A year and a half ago, Samsung introduced the first consumer SATA SSDs to use QLC NAND flash memory, squeezing an extra bit into each memory cell, yielding the 860 QVO to join the existing lineup of 860 EVO and PRO. Samsung's second-generation QLC NAND is now ready, and the 870 QVO is the first 870 model to be launched.

The industry put a lot of effort into preparing for the arrival of QLC NAND: beefing up error correction to compensate for lower write endurance, and tuning cache algorithms on consumer drives to forestall dealing with lower performance after SLC caches run out. But in spite of all the work it took to make QLC SSDs viable, they haven't made much of a splash and definitely aren't displacing TLC from the market yet.

The new Samsung 870 QVO SSDs arrive into a market landscape that doesn't look all that different to what the 860 QVO faced at the end of 2018. Samsung is no longer the only brand selling consumer SATA SSDs with QLC NAND, but it's not a big club. Most of the other QLC SATA drives are so low-end that the manufacturers won't commit to using any particular memory inside, and they're using whatever is cheapest at the moment. Some months, there's more to be saved by going with a DRAMless controller and TLC NAND that fell off the back of a truck, rather than using Samsung's strategy of pairing theoretically cheaper QLC NAND with a solid controller.

QLC NAND has had a bit more of an impact in the NVMe SSD market, where a few more brands are experimenting with using QLC to make big drives cheaper, rather than simply making cheap drives even cheaper. That's the same goal that Samsung had for the 860 QVO and now the 870 QVO: bringing multi-TB SSDs into the mainstream. Samsung's most visible contribution to that goal will be the introduction of the first 8TB consumer SATA SSD: the 8TB 870 QVO. That model is due to arrive a little bit later in August, and we don't have a sample yet. Today, we're looking at the 1TB and 4TB capacities of the 870 QVO.

Samsung MKX controller and LPDDR4 DRAM

The Samsung 870 QVO is an incremental update to the 860 QVO. The QLC NAND has been updated from Samsung's 64-layer V-NAND to their 92-layer V-NAND. For their NVMe product line, this change produced the 970 EVO Plus as a successor to the 970 EVO without even updating the controller, and the 970 PRO didn't get updated at all. The 870 QVO does bring a controller update, replacing the MJX with the MKX in Samsung's long line of SATA SSD controllers. Samsung hasn't disclosed any particular enhancements to their controller or firmware architecture, and we suspect this iteration is a more minor update than the last one. We know that the older MJX controller was already capable of supporting 8TB drive capacities, so that wasn't the driving force for this controller update.

Samsung 870 QVO Specifications
Capacity		1 TB	2 TB	4 TB	8 TB
Form Factor		2.5" 7mm SATA
Controller		Samsung MKX
NAND Flash		Samsung 1Tbit 92L 3D QLC
LPDDR4 DRAM		1 GB	2 GB	4 GB	8 GB
Max SLC Cache Size		42 GB	78 GB	78 GB	78 GB
Sequential Read		560 MB/s
Sequential Write	SLC	530 MB/s
	QLC	80 MB/s	160 MB/s
Random Read IOPS (4kB)	QD1	11k (SLC) 5k (QLC)	11k (SLC) 5k (QLC)
	QD32	98k (SLC) 45k (QLC)	98k (SLC) 74k (QLC)
Random Write IOPS (4kB)	QD1	35k (SLC) 22k (QLC)	35k (SLC) 34k (QLC)
	QD32	88k (SLC) 22k (QLC)	88k (SLC) 42k (QLC)
Power Consumption	Read	2.1 W	2.1 W	2.2 W	2.4 W
	Write	2.2 W	3.0 W	3.2 W	3.3 W
	Idle	30 mW	30 mW	35 mW	45 mW
	DevSlp	3 mW	4 mW	7 mW	10 mW
Warranty		3 years
Write Endurance		360 TB 0.3 DWPD	720 TB 0.3 DWPD	1440 TB 0.3 DWPD	2880 TB 0.3 DWPD
MSRP		$129.99 (13¢/GB)	$249.99 (12¢/GB)	$499.99 (12¢/GB)	TBA

Samsung continues to provide more detailed performance specifications than any other consumer SSD vendor. The big top-line numbers that everybody reports are hardly worth mentioning for a SATA drive; almost any drive can saturate the 6Gbps interface under ideal conditions, with random or sequential reads or writes at a high enough queue depth. Samsung goes the extra mile to provide specs for performance at queue depth 1, and performance after the SLC cache runs out. Some of those numbers look pretty brutal: sequential write speeds dropping to a mere 80MB/s for the 1TB model, and even random reads are considerably slower when accessing QLC data rather than the SLC cache. But overall, these specs are very similar to the 860 QVO. Random write performance at QD1 appears to have taken a bit of a hit, but everywhere else performance for the 870 QVO is rated to be equal or slightly better than its predecessor.

The 870 QVO product line still starts at 1TB, the minimum size needed to put 8 controller channels to work when using 1024Gbit NAND dies. That smallest capacity model comes with some significant performance deficits relative to the multi-TB models, in much the same fashion as 256GB TLC drives fall behind their larger counterparts, or 512GB capacities for high-performance NVMe SSDs. The most notable limitations of the 1TB 870 QVO are the post-cache write speed of 80 MB/s compared to 160 MB/s, and the cache size that is almost halved. Together that means the 1TB model is more at risk of exhibiting unacceptable performance when the SLC write cache runs out, but at 42GB this model's cache can still handle more writes than many users perform in an entire day of desktop usage.

Samsung's warranty for the 870 QVO is 0.3 drive writes per day for 3 years. This is comparable to many low-end consumer TLC drives and a step up from most other consumer QLC drives that are rated for 0.1 to 0.15 DWPD (sometimes over a 5-year warranty period, but that's still fewer total writes than the 870 QVO is rated for).

Introductory MSRPs for the 870 QVO are reduced from what the 860 QVO debuted at, with $50 off the 2TB and $100 off the 4TB. But that merely puts the 870 QVO's launch MSRP on par with the current street prices for the 860 QVO. And because it's Samsung, the pricing isn't low enough to rule out comparing against mainstream TLC SATA SSDs and entry-level NVMe SSDs, especially for the lower capacities.

In a way, that's good for this review, because the 870 QVO doesn't have much direct competition in the form of other large QLC SATA drives. Most of the SSDs that are considerably cheaper than the 860/870 QVOs are DRAMless SSDs, usually TLC but occasionally QLC. The cheapest entry-level NVMe SSDs are all either DRAMless with TLC, or use QLC with a more mainstream controller.

For this review, we are comparing the 870 QVO against the following:

• The 870 QVO's immediate predecessor, the 860 QVO
• The ADATA Ultimate SU750 and Patriot P200, two different DRAMless TLC SATA drives. The SU750 uses a Realtek controller and the 2TB P200 used a Maxio controller, both budget options.
• The Crucial MX500 and Samsung 860 EVO as mainstream SATA SSDs with TLC NAND. The MX500 has always been one of the most affordable mainstream SATA SSDs from a major brand, and the 860 EVO generally marks the most it makes sense to pay for SATA SSDs—any more, and the money would be better spent on a good NVMe SSD.
• The Intel 660p: one of the handful of QLC NVMe SSDs on the market. The 660p is being replaced by the newer 665p but both are still widely available. Even though the 665p is a bit faster, the 660p still has no trouble running circles around SATA drives under the right conditions.

We don't have any non-Samsung 4TB consumer SSDs to compare against; the Western Digital Blue is pretty much the only other 4TB consumer SATA SSD, and Sabrent's Phison E12-based Rocket and Rocket Q drives are the current options on the NVMe side. The rest of the 4+ TB SSD options are enterprise drives that lack SLC caching and idle power management and are way more expensive than anything else this review is looking at.

Ultimately, most of the competition against multi-TB SSDs comes from hard drives, the incumbent alternatives for high-capacity drives. Multi-TB hard drives are still vastly cheaper than multi-TB SSDs, but for many consumers the big question posed by drives like the 870 QVO is whether it's reasonable yet to move all their storage over to solid-state. That doesn't necessarily require SSDs to match hard drives on a $/GB basis, because there are plenty of advantages to SSDs that are worth paying at least a bit extra for.

Representing the hard drive market, we have a Western Digital WD Red 4TB, the pre-SMR WD40EFRX model. It's the newest and largest hard drive I have on hand, because my home office has been drowning in SSDs for years. (A nice problem to have.)

 

 

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 HD Programmable Power Module and accessories
• Thanks to StarTech for providing a RK2236BKF 22U rack cabinet.

Whole-Drive Fill

This test starts with a freshly-erased drive and fills it with 128kB sequential writes at queue depth 32, recording the write speed for each 1GB segment. This test is not representative of any ordinary client/consumer usage pattern, but it does allow us to observe transitions in the drive's behavior as it fills up. This can allow us to estimate the size of any SLC write cache, and get a sense for how much performance remains on the rare occasions where real-world usage keeps writing data after filling the cache.

Samsung 870 QVO 4TBSamsung 870 QVO 1TBSamsung 860 QVO 4TBSamsung 860 QVO 1TBSamsung 860 EVO 4TBSamsung 860 EVO 1TBCrucial MX500 1TBIntel SSD 660p 1TBADATA Ultimate SU750 1TBPatriot P200 2TBWD Red 4TB (5400RPM CMR)

The SLC caches on the 870 QVOs run out right on schedule, at 42 GB and 78 GB. Write performance drops precipitously but is stable thereafter, for the rest of the drive fill process. This behavior hasn't changed meaningfully from the 860 QVO.

Average Throughput for last 16 GB	Overall Average Throughput

The 870 QVO turns in scores that are very similar to its predecessor. The 1TB model averages similar write performance to a hard drive, albeit with very different performance characteristics along the way. The 4TB model manages to stay ahead of the hard drive's write performance for pretty much the entire run. Both capacities of QLC drives offer a mere fraction of the post-cache write speed of mainstream TLC drives, and even the 2TB DRAMless TLC drive offers much better sequential write performance for almost all of the test duration.

Working Set Size

Most mainstream SSDs have enough DRAM to store the entire mapping table that translates logical block addresses into physical flash memory addresses. DRAMless drives only have small buffers to cache a portion of this mapping information. Some NVMe SSDs support the Host Memory Buffer feature and can borrow a piece of the host system's DRAM for this cache rather needing lots of on-controller memory.

When accessing a logical block whose mapping is not cached, the drive needs to read the mapping from the full table stored on the flash memory before it can read the user data stored at that logical block. This adds extra latency to read operations and in the worst case may double random read latency.

We can see the effects of the size of any mapping buffer by performing random reads from different sized portions of the drive. When performing random reads from a small slice of the drive, we expect the mappings to all fit in the cache, and when performing random reads from the entire drive, we expect mostly cache misses.

When performing this test on mainstream drives with a full-sized DRAM cache, we expect performance to be generally constant regardless of the working set size, or for performance to drop only slightly as the working set size increases.

Samsung 870 QVO 4TBSamsung 870 QVO 1TBSamsung 860 QVO 4TBSamsung 860 QVO 1TBSamsung 860 EVO 4TBSamsung 860 EVO 1TBCrucial MX500 1TBIntel SSD 660p 1TBADATA Ultimate SU750 1TBPatriot P200 2TB

The 870 QVO clearly has improved read latency over its predecessors, and that's enough for the 1TB 870 to slightly outperform the ADATA SU750, a DRAMless TLC drive. The 4TB 870 QVO also shows a new behavior, with excellent random read performance at the very beginning of the test—better even that the TLC-based 860 EVOs. It looks like this test may have caught some data that was still being served from the SLC cache. Otherwise, the 870 QVOs don't care much about data locality for random reads, unlike many drives with limited or no DRAM cache.

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.

Average Data Rate
Average Latency	Average Read Latency	Average Write Latency
99th Percentile Latency	99th Percentile Read Latency	99th Percentile Write Latency
Energy Usage

The second-generation Samsung QLC drives offer slight performance increases over their predecessors on The Destroyer, but it's not enough to significantly change how a QVO rates against drives in other market segments. The 870 QVO is still trading wins against DRAMless SATA drives with TLC NAND, and clearly well behind the Intel QLC NVMe drive and the mainstream TLC SATA drives.

In general, the latency scores from the Samsung QVO drives are worse than from the DRAMless TLC drives, while the 4TB QVOs still have better overall throughput. The 1TB QVOs (both old and new) are prone to write latencies that are worse than the 5400RPM hard drive. Both capacities of the 870 QVO have worse read latency but better write latency scores than the 860 QVO.

Energy usage is a mixed bag. The 4TB 870 QVO is a slight improvement over its predecessor while the 1TB is a slight step backward. All of the QLC drives require substantially more energy to complete The Destroyer than mainstream TLC drives, and one of the DRAMless TLC drives comes out wa

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.

Average Data Rate
Average Latency	Average Read Latency	Average Write Latency
99th Percentile Latency	99th Percentile Read Latency	99th Percentile Write Latency
Energy Usage

The Heavy test is quite a bit shorter than The Destroyer, so the scores show more impact from the peak performance of SLC caching. The 4TB Samsung QVOs offer overall performance that is competitive with mainstream TLC SATA drives, when the test is run on an empty drive and they get to make use of their full SLC cache sizes. Those conditions also allow the Intel 660p to show off its NVMe performance, but the 1TB Samsung QVOs have the smallest SLC caches and worst post-cache performance, and this test is long enough for that to become a problem. When the drives are filled before running the test, all of the QLC models fall short of the mainstream TLC drives.

Performance has again changed very little from the 860 QVO to the 870 QVO. A few of the latency scores have regressed slightly, but not by enough to matter. Both capacities of the 870 QVO manage to outperform the hard drive on every performance metric; write latency comes close for the smaller, slower 870 QVO, but the read latency scores are all several times better than the hard drive can manage.

The 870 QVO still requires a lot more energy to complete the test than more high-end TLC-based drives. This mostly comes down to extra energy used as a result of the test taking longer, with write operations having the biggest impact. The 4TB QVO is again more efficient than the slower 1TB model.

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.

Average Data Rate
Average Latency	Average Read Latency	Average Write Latency
99th Percentile Latency	99th Percentile Read Latency	99th Percentile Write Latency
Energy Usage

The Light test really highlights the benefits of putting a high-end SSD controller in an otherwise entry-level product. Samsung's controllers set the bar, and on lighter workloads like this test the QLC NAND doesn't drag the 870 QVO down to the level of non-Samsung SATA drives unless the test is run on a full drive. However, such a light workload also puts the NVMe competition in its best possible light where the Intel 660p is three times faster overall.

Latency does still spike for the full-drive test runs on the QLC drives, especially when looking at 99th percentile latencies. But unlike the Intel 660p, those spikes aren't bad enough to bring the worst-case latencies of the Samsung QVOs up to hard drive levels of lag.

Aside from the extra energy used by the QLC drives on the full-drive test runs, the energy usage differences between SSDs on this test are pretty minor, and the 870 QVO shows no meaningful change from the 860 QVO.

Burst IO Performance

Our burst IO tests run at a queue depth of one and the amount of data transferred is limited to ensure that SLC write buffers don't fill up and controllers don't overheat. In between each burst there's enough idle time to keep the drive averaging a 20% duty cycle, allowing for some buffered writes and deferred garbage collection to be completed. The random read and write tests use 4kB operations and the sequential tests use 128kB operations. All the burst tests are confined to a 16GB portion of the drive, so DRAMless SSDs are not disadvantaged as much as they are for larger tests.

Random Read	Random Write
Sequential Read	Sequential Write

The Samsung 870 QVOs show significant improvement to QD1 random read performance, with the 1TB model still outperforming the 4TB model. QLC still imposes a bit of a performance penalty relative to mainstream SATA drives, but the biggest difference on display here is naturally from the NVMe drive serving the reads entirely out of its huge SLC cache.

For random writes at QD1, the 870 QVO is a stark regression from its predecessor, which was on par with the TLC-based 860 EVO. The 870 QVO is now clearly slower than mainstream TLC SSDs and is barely faster than the slower DRAMless competitor.

Sequential reads and writes at QD1 both show slight improvements, but these drives are almost all simply bumping against the limits of the SATA interface.

Sustained IO Performance

Our sustained IO tests measure performance on queue depths up to 32, but the scores reported here are only the averages for the low queue depths (1,2,4) that are most representative of real-world consumer workloads. Each queue depth is tested for up to one minute or 32GB, and the tests are confined to a 64GB span of the drive.

Random Read	Random Write
Sequential Read	Sequential Write

Random Read	Random Write
Sequential Read	Sequential Write

The most notable performance changes the 870 QVO brings to the sustained IO tests are from the 1TB model, which has greatly improved both random and sequential write performance. However, this comes at the cost of reduced random read performance, which is also a weakness for the 4TB model.

Power efficiency from the 870 QVOs during the sustained IO tests ranges from poor to average. Samsung's controller and LPDDR4 help keep power draw in check, but ultimately it takes more energy to operate slower, more complicated QLC NAND.

Random Read	Random Write
Sequential Read	Sequential Write

Looking at the big picture of the 870 QVO against all the other drives we've tested shows that the 870 QVO can reach the same top speeds as most SATA drives for three out of the four workload types. Random reads are the exception, where even high queue depths don't bring the 870 QVO up to the SATA limits during our sustained test, and the power draw is clearly on the high side there as well.

Random Read	Samsung 870 QVO 4TBSamsung 870 QVO 1TBSamsung 860 QVO 4TBSamsung 860 QVO 1TBSamsung 860 EVO 4TBSamsung 860 EVO 1TBCrucial MX500 1TBIntel SSD 660p 1TBADATA Ultimate SU750 1TBPatriot P200 2TBWD Red 4TB (5400RPM CMR)
Random Write	Samsung 870 QVO 4TBSamsung 870 QVO 1TBSamsung 860 QVO 4TBSamsung 860 QVO 1TBSamsung 860 EVO 4TBSamsung 860 EVO 1TBCrucial MX500 1TBIntel SSD 660p 1TBADATA Ultimate SU750 1TBPatriot P200 2TBWD Red 4TB (5400RPM CMR)
Sequential Read	Samsung 870 QVO 4TBSamsung 870 QVO 1TBSamsung 860 QVO 4TBSamsung 860 QVO 1TBSamsung 860 EVO 4TBSamsung 860 EVO 1TBCrucial MX500 1TBIntel SSD 660p 1TBADATA Ultimate SU750 1TBPatriot P200 2TBWD Red 4TB (5400RPM CMR)
Sequential Write	Samsung 870 QVO 4TBSamsung 870 QVO 1TBSamsung 860 QVO 4TBSamsung 860 QVO 1TBSamsung 860 EVO 4TBSamsung 860 EVO 1TBCrucial MX500 1TBIntel SSD 660p 1TBADATA Ultimate SU750 1TBPatriot P200 2TBWD Red 4TB (5400RPM CMR)

Compared to its predecessor, the 870 QVO brings slight improvements to random read performance, mainly at higher queue depths, while keeping power consumption almost unchanged. Random write performance has changed drastically for the 1TB QVO: the 1TB 870 is able to ramp up to much higher random write speeds, keeping pace with the 4TB model until the very end of the test when the smaller drive's cache finally runs out in spite of the idle time between phases of the test. The older 1TB 860 QVO's random write speed was constrained almost from the very beginning of the test. The 1TB 860 QVO also used to show a bit of fall-off in sequential read performance as the test reached higher capacities, and that behavior is gone with the 870 QVO. Sequential writes show a similarly drastic improvement for the 1TB 870 QVO, now able to generally keep pace with the larger model, which was not remotely the case for the previous generation.

Some of the big differences in write speed shown for the 1TB QVOs here may be an artifact of this test's size and duration, but even so it is clear that the smallest QV

Mixed IO Performance

Our tests of mixed read/write IO vary the workload from pure reads to pure writes at 10% increments. Each mix is tested for up to 1 minute or 32GB of data transferred. The mixed random IO test uses a queue depth of 4 while the mixed sequential IO test uses a queue depth of 1. The tests are confined to a 64GB span of the drive, and the drive is given up to one minute of idle time in between each mix tested.

Mixed Random IO	Mixed Sequential IO

The Samsung 870 QVO brings improved mixed sequential IO performance: a big boost to the 1TB model and a small increment for the 4TB. For mixed random IO, the 4TB model also gets a decent improvement, but the 1TB 870 QVO's performance is a clear regression. However, even with that regression the 870 QVO still performs far better on mixed random IO than the DRAMless SSDs.

Mixed Random IO	Mixed Sequential IO

Power efficiency scores for the mixed random IO test generally follow the same pattern as the raw performance scores, but the DRAMless SSDs aren't as far behind the QLC and mainstream TLC drives. Over on the sequential IO side, most of the Samsung SATA drives had similar overall performance, but they still have fairly substantial power efficiency differences. The 870 QVO significantly improves power efficiency on the mixed sequential IO test compared to its predecessor, and has almost caught up to the TLC-based 860 EVO.

Mixed Random IO	Samsung 870 QVO 4TBSamsung 870 QVO 1TBSamsung 860 QVO 4TBSamsung 860 QVO 1TBSamsung 860 EVO 4TBSamsung 860 EVO 1TBCrucial MX500 1TBIntel SSD 660p 1TBADATA Ultimate SU750 1TBPatriot P200 2TBWD Red 4TB (5400RPM CMR)
Mixed Sequential IO	Samsung 870 QVO 4TBSamsung 870 QVO 1TBSamsung 860 EVO 4TBSamsung 860 EVO 1TBSamsung 860 QVO 4TBSamsung 860 QVO 1TBCrucial MX500 1TBIntel SSD 660p 1TBADATA Ultimate SU750 1TBPatriot P200 2TBWD Red 4TB (5400RPM CMR)

On the mixed random IO test, the 870 QVOs show clear improvement on the more read-heavy half of the test. As the workload continues to get more write-heavy, the 4TB 870 loses its lead over the 4TB 860, and the 1TB 870 hits a performance wall when it runs out of cache where the 1TB 860 QVO had no trouble.

During the mixed sequential IO test, the Samsung SATA drives all show broadly similar performance across the range of workloads, all bottoming out near a 30% read/70% write mix. The outlier was the 1TB 860 QVO which had substantially worse performance across most of the test, including a much lower worst-case performance. The 1TB 870 QVO shows drastic improvement over its predecessor.

 

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.

The big obvious difference in the power measurements here between the 870 QVO and its predecessor is for the 4TB model. When we tested the 4TB 860 QVO, SATA link power management did work, but the drive itself never went to sleep because it was busy with background garbage collection work the entire time our testbed was trying to measure its idle power. The 4TB 870 QVO finished its cleanup work in the time allotted by this test and was truly idle during the measurements.

Idle wake-up performance is unsurprisingly quite similar for the various Samsung SATA SSDs, and the Crucial MX500 also takes about a millisecond to wake up from sleep. The two DRAMless SSDs show how budget drives often have broken power management, either by not sleeping at all (leading to near-instant wake-up), or by taking a bit long to wake up despite the sleep state still drawing quite a bit of power. And then there's the NVMe SSD which has a remarkably long wake-up time from this particular sleep state, but it was also drawing less power than the SATA drives in the slumber state.

Conclusion

QLC NAND flash memory is still something of a novelty, even for entry-level consumer SSDs. It provides cheaper, denser storage than mainstream TLC NAND, but building a well-rounded QLC SSD is a tougher challenge. In the same way that Samsung's EVO SSDs are usually the TLC drives to beat, the 870 QVO is the QLC SATA drive to beat. But most manufacturers aren't even trying, preferring to cut different corners when designing entry-level SSDs. Far more popular than using the relatively recent development of QLC NAND is the technique of using a DRAMless SSD architecture, eliminating the RAM buffer that Samsung instead splurges on to implement with the latest and greatest LPDDR versions.

So it should come as no surprise that the strengths and weaknesses of the 870 QVO fall in different areas that are typical for entry-level drives. The most acute performance problems occur when the drive is asked to write more data than can fit in its SLC cache, and then the abysmal write performance of QLC NAND is laid bare. By contrast, many entry-level DRAMless SSDs that use TLC NAND have decent sustained write performance, and most tend to suffer worst on random IO workloads.

Overall, it's hard to say whether the 870 QVO offers a better performance profile than other typical entry-level SATA SSDs. Its best-case performance is better but its worst case performance is worse. The 870 QVO does have the advantage that its weaknesses are a bit more predictable, since they almost all stem from the poor write speed of QLC NAND. DRAMless SSDs can be quite variable, as shown by the pair included in this review.

Compared to the 860 QVO, the original QLC SATA consumer SSD, the 870 QVO is an improvement in almost all respects, but only a modest incremental improvement. It smooths over some of the rough edges of the 860 QVO and doesn't bring too many new surprises. Samsung has definitely proven that consumer QLC SSDs are viable, even if they don't have a clear winner.

SATA SSD Price Comparison (June 30, 2020)
	0.5 TB	1 TB	2 TB	4 TB
Samsung 870 QVO		$129.99 (13¢/GB)	$249.99 (12¢/GB)	$499.99 (12¢/GB)
Samsung 860 QVO		$124.99 (12¢/GB)	$249.99 (12¢/GB)	$479.99 (12¢/GB)
ADATA SU750/SU760	$54.99 (11¢/GB)	$94.99 (9¢/GB)
ADATA SU800	$64.98 (13¢/GB)	$109.99 (11¢/GB)	$219.98 (11¢/GB)
Crucial BX500	$59.15 (12¢/GB)	$99.99 (10¢/GB)	$199.99 (10¢/GB)
Mushkin Source	$62.99 (13¢/GB)	$109.99 (11¢/GB)
SK Hynix Gold S31	$60.99 (12¢/GB)	$113.67 (11¢/GB)
Samsung 860 EVO	$77.99 (16¢/GB)	$139.99 (14¢/GB)	$323.54 (16¢/GB)	$619.99 (15¢/GB)
WD Blue 3D NAND/ SanDisk Ultra 3D	$64.99 (13¢/GB)	$114.99 (11¢/GB)	$226.88 (11¢/GB)	$539.99 (13¢/GB)
Crucial MX500	$69.99 (14¢/GB)	$114.99 (11¢/GB)	$229.99 (11¢/GB)
NVMe
Sabrent Rocket Q	$69.99 (14¢/GB)	$119.98 (12¢/GB)	$249.99 (12¢/GB)	$719.99 (18¢/GB)
Crucial P1	$59.99 (12¢/GB)	$104.99 (10¢/GB)	$299.99 (15¢/GB)
Intel 660p	$72.99 (14¢/GB)	$119.99 (12¢/GB)	$263.99 (13¢/GB)
Intel 665p		$129.99 (13¢/GB)	$309.99 (15¢/GB)

The 870 QVO probably shouldn't be judged solely as a competitor among entry-level consumer SSDs. It has a better business case focused on the niche of high-capacity SSDs, where there are fewer competitors and the cost savings of QLC NAND are more significant. Samsung has often been on the leading edge of consumer SSD capacity increases, having introduced 2TB and 4TB models when those still sounded a bit outrageous for a consumer-oriented product line. The most important new thing about the Samsung 870 QVO is the 8TB model that isn't actually here yet.

At 1TB and 2TB, there's usually a mainstream TLC drive to be found for less than the 870 QVO or 860 QVO. At 4TB, there are very few competitors, though at the moment the WD Blue/SanDisk Ultra 3D does appear to be a very compelling 8-12% premium to get TLC NAND. When the 8TB 870 QVO arrives, it will occupy a unique market position as the first consumer SATA SSD in that capacity class. (It is possible to buy a grey-market Micron enterprise QLC SATA drive for roughly similar pricing to what we expect for the 8TB 870 QVO, but that forfeits SLC caching and a manufacturer's warranty.)

The biggest problem with the 870 QVO is that Samsung is still using SATA. That's a shrinking market segment, but high-capacity drives are probably going to be one of the last areas where SATA still makes sense—secondary storage and NAS drives don't need the benefits of NVMe as badly. For primary storage duty, the 870 QVO is easily beaten by NVMe QLC drives that offer similar capacities and prices but much better performance overall. Here again, the 870 QVO fares better when looking at higher capacities, because the Sabrent Rocket Q is half again as expensive for 4TB (and probably also for the 8TB).

I can't really recommend the smaller two capacities of the 870 QVO given the plethora of alternatives. The larger models can almost win by default due to lack of competition, but it's hard to recommend them when so few consumers can justify buying so much SSD in the first place.