Possibility of new Samsung SATA SSDs to replace 850 series?

[JTToft]

Does anybody know or have any qualified guesses about when Samsung might introduce new SSDs to replace the 850 EVO and 850 Pro?

The EVO is nearly 1.5 years old and the Pro is coming up on 2 years in the next few months. Considering there were about 1.5 years between the 840 EVO and 850 EVO, and between 1.5 and 2 years between the 840 Pro and 850 Pro, it might not be unreasonable to expect the "860" to come relatively soon.

Any thoughts?

 

[satinsilverem2]

I really don't know what Samsung could do to make them better. More density is all I can think of. The Evo and Pro already saturates the SATA III bus, so I don't think they will get faster but it would be nice to see a 3-4TB version of the drives.

 

[\-V-/]

The Evo and Pro already saturates the SATA III bus, so I don't think they will get faster but it would be nice to see a 3-4TB version of the drives.

Not completely, but pretty close.

 

[JTToft]

I really don't know what Samsung could do to make them better. More density is all I can think of. The Evo and Pro already saturates the SATA III bus, so I don't think they will get faster but it would be nice to see a 3-4TB version of the drives.

- For sequential read and write, sure, nearly. But for random read/write they're both hovering around 100/300 MB/s. Lots of room for improvement in that area.

 

[Mr. Retrofire]

...for random read/write they're both hovering around 100/300 MB/s. Lots of room for improvement in that area.

Not if they use the SATA protocol. Newer SSDs use NVMe via PCIe. See also:

 

[JTToft]

Not if they use the SATA protocol. Newer SSDs use NVMe via PCIe.

- Yes, I'm aware of the benefits of the new interfaces and protocols. But is there any technical reason why SATA should be limited to the around 100 MB/s 4KB random reads and 300 MB/s 4KB random writes that the 850 Pro is able to muster up? From what I can understand, those performance measures are limited far more by the SSD controller than they are by the SATA interface.

Still, who knows what other sorts of optimisations Samsung might be able to come up with? Cost-efficiency seems a great candidate.

 

[treekram]

If you look at: http://www.anandtech.com/show/9702/samsung-950-pro-ssd-review-256gb-512gb/7

The random write benchmark of the PCIe 3.0 x4 NVMe drives (with the exception of the Intel 750) isn't that much faster than the SATA drives. Same with the random read benchmark, except the Intel 750 is slower than all of the Samsung's. Maybe it's the nature of the benchmark and the current technology?

In the "Final Words" section of the review of the Intel 750,
http://www.anandtech.com/show/9090/intel-ssd-750-pcie-ssd-review-nvme-for-the-client/10
they criticize Intel for concentrating on random IO performance to the detriment of sequential performance, so I don't think they're that surprised that random benchmarks fall in the SATA range for the PCIe 3.0 x4 NVMe drives. In other words, I wouldn't hold my breath waiting for significantly better performance in the random benchmarks with SATA drives.

 

[Mr. Retrofire]

...is there any technical reason why SATA should be limited to the around 100 MB/s 4KB random reads and 300 MB/s 4KB random writes that the 850 Pro is able to muster up?...

Yeah, the SATA protocol is not a low latency protocol compared to NVMe.

...
From:
https://en.wikipedia.org/wiki/NVM_Express#Comparison_with_AHCI

(SATA uses AHCI)

The Advanced Host Controller Interface (AHCI) interface comes with the benefit of wide software compatibility, but as a downside does not deliver optimal performance when used with SSDs connected via the PCI Express bus. As a logical interface, AHCI was developed back at the time when the purpose of a host bus adapter (HBA) in a system was to connect the CPU/memory subsystem with a much slower storage subsystem based on rotating magnetic media. As a result, AHCI introduces certain inefficiencies when used with SSD devices, which behave much more like DRAM than like spinning media.[4]

The NVMe device interface has been designed from the ground up, capitalizing on the low latency and parallelism of PCI Express SSDs, and complementing the parallelism of contemporary CPUs, platforms and applications. At a high level, the basic advantages of NVMe over AHCI relate to its ability to exploit parallelism in host hardware and software, manifested by the differences in command queue depths, efficiency of interrupt processing, the number of uncacheable register accesses, etc., resulting in various performance improvements.[4][27]:17–18
...

 

[treekram]

Yeah, the SATA protocol is not a low latency protocol compared to NVMe.

In the Anandtech article for the Samsung PCIe drives I linked in post #7, they had a benchmark random read and the Samsung AHCI PCIe drive was about 15% slower than the slowest Samsung NVMe PCIe drive, so there's more than just the protocol that explain speed differences in the Samsung SSD's (much faster in sequential benchmarks, not so much in random benchmarks).

Perhaps what's happening is that the fast PCIe NVMe drives get a lot of their speed advantage by parallelism and with a 4K random benchmark won't use that unless the block sizes are 1K - I think the block sizes are larger. In the case with Samsung, the NAND chips in the 850 and the PCIe drives are from the same family (maybe different densities) so unless you get faster NAND chips or smaller block sizes, you wont get much faster random read benchmarks?

 

[Mr. Retrofire]

In the Anandtech article for the Samsung PCIe drives I linked in post #7, they had a benchmark random read and the Samsung AHCI PCIe drive was about 15% slower than the slowest Samsung NVMe PCIe drive, so there's more than just the protocol that explain speed differences...

Current hardware and software does not use the full potential of NVMe. That is the reason why you see only a small difference in some benchmarks. I do not think, that they improve a dead legacy protocol like SATA/AHCI, if a replacement is available.

 

[Tobias Funke]

Both 1TB 850 Evo and Pro have been removed from Amazon.co.uk today.

Maybe a coincidence.