Wow, looks good to me … the motherboard, not the performance of the SSD.
I’m getting a sequential reading speed of 3524 MB/s and a write speed of 3225 MB/s with a Samsung 970 EVO Plus 1 TB SSD on a very simple Dell Optiplex 7070 SFF computer with a i7-9700 Intel CPU and 32 GB RAM-memory.
No, it’s not a WD SSD, but that is just a personal choice after many small problems in the past with WD SSD drives.
Samsung … better performance, better software, less troubles.
With x4 PCI Express® 2.0 bandwidth, M.2 supports up to 20Gbps data-transfer speeds.
At 1700mb / s you pretty much maxed out the the PCIE 2.0 x4 lanes. (I suppose the missing 300mb / s are used by data framing overhead* (headers etc.).)
You might as well have bought yourself a lower spec’ed NVMe SSD. Ah well, might you ever upgrade to a new MB then your SSD is up to it.
In the specs I do see a:
1 x PCIe 3.0/2.0 x16 (x16 mode)
If it’s not in use (by a discrete video card), you might add a 4 lanes M.2 to PCIe converter card to put your NVMe drive in order to enjoy its full speed.
* To determine the actual amount of data that can be transferred, the encoding technique must be understood. PCIe Gen 3.0 and PCIe Gen 4.0 use a 128b/130b encoding technique. Older generations such as PCIe Gen 2.0 use 8b/10b encoding. This encoding technique (PCie 4.0) transforms 128-bit data into 130-bit line code. This allows for reasonable clock recovery (which is the process of extracting timing information from a data stream) and ensures alignment of the datastream.Source
Exactly.
I took it out of my computer and put it on my Dell I5, which has PCIe 3.0.
Using Crystal the SSD reached the reading speed of 3500mb without any problem.
I was unlucky enough to buy the only motherboard (TUF H310M-PLUS) which is PCIe 2.0 haha. It is a recent card, but they are rare on the market, costing the same price as other 3.0.
I believe that this problem that I had will be increasingly rare.
Just found this thread when doing a search for UserBenchmark. Like you @Hermman, I am getting top performance on Crystal benchmark and around 1800 on UserBenchmark.
How can there be that much difference? Which one is right?