Quote:Western Digital on Tuesday formally announced its fourth-generation 3D NAND memory, developed as part of the Western Digital/Toshiba joint venture. The fourth-generation BiCS NAND flash chips from Western Digital feature 96 layers and will include several capacity points and will use TLC and QLC architectures. The company expects to start volume production of BiCS4 chips in 2018.

Quote:Toshiba America Electronic Components, Inc. (TAEC) today announced the latest generation of its BiCS FLASH three-dimensional (3D) flash memory. The newest BiCS FLASH device features 4-bit-per-cell, quadruple-level cell (QLC) technology and is the first 3D flash memory device to do so. Toshiba's QLC technology enables larger (768 gigabit) die capacity than the company's third-generation 512Gb 3-bit-per-cell, triple-level cell (TLC), and pushes the boundaries of flash memory technology.
...
Samples of Toshiba's groundbreaking QLC device began shipping earlier in June to SSD and SSD controller vendors for evaluation and development purposes. Additionally, samples will be showcased at the 2017 Flash Memory Summit, taking place from August 7-10 in Santa Clara, California.

Quote:Besides intention to produce 768 Gb 3D QLC NAND flash for the aforementioned devices, the most interesting part of Toshiba’s announcement is endurance specification for the upcoming components. According to the company, its 3D QLC NAND is targeted for ~1000 program/erase cycles, which is close to TLC NAND flash. This is considerably higher than the amount of P/E cycles (100 – 150) expected for QLC by the industry over the years. At first thought, it comes across a typo - didn't they mean 100?. But the email we received was quite clear:

- What’s the number of P/E cycles supported by Toshiba’s QLC NAND?
- QLC P/E is targeted for 1K cycles.

It is unclear how Toshiba managed to increase the endurance of its 3D QLC NAND by an order of magnitude versus initially predicted. What we do know is that signal processing is more challenging with QLC than it is with TLC, as each cell needs to accurately determine sixteen different voltage profiles (up from 2 in SLC, 4 in MLC, and 8 in TLC).

Quote:We're starting to hear about 96-layer BiCS FLASH with an expected date of the end of 2018. There are also whispers of QLC coming from Toshiba at that time. The E12 will overlap with Toshiba's next generation memory. 96-layer QLC is a story for another day, but mid-year or at Flash Memory Summit looks to be a good time for companies to preview products based on the technology.

Quote:Last but not least is the most interesting leak: In the middle of the chart is the Intel SSD 660p with 4-bit per cell (QLC) flash. The SSD 660p listing shows three capacities (512GB, 1TB and 2TB). The performance is much higher than we expected to see from QLC at right out of the gate. The leak says the 660p will achieve up to 1,800 MB/s sequential read and 1,100 sequential write speeds. The random performance clocks in at 150,000 IOPS for both reads and writes.

QLC was a hot topic at CES last week, but only behind closed doors. No one wanted to go on the record, but we know IMFT (Intel Micron Flash Technology) has it ready for the most part. Companies are excited about the cost-cutting technology but need controllers to pair with it. One source told us to expect 512GB QLC SSDs for around $100.

Quote:The controller will be part of Marvell's next generation controller product family addressing consumer, cloud data center, and enterprise SSDs. In the consumer space, manufacturers can built add-in card or M.2 form factors. The new product family will have controllers that can run in an enterprise-focused dual port PCIe 3.0 x2 configuration, most likely in a U.2 form factor with two lanes each going to separate nodes (computers in the same server chassis).

We first heard about the controller while talking with various companies about QLC NAND, four bits per cell. Every controller manufacturer is feverishly working to get hardware ready for the low endurance NAND that requires increased error correction technology. Marvell’s NANDEdge ECC technology is expected to be a key enabler of future QLC SSD solutions.

Quote:Samsung's stacks of flash have tripled in height since then—the company recently announced its forthcoming fifth-generation, 96-layer V-NAND. That flash may mark the end of line for Samsung's layer jenga. The company has hinted it will likely be seeking future gains through means other than adding more layers.

Quote:Applied Material's Sean Kang, speaking at Japan's International Memory Workshop (IMW), said that he expects future 3D NAND technologies to achieve 140 layers (up from today's leading 64-layer tech) by 2021. Increased numbers of layers will allow for increased die densities whilst keeping the same PCB real-estate and implementation area; at the same time, which is something the industry is craving for as data-sets only continue to increase in size. Before 2021 and its 140-layer NAND comes (which will require new fabrication materials), 90-layer solutions are expected to be launched this year, with a 20% decrease in layer height, down from its current 60 nm to 55 nm, which will allow for relatively stable stack heights, even as the number of layers increases significantly (by around 40% compared to 64-layer tech). Cheaper, more dense NAND tech - what isn't there to like?

Quote:One interesting point Micron made to me when we spoke was that the endurance needs of SSDs are actually decreasing, in many areas, rather than increasing. At first glance, this might seem counterintuitive. After all, the amount of data we collectively create each year has been growing for years. As it turns out, however, more advanced operating systems that return more data on how much data is actually written to drives per day in enterprise deployments has shown that the number of writes is lower, in some cases, than was previously thought.

Meanwhile, rapid growth in SSD capacities has meant that drives, generally speaking, are now much larger than they once were. This naturally decreases the number of drive writes per day that are practically going to be performed. We even referenced this idea earlier this year, when we noted that Nimbus’ 100TB SSD is so huge, you literally can’t perform one drive write per day if you assume that the SSD maintains its maximum rated transfer speed 24 hours per day.

When you put these trends together, you’ve got a potentially large market for SSDs in industries that have historically still been using HDDs, or might only be using SSDs for caching. The Ion 5210 QLC isn’t expected to replace TLC drives, but to serve as an adjunct to them, offering better than hard drive performance; significantly higher drive capacities, thanks to the 1.33x improvement in data stored per-cell; and a better overall price tag compared with MLC or TLC drives over the long term. Micron isn’t sharing more details than that at the moment, but the company has stated that it expects to give more information later this year.

Quote:Even as researchers expect 3D NAND flash to achieve the 140-layer level by 2021, technology and manufacturers still have to take all the intermediate steps before we're actually there. In that sense, Western Digital has just announced that they're well on their way in producing 96-layer 3D NAND and distributing it to customers. For now, the memory will be used for inexpensive storage solutions, but the idea is to eventually ramp um production for other, higher-performance products.

Quote:In their recent earnings call, Micron commented that they have 96-layer 3D NAND technology on track for volume shipments in the second half of 2018. Most of today's SSDs typically use 32-layer technology, with 64-layer flash chips used in some recent releases like the Crucial MX500. 96-layer is the third generation of 3D NAND and increases storage capacity per chip even further which allows smaller and more energy efficient mobile devices to be built. Of course it will be cheaper too, compared to current-generation 64 layer NAND, which should bring SSD pricing down even more, and of course generally help pricing of consumer products which use flash memory.

Quote:Samsung has delivered another salvo in the race to ever-increasing NAND flash density and performance. The company has begun mass production of its fifth-generation V-NAND memory with "over 90" (likely 96) layers per die.

Quote:Not to be upstaged, flash partners Toshiba and WD both announced their jointly-produced 3D QLC NAND yesterday in separate announcements. The announcements contained quite a bit more information, so we know that the new 96-layer BiCS4 flash has a density of 1.33 terabits2 per die.

SSDs come with up to 16 die per package, so that means Toshiba and WD will soon be able to pack in a whopping 2.66TB of storage into a single NAND package. The duo plans to begin sampling the BiCS4 QLC NAND in September and mass production is slated for early 2019. More importantly, WD's release clearly states that QLC SSDs will come to the consumer market under the SanDisk brand name. We expect Toshiba to follow suit with its own consumer SSDs.

SSDs have slowly chipped away at HDD market share over the last few years and WD even recently decided to shut down one of its primary HDD production plants due to reduced demand. SSD prices continue to plunge and many analyst firms predict even larger drops as the Chinese Yangtze River Storage Technology fabs begin to pump out 3D NAND. Samsung is also reportedly gearing up to increase its NAND production dramatically with a $2.6 billion increase in spending in 2019. Next year is shaping up to the year of QLC SSDs, which could be the final ingredient needed to push HDDs out of the primary storage role entirely.

Quote:The best part about the 660p is its performance numbers. The drive takes advantage of PCI-Express 3.0 x4, and offers (at least on paper), performance numbers identical to those of the pricier 700p. The drives read at speeds of up to 1800 MB/s, with up to 1100 MB/s writes. The 600p, in comparison, capped out at 560 MB/s sequential writes, while the 700p is only slightly higher, at 1200 MB/s. Random access speeds are up to 150,000 IOPS (both reads and writes). QLC pays off rich dividends to consumers by way of price/GB. The 660p 512 GB is expected to be priced at 113.90€ (0.22€/GB), the 1 TB variant at 197.75€ (0.20€/GB), and the 2 TB variant at 391.43€ (0.20€/GB). Not bad for launch prices, considering these are PCIe NVMe drives priced competitively with SATA SSDs.

Quote:Geizhals, a German online price comparison platform, revealed that the Intel 660p 512GB costs €112.90 ($131.11), while the larger 1TB and 2TB models go for €218.90 ($254.17) and €431.90 ($501.48), respectively. If we do some simple mathematics and divide the price by the capacity, it comes down to $0.25 per a gigabyte, which puts it right in the price range of a typical SATA SSD.

Intel hasn't announced the availability of the Intel 660p. Online speculation suggests the SSDs should launch in the second half of this year. PC-Canada expects to have them in stock by August 25. This year's installment of The Flash Memory Summit will take place on August 7. We'll be at the show, and it's possible that Intel could take the opportunity to officially present its 660p SSD lineup at the event (if so, we'll let you know).

Quote:Intel released the original 600p back in 2016 for nearly twice the price-per-GB as the 660p is today, but the drive was barely faster than the much cheaper SATA-based competition. The 660p changes that.

Intel's 660p is just $0.20-per-GB. That value is hard to ignore when the drive is the same price, if not cheaper, than the SATA-based competition. The 600p has half to one-third of the endurance of some competing drives, so its low price point does come at the cost of endurance. In reality, most consumers don’t need that much endurance if their average use case involves mostly office applications, web browsing, and content streaming. For heavier workloads, like frequent large file transfers or productivity applications, it is best to select an SSD with more endurance, like the NVMe Adata XPG SX8200 or the SATA Crucial MX500.

The 660p's included SSD Toolbox and five-year warranty are icing on the cake. The inclusion of 256-bit AES hardware encryption with Pyrite 1.0 and 2.0 support enables fast performance and tough security for the mobile market, and the thin single-sided M.2 2280 profile assures broad compatibility with laptops.

The 660p is a refined and welcome update to Intel's SSD 6 series. The 660p proves that Intel’s QLC NAND is ready for the mainstream and easily earns our top value recommendation for the low-endurance/budget segment. If you haven’t upgraded to an SSD yet and your workloads tend to be mundane, Intel’s SSD 660p is a good fit.

Quote:All in all, we now have three high-end NVMe SSDs that rival each other: the Samsung 970 Evo, the WD Black 2018 and the Toshiba XG6. Each model has its own strengths and weaknesses, but overall they perform similarly. Coming from a situation where Samsung had exclusive rights in the high-end segment, that is obviously a very pleasant situation for the consumer. Also pleasing to consumers is that Toshiba has the potential to price the XG6 well below the competition, thanks to the implementation of its 96-layer BiCS 4 flash memory.

However, for the time being we do not know the price of the Toshiba XG6, and it seems that you will probably have to wait a few months before you can buy this SSD or any spin-off under Toshiba's consumer brand OCZ. Just before our publication, Toshiba even informed us that mass production would probably not start until early 2019. However, due to the fact that Toshiba is the first to utilize the new 96-layer NAND, they promote themselves to a full player in the high-end segment. Therefore, we give the Toshiba XG6 a Hardware.Info Innovation Award.

Quote:IMFlash Technologies (IMFT), the Intel-Micron joint venture that manufactures NAND flash and 3D Xpoint memory for use in Intel and Micron end-user products, and Micron Technology-branded NAND flash supply to other SSD manufacturers, is facing a big hurdle with its QLC NAND flash manufacturing ramp-up, which if not checked, could influence SSD prices globally. The company is apparently seeing dangerously low yields of less than 50 percent for its 3D QLC NAND flash memory. This effectively makes its QLC NAND pricier (in terms of $/GB) than current-generation 3D TLC NAND.

The first victim of low yields of 3D QLC NAND flash is Intel's SSD 660p series, a mainstream NVMe SSD that brought 1 TB of storage under the $200-mark. Sources within IMFT tell Tweaktown that the company is seeing 48% yields in its 64-layer QLC NAND flash wafers (i.e. 52% of the wafer is unfit for further production). In contrast, 64-layer 3D TLC yields are above 90% (margin/incomplete dies are excluded from these figures). What's worse, the source predicts that the conditions may never get better with this generation.

Quote:According to DigiTimes, 3D TLC yields have only gotten off the ground in the beginning of this year - right around the time companies were rolling out their 3D QLC designs. And if TLC took longer than expected to achieve respectable yields, it seems that QLC memory will take even longer - we already knew that the Intel-Micron venture on QLC was facing less than 50% yields, but DigiTimes has now extended this struggle to what seems to be the entire NAND manufacturing industry (Samsung Electronics, SK Hynix, Toshiba/ Western Digital and Micron Technology/Intel). The result? Expected price fluctuations in the beginning of 2019, as predicted production volume fails to meet both projected and actual demand, with 3D TLC supplies having to cope with increased market demands.

Quote:The sun has risen a little early on Crucial's new P1 series SSD equipped with Micron's quad-level cell (QLC) NAND. Crucial hasn't announced the drive, but it's appeared on Amazon Japan as the "Crucial SSD M.2 500 GB P1 series Type 2280 PCIe 3.0 × 4 NVMe 5 year warranty CT 500 P 1 SSD 8 JP." More information has also become available via Crucial's UK website, but there still hasn't been an announcement for the U.S. market.

Quote:As traditional silicon scaling has stopped paying dividends for flash-storage density, NAND makers have packed more and more bits into their flash packages by layering more and more sheets of flash memory on top of one another. Today, SK Hynix is joining the elite 96-layer club with its "CTF-based 4D NAND flash." While that "4D" descriptor is purely fluff, the company is in fact producing many-layered charge-trap NAND (as opposed to the floating-gate tech favored by Intel and Micron). Those 96-layer stacks allow the company to pack 512 Gb (64 GB) of TLC flash into a single memory chip.
...
Hynix plans to introduce 1-TB SSDs using its 4D flash and in-house controllers later this year. Looking to 2019, the company plans to introduce UFS 3.0 mobile SSDs with this flash in the first half of the year, and enterprise SSDs with this technology in the second half. Hynix also plans to introduce 1-Tb flash packages using 96-layer technology and products using QLC NAND next year.

Quote:In October Samsung took the stage on its Tech Day event and announced its SSD roadmap. One of the key elements of that roadmap was the project to launch QLC (quad-level cell) SSDs, and now we've got more information on these products. Several European online retailers -French and Italian- have already listed the new Samsung 860 QVO units, which means their official availability is near us. The new SSD drives will feature the conventional 2.5-inch format with SATA interface, but the naming scheme changes from EVO or Pro to the new QVO, which stands for "Quality and Value Optimized SSD". Performance goes up to 550/520 MB per second for sequential read/write, and apparently these SSDs will feature 96,000 IOPS read and 89,000 IOPS write. There will be at least three variants: 1 TB (MZ-76Q1T0BW), 2 TB (MZ-76Q2T0BW) and 4 TB (MZ-76Q4T0BW), with prices of 117.50 euros, 225,96 euros and 451,93 euros (VAT excluded) according to those online retailers. Even with taxes included 19% would make 140, 270 and 540 euros), these are cheaper prices that the ones we can find on the Evo family (160, 380 and 850 euros at those storage capacities), for example. Some of these online shops mention December 2018 as the ETA.

Quote:The 860 QVO delivers acceptable performance for a SATA drive. It isn't amazingly fast, but 860 QVO will perform similarly to other TLC SSDs in most real-world applications. However, it lands on the low end of the spectrum during some tasks. For instance, the Intelligent TurboWrite cache eventually fills during large file transfers, which slows performance. But that shouldn't be an issue for the target market. After the initial operating system clone to the drive or the first transfer of your games library, the Intelligent TurboWrite cache should be large enough for normal daily use (especially if you buy a high-capacity model). If you need more performance, the obvious choice is to move to a faster model, like the EVO or the PRO.
...
The Samsung QVO lands at higher prices than we expected, especially given that the market is rapidly changing and SSD pricing is plummeting for the holidays. Just a few weeks ago 1TB SATA based SSDs were $160+ at a minimum, but now many popular options, including Samsung’s own 860 EVO, are priced under $130, and we expect prices to continue to plunge over the coming months. Honestly, why pay more for less? That’s not to say you get that much less out of the 860 QVO; it’s just that the 860 QVO's pricing is out of line with competing products.

It's hard to recommend the 860 QVO if it isn't retailing at least 10% below the other mainstream TLC SSDs. We also have to consider that the Intel 660p is just $165, so for $15 you get much more performance paired with a modern NVMe interface. As with all new releases, we expect some price corrections as the series matures, so it's possible the pricing situation could improve quickly.

Quote:The 860 QVO lands in the company of Toshiba's RC100, a tiny sliver of NVMe storage that attempted to scrape by without a dedicated DRAM cache. Even with all its advantages, the 860 QVO couldn't turn in a better showing. We can make excuses for the QLC drive's performance in IOMeter, since those tests are designed to expose drives' weakness by filling them to capacity and subjecting them to extreme conditions. Applications with similar access patterns are not at all suitable to be run on budget consumer drives like the 860 QVO. But there are no excuses for the drive's showing in our real-world RoboBench tests. Despite leaving the 860 QVO plenty of room to let its Intelligent TurboWrite caching work its magic, the drive still just couldn't deliver.

The drive did do well in our boot and load tests, so maybe it could find its niche there—but only if the price is right. In the graph below, the most compelling position is toward the upper left corner, where the price per gigabyte is low and performance is high.
...
In summary, the 860 QVO can boot Windows and load applications with SSD swiftness, but it seems unable to marshal its caching tricks well enough to yield consistently snappy file transfer speeds. At the price Samsung wants for it, the QVO is a tough pill to swallow. Instead, we'd suggest buying cheap TLC drives while the holiday sales bonanzas are yet running. Give QLC some more time to bake before you open up that particular oven.

Quote:There's two ways to interpret these results. One is that a PCIe drive has no right to be so slow, QLC is the worst, you can pry TLC from my cold, dead hands, etc. (substitute TLC for MLC or SLC depending on your personal level of snobbery). The other one is that the drive is just fine, and those lunatics at TR put this poor gumstick through tests no sane person would ask of a hundred-dollar consumer SSD. The drive was great in real-world tests, and that's all most people shopping client SSDs should care about.

And both perspectives are correct. The IOMeter tests we run ask far more of lowly consumer drives than their makers expect them to be used for. QLC may be verifiably worse for performance than TLC or MLC, but does it matter if it's still good enough for 99% of use cases? The P1's pseudo-SLC caching works great for ordinary workloads. So while the raw speeds of the QLC are slower than those from less dense NAND implementations, the end user will rarely be subjected to them.

But it's reasonable to expect a lower price in exchange for an empirically weaker drive. To quote a gerbil in the height of his wroth circa 2012: "TLC is a lower-performing product that offers 50% more storage for the same manufacturing cost. Unless I see 50% better cost/GB, we are being short-changed, and the lower performance is just adding insult to the injury." It may be an oversimplification, but change "TLC" to "QLC" and it's hard to argue with that sentiment. So let's see where this QLC drive lands in today's market. In the graph below, the most compelling position is toward the upper left corner, where the price per gigabyte is low and performance is high.

Even though we're well past the Black Friday and Cyber Monday feeding frenzies, SSD prices are still deliciously low, and that proves to be the P1's undoing. The drive is selling at its suggested price of $110 at Newegg, but that's just not good enough. Just as there were superior alternatives to the 860 QVO, there are superior alternatives to Crucial's P1. If you're looking to buy a PCIe SSD, Adata's SX8200 and Samsung's 970 EVO are both going for about the same 22 cents per gigabyte that the Crucial demands. And both of those drives put up speed figures that the P1 simply can't touch.
...
All things considered, Crucial's P1 actually gave us a nice surprise on the performance front. It couldn't withstand our crushing barrage of IOMeter tests, but not every drive needs to be that robust. Its caching tricks allowed it to maintain an appreciable lead over TLC SATA drives in real-world file transfers, and it delivered the snappy boot and load speeds that all SSDs enjoy. Its problem is purely one of pricing. Until drives with QLC flash start getting hit with the discount stick, our advice is the same as it was a few days ago: if you need an SSD and can afford a QLC drive, you should just take that money and buy a faster TLC bit bucket.

Quote:SK Hynix Inc., announced today that it has delivered samples of new 1Tb (Terabit) QLC (Quadruple Level Cell) product to major SSD (Solid State Drive) Controller companies. The Company applied its own QLC technology to its world's first 96-Layer "CTF (Charge Trap Flash) based 4D (Four-Dimensional) NAND Flash (or 4D NAND)." SK Hynix intends to expand its NAND portfolio to 96-layer-based 1Tb QLC products in time for the QLC market opening and strengthen its responsiveness to the next-generation high-density memory market.

Quote:Intel today at a press event in South Korea announced their plans for future product launches in the memory spaces. Optane is the name of the carriage Intel is pulling here - there's no novelty about that - and the company will be pushing a second generation release of Optane enterprise SSDs and Optane DC Persistent Memory modules. Most interesting for us down-to-earth PC enthusiasts, though - the market launch of 144 Layer QLC NAND in 2020, which should bring even lower pricing to NAND-based devices. Later, the company also plans to launch 144 layer TLC NAND solutions.

Quote:Last week, Intel's memory and storage group produced Intel QLC 3D NAND solid-state drive (SSD) number 10 million based upon the QLC NAND die built in Dalian, China. Production began in late 2018, and this milestone establishes QLC (quadruple-level cell memory) as a mainstream technology for high-capacity drives.

Quote:Going along with the current trend in NAND, Intel is stacking its flash memory to 144 layers. The technology is codenamed Keystone Harbor, and SSDs based on it are slated to debut in 2021.

However, Intel's plans for its SSDs don't end at high layer counts. According to Block & Files, Intel is also developing an SSD technology that stores five bits per cell, known as PLC. Few details are available at this time.

Intel's method of stacking memory cells up to 144 layers isn't unlike the work of Samsung's work in stacked NAND. Samsung is reportedly accelerating the development of 160-layer NAND using a "Double-Stack" technique. Chinese company YMTC is also building 128-layer QLC NAND using its own "Xtacking 2.0" architecture.

Quote:Just over two years ago, Micron became the first company to launch commercial products based on QLC (quad-level cell) technology. Intel followed with consumer-facing products later in 2018. These drives can hold up to four bits of data per cell, but they pay for the additional capacity with lower performance and lower longevity; QLC drives do not support nearly as many program/erase cycles as TLC drives, which are not as robust or as fast as MLC drives. The drop in SSD prices last year may have hampered QLC’s adoption — with all types of NAND storage cheap, QLC didn’t stand out as clearly — but manufacturers are still moving to adopt the technology. Recent leaks on Amazon indicate Samsung is planning to leap into the fray, with SSDs at 1TB and 8TB capacities, the latter priced at just $900.
...
At the highest end of the market, HDDs will continue to have advantages; nobody is going to drop a reasonably-priced 16TB SSD in the near future, while a 16TB HDD will set you back just $386 according to Newegg.

Pushing QLC capacities higher will help the drives in two specific ways. First, it allows for larger SLC caches. Most QLC drives use an SLC cache to improve performance, with the size of the cache being determined by the base capacity of the drive. Intel’s 512GB 660p SSD has an SLC cache of between 6GB – 76GB depending on how full the drive is, while the 2TB version has an SLC cache of 24-280GB. Samsung will provision its own caching system differently than Intel, but larger drive capacities typically mean larger caches, which means less chance of hitting low performance, even with large file writes.

Second, it indirectly improves drive longevity by increasing the total amount of reserve NAND on the drive and spreading the writes out to a larger number of blocks. SSDs use a technique known as wear-leveling to reduce wear-and-tear on specific areas of the drive. The larger your drive is, the more capacity you have to perform wear-leveling in the first place.

In short, larger QLC drives won’t be free from the weaknesses of the medium, but end-users will be less likely to encounter those weaknesses in day-to-day usage. At the same time, QLC drives — while slower than other SSDs — are much faster than spinning discs, as this THG review of the 660p shows:
...
With Intel now working on PLC (penta-level cell, or 5-bit) NAND, the push is on to shove magnetic disks out of the consumer market altogether, even with respect to capacity. The difference between the $154 price of an 8TB spinning disk and the $900 hypothetical price for an 8TB Samsung QLC drive is smaller, statistically, than the capacity and price differences between SSDs and HDDs that have already been overcome.
...
I don’t think hard drives are going to die — enterprise sales remain strong on the whole — but I do think we’ll see fewer and fewer of them in consumer products as time goes by, even at the low-end. That’s not a bad thing.

While I harbor a sneaking admiration for the way HDD manufacturers have repeatedly driven higher capacities over the decade, the single best upgrade to breathe new life into an older PC is an SSD swap. The only thing that even comes close is a RAM upgrade, and only in scenarios where the machine has far less RAM for a given task than it ought to have. In that context, you’re arguably correcting a configuration flaw as opposed to strictly “upgrading” a PC.

Quote:If we look at our synthetic testing, the Samsung 870 QVO does very well. It achieves excellent results, especially when it comes to random and sequential writes. Random and sequential reads are a little bit lower than some competing drives, but the differences are small. If the review ended here, I'd definitely give the Samsung 870 QVO a recommendation, but well, it does not.

Our real-life test results paint a completely different picture of the SSD. Here, the 870 QVO can barely keep up. On average it is the second-slowest SSD we ever tested, only the aging Crucial BX500 is slower. Let's look at individual results. Game startup times are barely affected in absolute terms, maybe a second slower, not worth worrying about. In relative terms the 870 QVO is still trailing the pack. In apps, things are different. For example, in WinRAR extraction, which sends a huge amount of small writes to the SSD, the Samsung drive is the slowest by a huge margin. Surprisingly, even read-heavy scenarios are affected; "Text Search in Files" has the Samsung drive as the third-slowest. Another important data point is ISO file copy, which simply copies a large file and has this SSD trail 20% behind the second-slowest drive. Steam game decryption has you in-game three minutes later than your friends with NVMe. Performance results are also very unpredictable. We do ten runs of our real-life test suite, with a pause between each test and a reboot after every full run. The 870 QVO is the only drive in our test group that showed huge swings in the data, which means it will subjectively appear to suddenly slow down by a lot, and performance will recover only after a while.

The underlying reason for these results is the QLC technology in the 870 QVO. Writing to QLC NAND cells directly is slow, so all modern SSDs operate some of their capacity in single-cell SLC mode, which is much faster to write to. The SLC cache will soak up bursts of write activity, and the drive flushes data from SLC to QLC in the background when idle. Our "Write Intensive Usage" tests do exactly that, and the results are shocking. While the drive starts out great, at 500 MB/s, write rates drop fairly quickly, to below 100 MB/s, which is slower than most hard drives! Some people even have Internet faster than that. This could mean that when downloading from Steam or EGS, or copying files over the network, you'll be bottlenecked by the SSD, not by the connection. Samsung also chose a relative small SLC cache of only 42 GB, which is only a fraction of what we're typically seeing on QLC SSDs.

It almost looks as though the 870 QVO was engineered specifically to perform well in synthetic benchmarks to get decent review scores. This could also be useful for support cases of people wondering about the low performance, as "CrystalDiskMark is showing normal results, RMA denied". I also have to question the legality of "up to 530 MB/s". Yeah sure, "up to".. problem is the real number is 80% (!) lower than the claim. Maybe it's legal, but it's just wrong to mislead customers like that.

If the price is right, these performance numbers could still be justified. Some people might consider the 870 QVO as a large-capacity HDD replacement. The most affordable 1 TB SATA SSDs retail for around $90; our favorites, the Crucial MX500 and Seagate 120, are $115, and last-generation's Samsung 860 QVO is $110. That's probably why Samsung is asking $115 for the 870 QVO. That price is completely unrealistic because pretty much every other SSD you can buy will be faster than the 870 QVO. Most budget SSDs on the market use TLC flash, which operates much faster than the QLC on the 870 QVO. For $115 you can even find decent M.2 NVMe drives that are in a completely different realm performance-wise. Of course, not everybody has a PC with M.2 NVMe support, or your slots are already filled with other drives. I still can't recommend the 870 QVO for that scenario; rather, buy a decent TLC SSD, it'll perform better. If you feel you don't need a lot of performance and can wait a bit, sure, buy the cheapest SSD you can find, you'll save $30 and it will still perform better than the Samsung 870 QVO.