Lexar hits 7 GB/s with new M.2 PCIe 4.0 SDD

[erek]

Hehe, Awesome performance!

"The read and write speed of SSDs are exceeding far beyond what we need in real life. Recently you've already seen M.2 SSDs perform at the 4GB/s and 5GB/s marker, well, no you may add to that Lexar, who showed an example running with read performance of 7 GB/sec."

https://www.guru3d.com/news-story/lexar-hits-7-gbs-with-new-m-2-pcie-4-sdd.html

 

[drescherjm]

I will be happier if the 54MB/s number was doubled.

 

[erek]

I will be happier if the 54MB/s number was doubled.

Here's the Phison E12 (Inland Premium) non-PCIe 4.0

 

[DejaWiz]

Remember when all the manufacturers of the first MoBos equipped with PCIe M.2 NVMe slots (Z97, iirc) used to advertise the theoretical max of 32 Gb/s speeds as the selling point?

NVMe drives don't even come close to that 32 Gb/s (4000 Mbyte/s)...and the newest NVMe drives (PCIe 4.0 used on MoBos like the AMD X570 chipset) aren't even approaching that PCIe 3.0 cap.


Lo and behold, they are still using the theoretical max as a selling point:

64 Gb/s = 8000 MB/s, which is 2.3x what this new Lexar is shown performing at. Using PC Mark 8, most NVMe PCIe 4 and Optane drives tested come in between 700 - 1700 MB/s:

source: http://www.thessdreview.com/our-reviews/nvme/corsair-mp600-pcie-4-0-nvme-ssd-review-1tb/4/







Only the Intel Optane 900 series and Samsung Z-NAND consistently score 1400-1600 MB/s...on a PCIe 3.0 bus interface, at that!

source: https://www.tomshardware.com/news/intel-optane-vs-samsung-z-nand-ssd,38987.html



Until NAND itself and the controllers used on these drives catches up to the near max limits of the bus interface, then I will always take the advertised performance figures as a hunk of salt being served.

 

[Rockenrooster]

Remember when all the manufacturers of the first MoBos equipped with PCIe M.2 NVMe slots (Z97, iirc) used to advertise the theoretical max of 32 Gb/s speeds as the selling point?

NVMe drives don't even come close to that 32 Gb/s (4000 Mbyte/s)...and the newest NVMe drives (PCIe 4.0 used on MoBos like the AMD X570 chipset) aren't even approaching that PCIe 3.0 cap.


Lo and behold, they are still using the theoretical max as a selling point:


View attachment 207668






64 Gb/s = 8000 MB/s, which is 2.3x what this new Lexar is shown performing at. Using PC Mark 8, most NVMe PCIe 4 and Optane drives tested come in between 700 - 1700 MB/s:


View attachment 207669

source: http://www.thessdreview.com/our-reviews/nvme/corsair-mp600-pcie-4-0-nvme-ssd-review-1tb/4/







Only the Intel Optane 900 series and Samsung Z-NAND consistently score 1400-1600 MB/s...on a PCIe 3.0 bus interface, at that!

View attachment 207670

source: https://www.tomshardware.com/news/intel-optane-vs-samsung-z-nand-ssd,38987.html



Until NAND itself and the controllers used on these drives catches up to the near max limits of the bus interface, then I will always take the advertised performance figures as a hunk of salt being served.

I see what you mean (I love me some random read/write speed), but they were saying max sequential speed, not average over a few different tests...
7GB/sec is pretty close to 8GB/sec.

 

[Shoganai]

 

[DanNeely]

I see what you mean (I love me some random read/write speed), but they were saying max sequential speed, not average over a few different tests...
7GB/sec is pretty close to 8GB/sec.

Especially once you factor in protocol overhead. It's kept 3.0 x4 drives to about 3.5GB/sec on paper specs; 4.0 x4 at 7GB/sec is about the most we can expect.

 

[power666]

The thing is that SSD manufactures road the SATA3 wagon for while as both NAND performance and per chip capacity increased. The result wasn't an increase in speed (which was limited by SATA3 speeds anyway) or capacities but rather they cut costs by reducing the number of NAND channels in the design as well as the number of NAND chips per channel. Cause in point is that the first generation of SATA3 SSD's typically had eight channels vs. the common two nowadays with four being for performance or high capacity drives.

High performance NVMe cards in PCIe form factor use different controllers than consumer drives (not the M.2 carriers) tend to have at least 8 channels with a large DRAM buffer which is why they are so much faster. These cards tend to be 8x PCIe 3.0 (with 4.0 coming soon) as they could otherwise saturate a 4x wide link.

In short, it is possible to build NVMe drives that saturate the 4x PCIe link of the M.2 form factor but there little need to do so.