these sequential tests, the 800P drives don’t shine particularly brightly. The 118 GB is slower than some speedy SATA drives, while the smaller drive is scarcely faster than Intel’s 760p and its 3D NAND. Let’s see if random response times look any better.
In fact, they look a heck of a lot better. The Optane 800P drives post the lowest random-read latencies we’ve ever seen, period, and by no small margin. Their read response times are one-fifth that of the $1250 Samsung 960 Pro 2TB. Absurd. Write response times are also blazingly fast, just not record-setting.
Intel’s talk of responsiveness was no idle boast. Let’s see if the rest of the company’s claims bear out in our remaining tests.
Sustained and scaling I/O rates
Our sustained IOMeter test hammers drives with 4KB random writes for 30 minutes straight. It uses a queue depth of 32, a setting that should result in higher speeds that saturate each drive’s overprovisioned area more quickly. This lengthy—and heavy—workload isn’t indicative of typical PC use, but it provides a sense of how the drives react when they’re pushed to the brink.
We’re reporting IOps rather than response times for these tests. Click the buttons below the graph to switch between SSDs.
What a sight to behold. Intel said the 800P would give “fresh-out-of-box” performance regardless of how much of a beating it might be taking, and our graphs completely agree. The drives’ high peak write speed remains remarkably constant throughout the 30 minute period. The 118 GB drive suffers frequent dips to a lower rate, but it always bounces back up to its initial higher speed, spending the lion’s share of the test period there.
The Optane 800P drives broke the x-axis scale on our steady-state graph, doubling the previously unchallenged DC P3700. As our graph of our entire test period showed, these drives barely fall from their peak rates under a sustained workload. Bravo, 800P.
Our final IOMeter test examines performance scaling across a broad range of queue depths. We ramp all the way up to a queue depth of 128. Don’t expect AHCI-based drives to scale past 32, though—that’s the maximum depth of their native command queues.
For this test, we use a database access pattern comprising 66% reads and 33% writes, all of which are random. The test runs after 30 minutes of continuous random writes that put the drives in a simulated used state. Click the buttons below the graph to switch between the different drives. And note that the P3700 plot uses a much larger scale.
The Optane drives forced us to use the much-larger vertical scale we’ve only had to break out for Intel’s data-center-class P3700. Intel’s press materials were careful to emphasize the 800P’s performance at low queue depths, and here we see why. The graphs flatline after QD4. But given the dizzying height of that line, it’s nothing to complain about, as the next graphs will show.
These graphs may flatten out, but they flatten out so far above any other drive we can compare against that it’s a moot point. The 760p is a fast drive, remember, but it looks pathetic alongside the Optane 800P. Only the data-center-class P3700 manages to crest above the 800P 118 GB, and it took it until QD64 to do so. Client workloads are highly unlikely to ever exploit that kind of parallelism, and the incredible performance of these Optane SSDs from QD1 to QD4 bolsters Intel’s claims about Optane’s unique characteristics.
Our sustained and scaling tests really hammer the point home that 3D Xpoint is a fundamentally different beast than NAND. The limitations and degradations that traditional SSDs are subject to just don’t apply to Intel’s new mojo. On the next page, we’ll put away IOMeter and see how the 800P handles real-world workloads.
TR RoboBench — Real-world transfers
RoboBench trades synthetic tests with random data for real-world transfers with a range of file types. Developed by our in-house coder, Bruno “morphine” Ferreira, this benchmark relies on the multi-threaded robocopy command build into Windows. We copy files to and from a wicked-fast RAM disk to measure read and write performance. We also cut the RAM disk out of the loop for a copy test that transfers the files to a different location on the SSD.
Robocopy uses eight threads by default, and we’ve also run it with a single thread. Our results are split between two file sets, whose vital statistics are detailed below. The compressibility percentage is based on the size of the file set after it’s been crunched by 7-Zip.
|Number of files||Average file size||Total size||Compressibility|
The media set is made up of large movie files, high-bitrate MP3s, and 18-megapixel RAW and JPG images. There are only a few hundred files in total, and the data set isn’t amenable to compression. The work set comprises loads of TR files, including documents, spreadsheets, and web-optimized images. It also includes a stack of programming-related files associated with our old Mozilla compiling test and the Visual Studio test on the next page. The average file size is measured in kilobytes rather than megabytes, and the files are mostly compressible.
RoboBench’s write and copy tests run after the drives have been put into a simulated used state with 30 minutes of 4KB random writes. The pre-conditioning process is scripted, as is the rest of the test, ensuring that drives have the same amount of time to recover.
Let’s take a look at the media set first. The buttons switch between read, write, and copy results.
RoboBench often echoes what our first set of IOMeter tests reveal. The 800P runs very quickly despite only running off two PCIe lanes. Its write speeds are fine, but they hover only a little bit beyond the reach of fast SATA drives.
The work set might better showcase the 800P drives’ stellar random performance. Let’s find out.
And indeed it does. The 800P duo grabs records for the single-threaded read and copy tests, and the margin of victory is substantial. Their write performances are also good, if not quite as dominant.
RoboBench reaffirmed that sequential writes are the 800P’s weak point, but that weakness is offset by the drive’s insane random performance. Ordinarily the next page would host our boot and load tests, but as we cautioned in our introduction, our test rigs just can’t hack it with booting Optane. So flip to the next page to read about our test methods, or skip ahead to our final thoughts.
Test notes and methods
Here are the essential details for all the drives we tested:
|Adata Premier SP550 480GB||SATA 6Gbps||Silicon Motion SM2256||16-nm SK Hynix TLC|
|Adata Ultimate SU800 512GB||SATA 6Gbps||Silicon Motion SM2258||32-layer Micron 3D TLC|
|Adata Ultimate SU900 256GB||SATA 6Gbps||Silicon Motion SM2258||Micron 3D MLC|
|Adata XPG SX930 240GB||SATA 6Gbps||JMicron JMF670H||16-nm Micron MLC|
|Corsair MP500 240GB||PCIe Gen3 x4||Phison 5007-E7||15-nm Toshiba MLC|
|Crucial BX100 500GB||SATA 6Gbps||Silicon Motion SM2246EN||16-nm Micron MLC|
|Crucial BX200 480GB||SATA 6Gbps||Silicon Motion SM2256||16-nm Micron TLC|
|Crucial MX200 500GB||SATA 6Gbps||Marvell 88SS9189||16-nm Micron MLC|
|Crucial MX300 750GB||SATA 6Gbps||Marvell 88SS1074||32-layer Micron 3D TLC|
|Crucial MX500 500GB||SATA 6Gbps||Silicon Motion SM2258||64-layer Micron 3D TLC|
|Crucial MX500 1TB||SATA 6Gbps||Silicon Motion SM2258||64-layer Micron 3D TLC|
|Intel X25-M G2 160GB||SATA 3Gbps||Intel PC29AS21BA0||34-nm Intel MLC|
|Intel 335 Series 240GB||SATA 6Gbps||SandForce SF-2281||20-nm Intel MLC|
|Intel 730 Series 480GB||SATA 6Gbps||Intel PC29AS21CA0||20-nm Intel MLC|
|Intel 750 Series 1.2TB||PCIe Gen3 x4||Intel CH29AE41AB0||20-nm Intel MLC|
|Intel DC P3700 800GB||PCIe Gen3 x4||Intel CH29AE41AB0||20-nm Intel MLC|
|Mushkin Reactor 1TB||SATA 6Gbps||Silicon Motion SM2246EN||16-nm Micron MLC|
|OCZ Arc 100 240GB||SATA 6Gbps||Indilinx Barefoot 3 M10||A19-nm Toshiba MLC|
|OCZ Trion 100 480GB||SATA 6Gbps||Toshiba TC58||A19-nm Toshiba TLC|
|OCZ Trion 150 480GB||SATA 6Gbps||Toshiba TC58||15-nm Toshiba TLC|
|OCZ Vector 180 240GB||SATA 6Gbps||Indilinx Barefoot 3 M10||A19-nm Toshiba MLC|
|OCZ Vector 180 960GB||SATA 6Gbps||Indilinx Barefoot 3 M10||A19-nm Toshiba MLC|
|Patriot Hellfire 480GB||PCIe Gen3 x4||Phison 5007-E7||15-nm Toshiba MLC|
|Plextor M6e 256GB||PCIe Gen2 x2||Marvell 88SS9183||19-nm Toshiba MLC|
|Samsung 850 EV0 250GB||SATA 6Gbps||Samsung MGX||32-layer Samsung TLC|
|Samsung 850 EV0 1TB||SATA 6Gbps||Samsung MEX||32-layer Samsung TLC|
|Samsung 850 Pro 512GB||SATA 6Gbps||Samsung MEX||32-layer Samsung MLC|
|Samsung 860 Pro 1TB||SATA 6Gbps||Samsung MJX||64-layer Samsung MLC|
|Samsung 950 Pro 512GB||PCIe Gen3 x4||Samsung UBX||32-layer Samsung MLC|
|Samsung 960 EVO 250GB||PCIe Gen3 x4||Samsung Polaris||32-layer Samsung TLC|
|Samsung 960 EVO 1TB||PCIe Gen3 x4||Samsung Polaris||48-layer Samsung TLC|
|Samsung 960 Pro 2TB||PCIe Gen3 x4||Samsung Polaris||48-layer Samsung MLC|
|Samsung SM951 512GB||PCIe Gen3 x4||Samsung S4LN058A01X01||16-nm Samsung MLC|
|Samsung XP941 256GB||PCIe Gen2 x4||Samsung S4LN053X01||19-nm Samsung MLC|
|Toshiba OCZ RD400 512GB||PCIe Gen3 x4||Toshiba TC58||15-nm Toshiba MLC|
|Toshiba OCZ VX500 512GB||SATA 6Gbps||Toshiba TC358790XBG||15-nm Toshiba MLC|
|Toshiba TR200 480GB||SATA 6Gbps||Toshiba TC58||64-layer Toshiba BiCS TLC|
|Toshiba XG5 1TB||PCIe Gen3 x4||Toshiba TC58||64-layer Toshiba BiCS TLC|
|Transcend SSD370 256GB||SATA 6Gbps||Transcend TS6500||Micron or SanDisk MLC|
|Transcend SSD370 1TB||SATA 6Gbps||Transcend TS6500||Micron or SanDisk MLC|
We tested the Optane SSD 800P duo using Asus’ Hyper M.2 X4 PCIe 3.0 adapter card. All the SATA SSDs were connected to the motherboard’s Z97 chipset. The Plextor M6e was connected to the Z97 via the motherboard’s M.2 slot, which is how we’d expect most folks to run that drive. Since the XP941, 950 Pro, RD400, and 960 Pro require more lanes, they were connected to the CPU via our PCIe adapter card. The 750 Series and DC P3700 were hooked up to the CPU via the same full-sized PCIe slot.
We used the following system for testing:
|Processor||Intel Core i5-4690K 3.5GHz|
|Platform hub||Intel Z97|
|Platform drivers||Chipset: 10.0.0.13
|Memory size||16GB (2 DIMMs)|
|Memory type||Adata XPG V3 DDR3 at 1600 MT/s|
|Audio||Realtek ALC1150 with 184.108.40.20644 drivers|
|System drive||Corsair Force LS 240GB with S8FM07.9 firmware|
|Storage||Crucial BX100 500GB with MU01 firmware
Crucial BX200 480GB with MU01.4 firmware
Crucial MX200 500GB with MU01 firmware
Intel 335 Series 240GB with 335u firmware
Intel 730 Series 480GB with L2010400 firmware
Intel 750 Series 1.2GB with 8EV10171 firmware
Intel DC P3700 800GB with 8DV10043 firmware
Intel X25-M G2 160GB with 8820 firmware
Plextor M6e 256GB with 1.04 firmware
OCZ Trion 100 480GB with 11.2 firmware
OCZ Trion 150 480GB with 12.2 firmware
OCZ Vector 180 240GB with 1.0 firmware
OCZ Vector 180 960GB with 1.0 firmware
Samsung 850 EVO 250GB with EMT01B6Q firmware
Samsung 850 EVO 1TB with EMT01B6Q firmware
Samsung 850 Pro 500GB with EMXM01B6Q firmware
Samsung 950 Pro 512GB with 1B0QBXX7 firmware
Samsung XP941 256GB with UXM6501Q firmware
Transcend SSD370 256GB with O0918B firmware
Transcend SSD370 1TB with O0919A firmware
|Power supply||Corsair AX650 650W|
|Case||Fractal Design Define R5|
|Operating system||Windows 8.1 Pro x64|
Thanks to Asus for providing the systems’ motherboards, to Intel for the CPUs, to Adata for the memory, to Fractal Design for the cases, and to Corsair for the system drives and PSUs. And thanks to the drive makers for supplying the rest of the SSDs.
We used the following versions of our test applications:
- IOMeter 1.1.0 x64
- TR RoboBench 0.2a
- Avidemux 2.6.8 x64
- LibreOffice 4.3.2
- GIMP 2.8.14
- Visual Studio Express 2013
- Batman: Arkham Origins
- Tomb Raider
- Middle Earth: Shadow of Mordor
Some further notes on our test methods:
- To ensure consistent and repeatable results, the SSDs were secure-erased before every component of our test suite. For the IOMeter database, RoboBench write, and RoboBench copy tests, the drives were put in a simulated used state that better exposes long-term performance characteristics. Those tests are all scripted, ensuring an even playing field that gives the drives the same amount of time to recover from the initial used state.
- We run virtually all our tests three times and report the median of the results. Our sustained IOMeter test is run a second time to verify the results of the first test and additional times only if necessary. The sustained test runs for 30 minutes continuously, so it already samples performance over a long period.
- Steps have been taken to ensure the CPU’s power-saving features don’t taint any of our results. All of the CPU’s low-power states have been disabled, effectively pegging the frequency at 3.5GHz. Transitioning between power states can affect the performance of storage benchmarks, especially when dealing with short burst transfers.
The test systems’ Windows desktop was set at 1920×1080 at 60Hz. Most of the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.
Intel’s Optane SSD 800P breaks new ground in our storage test results. Never before have we seen such breathtaking random response times. Nor have we seen a drive stubbornly cling to its peak random-write speeds while writing its entire capacity. The 800P fulfills all the claims Intel makes for it in its marketing materials. Sadly, being unable to boot the drive on our aging test platform means we don’t have enough results to compare it to the rest of our crew in overall performance rankings. And we can’t graph its price-to-performance ratio on our scatter plots, either. But we’ll soldier on and conduct our final analysis even without visual aids.
Even with this Optane duo’s world-class performance in certain workloads, $129 and $199 are incredibly dear prices for 58-GB and 118-GB SSDs. $129 can easily buy a 500-GB-class SATA drive, and $199 is enough to move up to a 500-GB-class NVMe stick or get most of the way to a terabyte of NAND. PC gamers and demanding users don’t just need speed, they need capacity, and the Optane SSD 800P can only fulfill one of those demands.
For the curious, 58 GB might be enough room for a Windows installation and some applications, but it won’t leave a lot of breathing room. The 118-GB drive might offer room for a small music collection or one or two older AAA games. No matter how you slice it, allocating the money to good old NAND instead of 3D Xpoint gets builders a lot more space to stretch their legs, and we imagine cost-per-gigabyte will remain the primary roadblock to Optane adoption.
DIY builders’ wallets are already strained by the ongoing price insanity for graphics cards and RAM, and the 800P duo’s prices are too high too squeeze into prebuilts the way Optane Memory might. The 800P is a non-starter as an upgrade path for older systems that might not boot from an NVMe SSD, too. The drives’ low-queue-depth performance is nice, to be sure, but it’s certainly not worth the price of admission as secondary storage. Cost and capacity are more important considerations in that use case, as we see it.
Intel suggests that RAIDing the larger 800P could be a route to higher capacities and even higher performance from Optane, but we see little point to that exercise when the 280-GB Optane SSD 900P lists for less money than a pair of 118-GB 800Ps, offers even more space, and doesn’t have to contend with any potential headaches of a RAID.
Nonetheless, it’s hard not to get excited for the possibilities that 3D Xpoint opens up. Despite our inability to put it to the test this time around, we’ve seen how Optane’s blistering response times translate into less time waiting around for stuff to load. And the fact that the drives suffer no performance loss as they fill up alleviates a pain point we’ve all had to deal with over the years. NAND is great tech, to be sure, but being forced to treat a portion of a drive’s capacity as unusable to preserve its performance can get frustrating.
Overall, the high price tag and limited capacity of this duo makes it difficult to recommend the Optane 800Ps for all but those curious about the potential of this technology. Intel has got a good thing going here, but it will take some sticker-slashing before we can start heaping the awards on Optane. Since 3D Xpoint is Intel and Micron’s proprietary tech, though, it’s highly unlikely any Optane competitors will surface any time soon to exert price pressure. Micron’s QuantX might eventually burst forth from the shadows, but QuantX products will likely be for the data center when they do appear. For the time being, moving beyond NAND in the hobbyist market will remain a rich man’s game.