Sometimes my RAID 0 SATA SSD array is extremely slow. Then, maybe 15min later or after rebooting, it works fine again.
This is how it looks when it's slow:
This is the normal speed:
My relevant specs are:
Ryzen 7 1800x
Asus Crosshair VI Hero (RAID on X370 chipset)
2x Radeon R7 240GB SSDs (in RAID 0)
Windows 10 Pro N 1709
AMD AHCI RAID driver 9.2.0.0070
RAID 0 with Read + Write Back Cache
Your Windows build looks to be outdated, please update to the latest build and setup Raid.
Does system performance seem sluggish, or is it just this one test showing low results?
It's a really bad idea to RAID0 your boot drive, even though SSDs are far more reliable than HDDs it's still a bad idea. Also, you shouldn't always take benchmark results at face value, especially if you only run them once. If they're to be believed your sequential performance is better than mine, despite me having Samsung 850 EVOs which are -much- faster than the R7 SSDs you use.
Comparison: 512GB 960 Pro:
I've been trying to update to the newest feature version of Windows, but it keeps failing. It's on the most recent version of 1709 though (KB4462932, released October 18, 2018).
Maybe I will rebuild the array and reinstall Windows in a few months, if there is no other solution.
The system is sluggish when it happens, I just ran the benchmark to "prove" that the issue is indeed related to the storage.
It's not necessarily related to the drives. It could be corrupt system files, run "sfc /scannow" and "DISM.exe /Online /Cleanup-image /Restorehealth" without quotes at the administrator command prompt or powershell.
I'm completely aware of the (usual) consequences of running RAID 0. I have no data on it that I mind losing, which isn't backed up constantly.
It depends if you're running the SATA version of 850 EVO and if you're running hardware RAID or a software emulation of RAID (like Windows Storage Spaces). It can have various other reasons why results from one system to another might differ.
Generally, sequential speeds of SATA SSDs is often limited by the SATA bandwidth.
According to this review, my SSDs should perform similarly to SATA 850 EVOs:
I get similar results for 5 runs, it seems to be accurate. Also keep in mind that you're running a different version of CrystalDiskMark with the Seq Q1T1 test replaced by a 4KiB Q8T8 test and in a different order.
Yes I'm using the newer version, as you should be. Also no, it's not, they're 30% faster.
Which is why Anandtech came to this conclusion:
These drives were -never- a good idea, costing more than a Samsung Evo drive, yet performing worse. That's why AMD quickly discontinued them, and why https://www.amd.com/en-us/products/solid-state-drives does not exist anymore.