Historically, HP has not been a big player in the consumer storage market. Recently however, the HP brand has been used for several SSDs based on reference designs for Silicon Motion's SSD controllers.

Last year, the HP S700 and S700 Pro were relatively uninteresting low-end SATA SSDs in a very crowded market. This year, we have a pair of HP SSDs featuring Silicon Motion's second-generation NVMe SSD controllers and the latest Intel/Micron 64-layer 3D TLC NAND. Silicon Motion has made huge improvements to their NVMe controller design and produced a whole family of controllers. The HP EX920 uses the high-end SM2262 controller, while the HP EX900 uses the entry-level SM2263XT.

Our first hands-on experience with the SM2262 controller was at Flash Memory Summit 2017 last August, where Silicon Motion gave us the opportunity to play with early engineering samples. Between the time constraints of benchmarking during a trade show and the firmware bugs that existed at the time, we weren't able to get a good picture of the SM2262's performance, but it was clear that there was at least some truth to the very bold performance claims Silicon Motion was making - claims we viewed with skepticism given how disappointing their SM2260 first-generation NVMe controller was in drives like the Intel 600p.

The first SM2262 drive to come to market was the Intel 760p, which we liked at the time of its release, but we didn't get to test it in high enough capacities to really stress the SM2262 controller. The allure of the 760p was diminished when the price shot up shortly after release, but that made room for drives like the HP EX920.

HP EX920 Specifications
Capacity		256 GB	512 GB	1 TB
Controller		Silicon Motion SM2262
NAND Flash		Intel/Micron 64L 3D TLC
Form-Factor, Interface		double-sided M.2 2280 PCIe 3 x4 NVMe 1.3
Sequential Read		3200 MB/s	3200 MB/s	3200 MB/s
Sequential Write		1200 MB/s	1600 MB/s	1800 MB/s
Random Read		180k IOPS	340k IOPS	350k IOPS
Random Write		250k IOPS	260k IOPS	250k IOPS
Power Consumption	Active	4.29 W	5.61 W	6.23 W
	Idle	0.73 W	0.73 W	0.73 W
Warranty		5 years
Write Endurance		160 TB	320 TB	650 TB
Price		$108.89 (43¢/GB)	$175.99 (34¢/GB)	$289.99 (28¢/GB)

The HP EX920 is currently the flagship of HP's SSD brand, but it is not entirely positioned like a premium NVMe SSD. The performance specs certainly place it in that market segment, but the pricing puts it below competitors like Samsung's 970 series or the Western Digital WD Black. HP initially released the EX920 with a three year warranty and 0.8 drive writes per day endurance rating (providing the same TBW as a typical five-year/0.5 DWPD SSD), but has since extended the warranty to a full five years and updated the drive's packaging.

Our 1TB HP EX920 sample is a double-sided M.2 card with two NAND packages and one DRAM package on each side. There are some passive power regulation components on the back side, so even the lower capacities that don't need to populate the back with memory still need a few millimeters of clearance that not all laptops offer. The controller chip bears HP's logo, but underneath is the standard SM2262 controller. Silicon Motion is using the same plated copper heatspreader that the SM2260 controller featured, and the EX920 doesn't add any other form of heatsink or heatspreader.

The SM2262 controller uses a dual-core ARM CPU, but that's actually one of the least important parts of the controller design. Silicon Motion has put a lot of effort into improving their LDPC error correction engine to support higher data rates. The SM2260 already used an 8-channel NAND interface and a PCIe 3 x4 host interface, but it couldn't come close to saturating the PCIe link, especially when paired with the relatively slow first-generation Intel/Micron 3D TLC NAND.

The Competition

There are now several SM2262-based SSDs on the market. In addition to the Intel 760p, the HP EX920 also faces competition from the ADATA XPG SX8200 and the related XPG GAMMIX S11 (the SX8200, but with a heatsink). Several of the usual suspects also now have high-end NVMe drives with 3D TLC NAND: Samsung's 970 EVO, Western Digital's WD Black and SanDisk Extreme Pro, Plextor's M9Pe family.

High-end options are a little slim at the moment from brands that usually combine Toshiba NAND and Phison controllers: the low-end Phison E8 controller has been shipping for months, but the high-end Phison E12 controller is still in the sampling phase. Like Silicon Motion, Phison's first-generation E7 NVMe controller wasn't really able to compete in the high end when it was new, and without 3D NAND support the Phison E7 is completely obsolete. Toshiba's own XG5 OEM SSD also still hasn't seen a retail release to replace the Toshiba OCZ RD400, and at this point it looks like their next retail NVMe SSD will probably be based on the successor to the XG5.

AnandTech 2018 Consumer SSD Testbed
CPU	Intel Xeon E3 1240 v5
Motherboard	ASRock Fatal1ty E3V5 Performance Gaming/OC
Chipset	Intel C232
Memory	4x 8GB G.SKILL Ripjaws DDR4-2400 CL15
Graphics	AMD Radeon HD 5450, 1920x1200@60Hz
Software	Windows 10 x64, version 1709
	Linux kernel version 4.14, fio version 3.6
	Spectre/Meltdown microcode and OS patches current as of May 2018

• Thanks to Intel for the Xeon E3 1240 v5 CPU
• Thanks to ASRock for the E3V5 Performance Gaming/OC
• Thanks to G.SKILL for the Ripjaws DDR4-2400 RAM
• Thanks to Corsair for the RM750 power supply, Carbide 200R case, and Hydro H60 CPU cooler
• Thanks to Quarch for the XLC Programmable Power Module and accessories
• Thanks to StarTech for providing a RK2236BKF 22U rack cabinet.

AnandTech Storage Bench - The Destroyer

The Destroyer is an extremely long test replicating the access patterns of very IO-intensive desktop usage. A detailed breakdown can be found in this article. Like real-world usage, the drives do get the occasional break that allows for some background garbage collection and flushing caches, but those idle times are limited to 25ms so that it doesn't take all week to run the test. These AnandTech Storage Bench (ATSB) tests do not involve running the actual applications that generated the workloads, so the scores are relatively insensitive to changes in CPU performance and RAM from our new testbed, but the jump to a newer version of Windows and the newer storage drivers can have an impact.

We quantify performance on this test by reporting the drive's average data throughput, the average latency of the I/O operations, and the total energy used by the drive over the course of the test.

The average data rate from the 1TB HP EX920 on The Destroyer is a bit disappointing, falling below the lower-capacity Intel 760p and only a little over half as fast as current 1TB high-end NVMe SSDs. The EX920 is still substantially faster than SATA SSDs or low-end NVMe SSDs like the Phison E8-based Kingston A1000, but the EX920 is definitely not in the same league as Samsung's drives or the current WD Black.

The average and 99th percentile latency scores from the HP EX920 cast the drive in a slightly better light than the average data rate did, but it is still falling short of the high-end NVMe drives. The EX920 scores better than the smaller Intel 760p on these two metrics, even though the Intel drive delivered a higher average data rate.

The HP EX920's average read latency is about twice as high as the high-end NVMe SSDs. The average write latency of the EX920 is also worse than the Samsung drives and WD Black by about the same factor, but the boundary for the top tier of drives is not as sharp since the Plextor M9Pe and the aging Toshiba OCZ RD400 are also lagging behind some.

The 99th percentile read latency of the HP EX920 is decent, though it could also be said that this is simply a matter of Samsung's drives being uncharacteristically unimpressive while the WD Black sets a high standard. The 99th percentile write latency of the EX920 is clearly a problem, with more than 40% higher latency than any drive in this bunch that isn't using a Silicon Motion controller. The EX920 and Intel 760p fare much better than the previous generation Intel 600p thanks to improvements to both the flash and the controller, but it is clear that Silicon Motion still needs to work of their QoS.

The energy usage of the HP EX920 when running The Destroyer is a bit on the high side by M.2 NVMe SSD standards, but it doesn't stand out from other power-hungry drives including the Samsung 970 EVO. Unlike the 970 EVO, the EX920 doesn't justify its high power consumption with high performance, though even the 970 EVO's power consumption is hard to excuse when the WD Black offers the same performance for a little over half of the energy usage. NVMe power efficiency is still a work in progress for the industry as a whole.

AnandTech Storage Bench - Heavy

Our Heavy storage benchmark is proportionally more write-heavy than The Destroyer, but much shorter overall. The total writes in the Heavy test aren't enough to fill the drive, so performance never drops down to steady state. This test is far more representative of a power user's day to day usage, and is heavily influenced by the drive's peak performance. The Heavy workload test details can be found here. This test is run twice, once on a freshly erased drive and once after filling the drive with sequential writes.

When the ATSB Heavy test is run on a freshly-erased HP EX920, the average data rate is as high as we've seen from a flash-based SSD. However, the drive seems to be relying on aggressive SLC caching to achieve this stellar performance, because when the test is run on a full drive, the average data rate drops by almost half, leaving the EX920 with slightly sub-par full-drive performance.

The average and 99th percentile latency from the HP EX920 is top-notch when the Heavy test is run on an empty drive, but both latency figures roughly double when the test is run on a full drive. This leaves the EX920 as one of the slower drives in the high-end NVMe category.

On The Destroyer, the HP EX920's average read latency was twice that of the other high-end NVMe SSDs. Here on the Heavy test, the EX920 is quite competitive, though there's still a fairly large difference between the empty and full drive performance. The average write latency from the EX920 isn't as impressive even when the test is run on an empty drive, but it's still good, and much better than the low-end NVMe SSDs and SATA SSDs.

The 99th percentile read latency of the HP EX920 on the Heavy test is great. Even on the much slower full-drive test run, the EX920 manages to control read latency better than Samsung's SSDs. The situation is quite different for 99th percentile write latency: when the test is run on an empty drive, the EX920 performs on par with other high-end NVMe SSDs, but when the drive is full the EX920 offers worse QoS than some low-end NVMe SSDs and our comparison SATA drive.

As with The Destroyer, the HP EX920's energy usage on the Heavy test puts it in the more power-hungry tier of NVMe SSDs, but this time it has the performance to match, and the overall spread among M.2 NVMe SSDs isn't as broad.

AnandTech Storage Bench - Light

Our Light storage test has relatively more sequential accesses and lower queue depths than The Destroyer or the Heavy test, and it's by far the shortest test overall. It's based largely on applications that aren't highly dependent on storage performance, so this is a test more of application launch times and file load times. This test can be seen as the sum of all the little delays in daily usage, but with the idle times trimmed to 25ms it takes less than half an hour to run. Details of the Light test can be found here. As with the ATSB Heavy test, this test is run with the drive both freshly erased and empty, and after filling the drive with sequential writes.

The average data rates from the HP EX920 on the Light test aren't particularly impressive, with most high-end NVMe SSDs turning in higher scores. But the EX920 doesn't fall behind by enough to worry about-it's still several times faster than SATA drives, and close enough to the fastest drives that the difference isn't noticeable during ordinary usage.

The average and 99th percentile latency scores from the HP EX920 on the Light test aren't the best, but they're definitely good enough. Even the full-drive latency is not a problem, unlike for the smaller Intel 760p or 600p.

Average read and write latencies from the HP EX920 are good, especially when the test is run on a full drive. Both latency measurements stay at just a fraction of the latency of the Crucial MX500 SATA SSD.

The 99th percentile read and write latency scores from the HP EX920 are great, especially for the full-drive test run where the EX920 keeps tighter control on read latencies than any other flash-based SSD we've tested.

On the Light test, the HP EX920 doesn't fall quite as clearly into the power-hungry club. Its total energy usage on the Light test is still a bit above average, especially for the full-drive test run, but it is clearly using less energy than the M9Pe or 970 EVO.

Random Read Performance

Our first test of random read performance uses very short bursts of operations issued one at a time with no queuing. The drives are given enough idle time between bursts to yield an overall duty cycle of 20%, so thermal throttling is impossible. Each burst consists of a total of 32MB of 4kB random reads, from a 16GB span of the disk. The total data read is 1GB.

The HP EX920 surprises with a new record high burst random read performance from a flash-based SSD. It's still nothing compared Intel's Optane SSDs, but it's a substantial improvement over the competition. Even the MLC-based drives can't beat the EX920 on this test.

Our sustained random read performance is similar to the random read test from our 2015 test suite: queue depths from 1 to 32 are tested, and the average performance and power efficiency across QD1, QD2 and QD4 are reported as the primary scores. Each queue depth is tested for one minute or 32GB of data transferred, whichever is shorter. After each queue depth is tested, the drive is given up to one minute to cool off so that the higher queue depths are unlikely to be affected by accumulated heat build-up. The individual read operations are again 4kB, and cover a 64GB span of the drive.

On the longer random read test, the HP EX920 no longer stands out, but it's securely within the high-end performance bracket and isn't too far behind the fastest TLC SSD.

Power Efficiency in MB/s/W	Average Power in W

The power efficiency of the EX920 during the sustained random read test is poor, but the Samsung 970 EVO's score is almost as bad. Both drives need to improve a lot to match the WD Black or even the Toshiba XG5, which are a bit slower but draw a full 1W less power during the test.

HP EX920 1TBIntel SSD 760p 512GBSamsung 960 EVO 1TBSamsung 960 PRO 2TBSamsung 970 EVO 1TBOCZ RD400 1TBWD Black 1TB (3D NAND)Toshiba XG5 1TBPlextor M9PeGN 1TBCrucial MX500 1TBKingston A1000 960GBIntel Optane SSD 900P 280GBPatriot Hellfire 480GBIntel SSD 600p 512GB

The random read performance of the HP EX920 is very competitive at QD1 and QD2, but by QD4 the top drives are starting to pull away. At the highest queue depths, the EX920 is performing far below most other 1TB NVMe SSDs, but fortunately this poor scalability has little relevance to real-world client/consumer workloads.

Random Write Performance

Our test of random write burst performance is structured similarly to the random read burst test, but each burst is only 4MB and the total test length is 128MB. The 4kB random write operations are distributed over a 16GB span of the drive, and the operations are issued one at a time with no queuing.

The burst random write performance of the HP EX920 isn't quite the fastest we've measured, but it's still excellent. The Intel 760p managed a few percent faster with the same controller, but it's the WD Black that currently seems to have the fastest SLC write cache.

As with the sustained random read test, our sustained 4kB random write test runs for up to one minute or 32GB per queue depth, covering a 64GB span of the drive and giving the drive up to 1 minute of idle time between queue depths to allow for write caches to be flushed and for the drive to cool down.

On the longer random write test, the HP EX920 isn't quite in the top tier of drives, but it still performs better than most of the older high-end NVMe SSDs like the RD400 and XG5.

Power Efficiency in MB/s/W	Average Power in W

The power efficiency of the HP EX920 during the sustained random write test is good, but can't compete with the WD Black and Toshiba XG5. It's still a huge improvement over Silicon Motion's previous generation as represented here by the Intel 600p.

HP EX920 1TBIntel SSD 760p 512GBSamsung 960 EVO 1TBSamsung 960 PRO 2TBSamsung 970 EVO 1TBOCZ RD400 1TBWD Black 1TB (3D NAND)Toshiba XG5 1TBPlextor M9PeGN 1TBCrucial MX500 1TBKingston A1000 960GBIntel Optane SSD 900P 280GBPatriot Hellfire 480GBIntel SSD 600p 512GB

Like most other high-end NVMe SSD, the EX920 has mostly saturated its random write potential by QD4, with only a little more performance available by going all the way to QD32. The problem for the EX920 is that its limit is quite a bit lower than most of its competition, so it quickly falls behind as queue depths increase.

Sequential Read Performance

Our first test of sequential read performance uses short bursts of 128MB, issued as 128kB operations with no queuing. The test averages performance across eight bursts for a total of 1GB of data transferred from a drive containing 16GB of data. Between each burst the drive is given enough idle time to keep the overall duty cycle at 20%.

The HP EX920 breaks another record by delivering burst sequential read speeds of almost 2.5GB/s, when no other TLC-based SSD we've tested has yet hit 2GB/s on this test. The EX920 is even beating the MLC and 3D XPoint-based drives on this test.

Our test of sustained sequential reads uses queue depths from 1 to 32, with the performance and power scores computed as the average of QD1, QD2 and QD4. Each queue depth is tested for up to one minute or 32GB transferred, from a drive containing 64GB of data. This test is run twice: once with the drive prepared by sequentially writing the test data, and again after the random write test has mixed things up, causing fragmentation inside the SSD that isn't visible to the OS. These two scores represent the two extremes of how the drive would perform under real-world usage, where wear leveling and modifications to some existing data will create some internal fragmentation that degrades performance, but usually not to the extent shown here.

On the longer sequential read test with higher queue depths, the Samsung 960 PRO catches up to the EX920 and Samsung's TLC-based competitors come close. However, this only applies to a clean drive and data that was written sequentially. When the data has been fragmented by random writes, the sequential read speed of the EX920 drops by a lot, leaving it merely average and much slower than the Samsung drives (or the Optane SSD that is immune to this form of fragmentation).

Power Efficiency in MB/s/W	Average Power in W

With such high sequential read performance on a clean drive, it's no surprise to see the HP EX920 near the top of the efficiency chart, even though the drive is pulling more than 5W. On fragmented data the EX920 ends up worse off than any drive that isn't using a Silicon Motion controller, because the sequential read speed tanks while power consumption remains about the same.

HP EX920 1TBIntel SSD 760p 512GBSamsung 960 EVO 1TBSamsung 960 PRO 2TBSamsung 970 EVO 1TBOCZ RD400 1TBWD Black 1TB (3D NAND)Toshiba XG5 1TBPlextor M9PeGN 1TBCrucial MX500 1TBKingston A1000 960GBIntel Optane SSD 900P 280GBPatriot Hellfire 480GBIntel SSD 600p 512GB

The HP EX920 gets up to speed very quickly, reaching full sequential read speed at QD2 when several high-end NVMe SSDs require quite high queue depths to deliver their best performance. The EX920 is eventually surpassed by numerous competitors, but at low to mid queue depths it is definitely the fastest.

Sequential Write Performance

Our test of sequential write burst performance is structured identically to the sequential read burst performance test save for the direction of the data transfer. Each burst writes 128MB as 128kB operations issued at QD1, for a total of 1GB of data written to a drive containing 16GB of data.

The burst sequential write speed of the HP EX920 is great, but it is overshadowed a bit by the large lead Samsung's 970 EVO has on this test.

Our test of sustained sequential writes is structured identically to our sustained sequential read test, save for the direction of the data transfers. Queue depths range from 1 to 32 and each queue depth is tested for up to one minute or 32GB, followed by up to one minute of idle time for the drive to cool off and perform garbage collection. The test is confined to a 64GB span of the drive.

On the longer sequential write test, the HP EX920 falls a bit further behind as the scores of the top-tier drives are a bit more spread out than for the burst test. The EX920 is still providing performance that is worthy of a high-end NVMe SSD, but it can't match Samsung or WD.

Power Efficiency in MB/s/W	Average Power in W

The power efficiency of the HP EX920 on the sustained sequential write test is good, but several drives deliver higher performance while drawing less power, and the 970 EVO offers substantially higher performance with only slightly higher power draw. The most efficient drive is still the Toshiba XG5, which performs the same as the EX920 on this test but draws only 3W while the EX920 requires more than 5W.

HP EX920 1TBIntel SSD 760p 512GBSamsung 960 EVO 1TBSamsung 960 PRO 2TBSamsung 970 EVO 1TBOCZ RD400 1TBWD Black 1TB (3D NAND)Toshiba XG5 1TBPlextor M9PeGN 1TBCrucial MX500 1TBKingston A1000 960GBIntel Optane SSD 900P 280GBPatriot Hellfire 480GBIntel SSD 600p 512GB

Like almost all drives in this class, the EX920 shows very consistent performance across most of the test duration, with the only hint of SLC write caching effects showing up in the slight drop in performance after QD2. The Toshiba XG5 is the most notable exception, as it takes a bit longer to get up to speed.

Mixed Random Performance

Our test of mixed random reads and writes covers mixes varying from pure reads to pure writes at 10% increments. Each mix is tested for up to 1 minute or 32GB of data transferred. The test is conducted with a queue depth of 4, and is limited to a 64GB span of the drive. In between each mix, the drive is given idle time of up to one minute so that the overall duty cycle is 50%.

None of the flash-based SSDs can come close to the performance of Intel's Optane SSDs on the mixed random I/O test. But setting that aside and looking only at the drives that are priced close enough to be in direct competition, the HP EX920's performance is great: it narrowly beats the WD Black and is surpassed only by Samsung's top drives. The 1TB EX920 is substantially faster than the smaller Intel 760p and more than twice as fast as the Crucial MX500 SATA SSD.

Power Efficiency in MB/s/W	Average Power in W

The power efficiency of the EX920 during the mixed random I/O test is nothing special. As with many other tests, the HP draws more power than most drives, leading to poor efficiency when it can't also deliver top performance. The WD Black offers about the same performance with a 2W power draw instead of the 3W the HP requires, which allows the WD Black to beat even the super-fast Optane SSD for efficiency.

HP EX920 1TBIntel SSD 760p 512GBSamsung 960 EVO 1TBSamsung 960 PRO 2TBSamsung 970 EVO 1TBOCZ RD400 1TBWD Black 1TB (3D NAND)Toshiba XG5 1TBPlextor M9PeGN 1TBCrucial MX500 1TBKingston A1000 960GBIntel Optane SSD 900P 280GBPatriot Hellfire 480GBIntel SSD 600p 512GB

The HP EX920's performance accelerates throughout the mixed random I/O test as the proportion of writes increases. What holds it back is the lack of a large performance spike in the final phase of the test when the workload is 100% writes. The EX920 ends the test about 200MB/s slower than it would if its performance scaled as well as the 970 EVO.

Mixed Sequential Performance

Our test of mixed sequential reads and writes differs from the mixed random I/O test by performing 128kB sequential accesses rather than 4kB accesses at random locations, and the sequential test is conducted at queue depth 1. The range of mixes tested is the same, and the timing and limits on data transfers are also the same as above.

The HP EX920 doesn't offer the fastest performance on the mixed sequential I/O test, but it does maintain an average that's well above 1GB/s and close enough to the top drives that the difference would be barely noticeable without our benchmarking tools. The EX920 is substantially faster than drives like the Toshiba XG5 and Plextor M9Pe that also use 64L 3D TLC, and it's almost four times faster than the Crucial MX500 SATA SSD.

Power Efficiency in MB/s/W	Average Power in W

As usual, the HP EX920 draws quite a bit of power-almost as much as the Samsung 970 EVO. Since the EX920 doesn't offer quite the same level of performance, its efficiency score suffers but remains ahead of the older generation of high-end NVMe SSDs.

HP EX920 1TBIntel SSD 760p 512GBSamsung 960 EVO 1TBSamsung 960 PRO 2TBSamsung 970 EVO 1TBOCZ RD400 1TBWD Black 1TB (3D NAND)Toshiba XG5 1TBPlextor M9PeGN 1TBCrucial MX500 1TBKingston A1000 960GBIntel Optane SSD 900P 280GBPatriot Hellfire 480GBIntel SSD 600p 512GB

Like most high-end SSDs, the HP EX920's worst performance comes in the second half of this test, when there are more writes than reads but not quite enough writes to make the most of write combining and caching. The EX920 again shows less improvement than most drives when the workload finally transitions to pure writes, but the solid performance during the read-heavy half of the test and a worst case performance that barely dips below 1GB/s helps the EX920 maintain a great overall average.

Power Management Features

Real-world client storage workloads leave SSDs idle most of the time, so the active power measurements presented earlier in this review only account for a small part of what determines a drive's suitability for battery-powered use. Especially under light use, the power efficiency of a SSD is determined mostly be how well it can save power when idle.

For many NVMe SSDs, the closely related matter of thermal management can also be important. M.2 SSDs can concentrate a lot of power in a very small space. They may also be used in locations with high ambient temperatures and poor cooling, such as tucked under a GPU on a desktop motherboard, or in a poorly-ventilated notebook.

HP EX920 NVMe Power and Thermal Management Features
Controller	Silicon Motion SM2262
Firmware	SVN105
NVMe Version	Feature	Status
1.0	Number of operational (active) power states	3
1.1	Number of non-operational (idle) power states	2
	Autonomous Power State Transition (APST)	Supported
1.2	Warning Temperature	70°C
	Critical Temperature	80°C
1.3	Host Controlled Thermal Management	Supported
	Non-Operational Power State Permissive Mode	Not Supported

The HP EX920 implements most of the available NVMe features pertaining power and thermal management. The only omission is the optional Non-Operational Power State Permissive Mode, which lets the OS control whether the SSD can perform power-hungry background garbage collection even when it is supposed to be in a low-power idle state. This feature is only really useful for systems that are trying to maximize battery life even at the expense of degrading disk performance, and it's not a high priority for a performance-oriented drive like the EX920. The warning and critical temperature thresholds used by the EX920 are slightly lower than most competitors, but thermal throttling shouldn't be a big problem. The idle power states implemented by the EX920 advertise very low power draw and quick transitions in and out of idle.

HP EX920 NVMe Power States
Controller	Silicon Motion SM2262
Firmware	SVN105
Power State	Maximum Power	Active/Idle	Entry Latency	Exit Latency
PS 0	9 W	Active	-	-
PS 1	4.6 W	Active	-	-
PS 2	3.8 W	Active	-	-
PS 3	45 mW	Idle	2 ms	2 ms
PS 4	4 mW	Idle	6 ms	6 ms

Note that the above tables reflect only the information provided by the drive to the OS. The power and latency numbers are often very conservative estimates, but they are what the OS uses to determine which idle states to use and how long to wait before dropping to a deeper idle state.

Idle Power Measurement

SATA SSDs are tested with SATA link power management disabled to measure their active idle power draw, and with it enabled for the deeper idle power consumption score and the idle wake-up latency test. Our testbed, like any ordinary desktop system, cannot trigger the deepest DevSleep idle state.

Idle power management for NVMe SSDs is far more complicated than for SATA SSDs. NVMe SSDs can support several different idle power states, and through the Autonomous Power State Transition (APST) feature the operating system can set a drive's policy for when to drop down to a lower power state. There is typically a tradeoff in that lower-power states take longer to enter and wake up from, so the choice about what power states to use may differ for desktop and notebooks.

We report two idle power measurements. Active idle is representative of a typical desktop, where none of the advanced PCIe link or NVMe power saving features are enabled and the drive is immediately ready to process new commands. The idle power consumption metric is measured with PCIe Active State Power Management L1.2 state enabled and NVMe APST enabled if supported.

The active idle power consumption of the HP EX920 is pretty good compared to the competition. The 512GB Intel 760p draws a little less by virtue of having less DRAM to keep powered, and the Kingston A1000 with its PCIe x2 controller has the significant natural advantage of driving far few I/O signals. Otherwise, the EX920 represents about a 40% reduction from the typical active idle power for a 1TB high-end NVMe SSD.

When all the NVMe power features are turned on, the EX920 and the Intel 760p are in a class of their own, with power consumption dropping under 10mW even on our desktop testbed that often prevents NVMe SSDs from reaching their lowest idle states. The only way to get significantly lower idle power from an SSD is to use a well-engineered laptop that ensures full platform-level support for SATA DEVSLP (not available on any desktop) or PCIe ASPM.

The HP EX920's advertised power state transition latencies are proven to be extremely optimistic as the drive takes 54ms instead of just 6ms to wake up from its deepest idle state. But given just how low power consumption is in that idle state, 54ms of wake-up latency is a fair tradeoff and still fast enough to allow the OS to set a fairly aggressive power management policy.

Conclusion

The HP EX920 is not exactly priced like a top of the line SSD, even though it is currently the top model under the HP brand. It has consistently been selling for lower prices than the flagships from more established SSD brands like Samsung and Western Digital.

When it comes to performance, the HP EX920 is also not quite at the top, but it's close enough. On a few tests, the EX920 manages to set new performance records, but on some of our most challenging tests it falls behind the competition and comes close to dropping into the low-end NVMe segment, which is mostly occupied by drives with smaller & cheaper controllers. But even when the EX920 doesn't exactly shine, it stays well ahead of SATA drives, and the overall performance picture makes it clear that the EX920 deserves to be classified as a high-end NVMe SSD. The EX920's weaknesses are relative; it's never truly slow.

The power consumption of the HP EX920 is fairly high on almost every test. This is often justified by correspondingly high performance, but the EX920 usually ends up being a bit less efficient under load than the top competitors. Samsung's 970 EVO draws at least as much power as the EX920, but almost always offers at least a bit more performance. But right now, the drive to beat for efficiency is the WD Black, which is on average a bit faster than the HP EX920 but draws substantially less power than any other high-end NVMe SSD, usually putting even last year's Toshiba XG5 to shame.

However the HP EX920 does have one significant advantage in power consumption: it is one of the few NVMe SSDs that reaches extremely low idle power levels on our testbed. Far too many NVMe SSDs have power management bugs or compatibility limitations that keep their idle power far above advertised levels when measured on our desktop testbed. I'm sure that most of these drives can hit their rated single-digit mW ratings in the lab with a carefully configured notebook test system, but the HP EX920 drops down to a 9mW idle with no trouble at all. The Silicon Motion SM2262 controller used in the HP EX920 and the Intel 760p is the only NVMe SSD controller we've encountered that entirely follows through on its promises of low idle power. For many use cases, the excellent and reliable idle power savings will more than make up for any deficiencies in load power consumption.

Overall this review is our second look at the Silicon Motion SM2262 controller, after the Intel 760p that features unspecified Intel-specific customizations to both the firmware and the controller hardware itself. The HP EX920 puts HP's logo on the controller instead of the Silicon Motion logo, but I suspect that the drive is functionally identical to Silicon Motion's reference design – HP hasn't even removed the activity LED that ends up covered by the EX920's sticker.

The EX920 makes it abundantly clear that Silicon Motion has recovered from the shortcomings of their first NVMe controller that was so disappointing in drives like the Intel 600p. In a way, Silicon Motion has recaptured the same status they held several years ago, before NVMe and when mainstream SATA SSDs still used MLC NAND. At that time, Silicon Motion's SM2246EN controller allowed drives like the Crucial BX100 to offer performance that was pretty close to the top SSDs, but with substantially lower price tags. There was almost no reason to consider the top SATA SSDs when the mid-range was so good, and the same may now be true for the NVMe SSD market.

NVMe SSD Price Comparison
	240-256GB	400-512GB	960-1200GB
HP EX920	$108.89 (43¢/GB)	$175.99 (34¢/GB)	$289.99 (28¢/GB)
ADATA XPG SX8200	$78.99 (33¢/GB)	$159.99 (33¢/GB)	$349.99 (36¢/GB)
Intel SSD 760p	$122.31 (48¢/GB)	$218.00 (43¢/GB)	$399.95 (39¢/GB)
Samsung 970 EVO	$107.99 (43¢/GB)	$197.99 (40¢/GB)	$397.99 (40¢/GB)
Western Digital WD Black (3D NAND)	$104.99 (42¢/GB)	$199.99 (40¢/GB)	$399.07 (40¢/GB)
Kingston A1000	$69.99 (29¢/GB)	$144.77 (30¢/GB)	$279.99 (29¢/GB)
MyDigitalSSD SBX	$69.99 (27¢/GB)	$139.99 (27¢/GB)	$299.99 (29¢/GB)
Toshiba RC100	$79.99 (33¢/GB)	$154.99 (32¢/GB)
HP EX900	$89.99 (36¢/GB)	$159.99 (32¢/GB)
SATA Drives:
Crucial MX500	$69.99 (28¢/GB)	$109.99 (22¢/GB)	$199.99 (20¢/GB)
Samsung 860 EVO	$79.99 (32¢/GB)	$113.89 (23¢/GB)	$237.99 (24¢/GB)
WD Blue 3D NAND	$69.99 (28¢/GB)	$109.99 (22¢/GB)	$220.00 (22¢/GB)

The 1TB HP EX920 that we tested is the capacity with the best pricing. Samsung and Western Digital have their high-end 1TB NVMe SSDs around $400, but the HP EX920 is under $300, putting it close to low-end NVMe SSDs based on the Phison E8 controller. ADATA's similar SM2262 drive, the XPG SX8200, is currently much cheaper at sub-TB capacities, and while the Intel 760p has much higher prices across the capacity range.

With no pressure from Phison-based drives until E12 products start shipping, that leaves the rest of the high-end NVMe market to the Western Digital WD Black and Samsung 970 EVO. They are both competitively priced against the HP EX920 in the 256GB capacity class, but the larger models are selling for premium prices. It is very easy to recommend the 1TB HP EX920 instead of a slightly faster but much more expensive M.2 NVMe SSD. (Ultra-high end drives like the Samsung 970 PRO and Intel Optane SSDs are for consumers who don't look at price tags.) For smaller capacities, the ADATA SX8200 is probably a solid choice for high performance, while the MyDigitalSSD SBX offers many NVMe benefits at mainstream SATA prices.