Though I really feel like I’m doing every part proper with the testbed setup, the truth that all of those drives hit a wall at round 7160 MB/s suggests a limitation someplace. Maybe that is the sensible restrict of a PCIe 4.0 x4 drive on this platform, although the usual theoretically permits for nearer to eight GB/s (~7877 MB/s), there may be overhead to contemplate.
We transfer on from the real-world simulation of the PCMark 10 checks above, to the strictly artificial CrystalDiskMark resutls under. As traditional, I’ve run the checks with a single thread and at 4 completely different queue depths – 1, 2, 4, and eight.
We start with sequential outcomes: