QLC Goes To 8TB: Samsung 870 QVO and Sabrent Rocket Q 8TB SSDs Reviewed
by Billy Tallis on December 4, 2020 8:00 AM ESTWhole-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.
The Sabrent Rocket Q takes the strategy of providing the largest practical SLC cache size, which in this case is a whopping 2TB. The Samsung 870 QVO takes the opposite (and less common for QLC drives) approach of limiting the SLC cache to just 78GB, the same as on the 2TB and 4TB models.
Average Throughput for last 16 GB | Overall Average Throughput |
Both drives maintain fairly steady write performance after their caches run out, but the Sabrent Rocket Q's post-cache write speed is twice as high. The post-cache write speed of the Rocket Q is still a bit slower than a TLC SATA drive, and is just a fraction of what's typical for TLC NVMe SSDs.
On paper, Samsung's 92L QLC is capable of a program throughput of 18MB/s per die, and the 8TB 870 QVO has 64 of those dies, for an aggregate theoretical write throughput of over 1GB/s. SLC caching can account for some of the performance loss, but the lack of performance scaling beyond the 2TB model is a controller limitation rather than a NAND limitation. The Rocket Q is affected by a similar limitation, but also benefits from QLC NAND with a considerably higher program throughput of 30MB/s per die.
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.
The Sabrent Rocket Q's random read performance is unusually unsteady at small working set sizes, but levels out at a bit over 8k IOPS for working set sizes of at least 16GB. Reads scattered across the entire drive do show a substantial drop in performance, due to the limited size of the DRAM buffer on this drive.
The Samsung drive has the full 8GB of DRAM and can keep the entire drive's address mapping mapping table in RAM, so its random read performance does not vary with working set size. However, it's clearly slower than the smaller capacities of the 870 QVO; there's some extra overhead in connecting this much flash to a 4-channel controller.
150 Comments
View All Comments
Oxford Guy - Tuesday, December 8, 2020 - link
Samsung doing it isn’t shocking. It’s Samsung after all.Palorim12 - Monday, December 14, 2020 - link
idk if you remember when they did the change from TLC to 3-bit MLC, but I member. It was after all that stuff went down with the 840 EVO. Despite all TLC having this issue across all brands, Samsung was the first to push TLC, so when the slow down issue creeped up, Samsung got the brunt of the complaints, and ppl to this day will use that as a reason as why Samsung "sucks", despite the fact that the issue started creeping up on sandisk and other TLC drives that had entered the market much later after Samsung did. And by the time Samsung figured out the problem and fixed it, all the other manufacturers copied the fix and then really started pushing their own TLC products.And TBH, TLC products since then have been pretty good. I recommend 850 EVOs, and now 860 EVOs to all my friends who want to switch to SSDs but are worried about the price. I've only recommended 2-bit MLC drives to ppl who I know will hit the drive had with writes with the type of work they do.
at_clucks - Wednesday, December 9, 2020 - link
@shabby, other companies have done worse if you ask me, like switching from MLC to TLC mid way through a product's run. Good luck with identifying the exact type of NAND based on decoding a SN without having the decoder ring, especially when the product is still in the store's warehouse.phoenix_rizzen - Monday, December 7, 2020 - link
Switching to a number would really simplify things.1LC
2LC
3LC
4LC
...
But since when has logic been part of marketing?
Billy Tallis - Monday, December 14, 2020 - link
I've already pretty much decided that if we ever get real products that store 5 bits per cell, I won't use any abbreviations that don't include the numeral 5. Stuff like 3bpc, 4bpc, 5bpc would make a lot more sense than current industry conventions.redzo - Tuesday, December 22, 2020 - link
This. It's been a long time since my last post at anand.Consumers have no idea of what they are purchasing. They are basically sheep.
QVO is nice if it is priced right. It should be priced way less.
I just purchased a 3d nand TLC 1TB for less than an intel/crucial/samsung qvo equivalent. This is not right. Manufacturers of NAND flash and product manufacturers are taking advantage of misinformed consumers.
More so. Most products are missing important specs like controller model, dramless or not, or even NAND type. This is just ridiculous.
dontlistentome - Saturday, December 5, 2020 - link
If you want MLC or TLC then buy it - they cost more because they cost more to make. I've just bought a 2TB SSD for the old man - paid the 15% or so extra for TLC over QLC.There's no conspiracy here or evil manufacturers. They do R&D then offer a product and see if consumers buy it. Almost all consumers, even those that claim not to be are driven pricipally by price, hence QLC being populat when the buyer looks at the ticket.
Oxford Guy - Sunday, December 6, 2020 - link
Economy of scale makes your comment fail.DigitalFreak - Monday, December 7, 2020 - link
+1 for your mad rhyming skills.Kangal - Tuesday, December 8, 2020 - link
Lmao.But for real, I thought we would have hit 8TB Sata-SSDs like last year for around USD $650. Instead I'm seeing these still yet to launch proper, and priced around $1,000. It's definitely true the market isn't dominated as much by the consumers, as it is dominated by the actual suppliers.