Crosshair VIII Hero X570 RAID 0 question

[jrrog1975]

I have 2 RAID 0 arrays configured on my Crosshair VIII Hero X570, using Corsair MP600 1TB SSDs. The first array is on the motherboard m.2 slots. The second array is on the Asus HYPER M.2 X16 GEN 4 card installed in the 2nd PCIe slot. I assumed the benchmark results on the Asus card would be at least as good or better than the motherboard RAID 0. However, the sequential read speed on the motherboard array is around 8571 MB/s and the Asus Hyper card array is around 5076 MB/s. Why would the motherboard RAID be faster?

[RedSector73]

My best guess is that using the x570 chipset rather than direct to the CPU. If you wanted to test this, could try placing it in the first slot normally for the graphics card and if correct the speed of the second array would become faster.

So, why the motherboard. Because one is controlled by the CPU and the other by the x570 chipset. The second array is all on controlled and guessing x570 as you don't mention placement on the board.

Hope you find this of help.

[jrrog1975]

RedSector73 wrote:
My best guess is that using the x570 chipset rather than direct to the CPU. If you wanted to test this, could try placing it in the first slot normally for the graphics card and if correct the speed of the second array would become faster.

So, why the motherboard. Because one is controlled by the CPU and the other by the x570 chipset. The second array is all on controlled and guessing x570 as you don't mention placement on the board.

Hope you find this of help.

The Asus card is in the 2nd PCIe slot, which is configured for x4 + x4 bifurcation. The GPU in slot 1 is now x8, leaving x8 for the 2 SSDs on the Asus card. Given that the 2nd RAID 0 array (on the Asus card) has 8 lanes direct to the CPU, shouldn't it be faster? Or does the bifurcation introduce some sort of latency?

[RedSector73]

https://cdn.wccftech.com/wp-content/uploads/2019/05/AMD-X570-Chipset-Details-and-Specs_1-1480x832.pn...

https://cdn.wccftech.com/wp-content/uploads/2019/05/AMD-X570-Chipset-Details-and-Specs_2-1480x832.pn...

jrrog1975 wrote:
The Asus card is in the 2nd PCIe slot, which is configured for x4 + x4 bifurcation. The GPU in slot 1 is now x8, leaving x8 for the 2 SSDs on the Asus card. Given that the 2nd RAID 0 array (on the Asus card) has 8 lanes direct to the CPU, shouldn't it be faster? Or does the bifurcation introduce some sort of latency?

I would think the motherboard will have better RAID unless you are using an enterprise RAID controller with it own dedicated RAM. What I was trying to say is that natively your board is using CPU for one NVMe and X570 chipset for the other in RAID.

The controller you are using is supplying the RAID and was concerned you might by putting that through the X570 controller rather than through PCIE directly to CPU. I assume you mean you are using the second PCIE which would be the one used for a second graphics card and not the second PCIE after the first (which would be controlled by x570 and would/could impose latency).

I read from this that the RAID controller itself maybe the weak point, if you are connected to the CPU PCIE Lanes.