AHCI vs RAID - Performance Test

I always used to think that AHCI would give better performance in a single SSD setup then RAID. Some companies like Dell always have the BIOS setup in RAID mode even if the user has a single drive setup/not using any RAID Arrays.

So I grabbed my wife's Dell Vostro 5470 laptop which has a Crucial MX100 512GB SSD and installed windows in RAID mode then formatted and installed it in AHCI and here are the results. Very interesting, RAID was slightly better and even had lower latency!



@tilleroftheearth

 

Very interesting, it depends on the RAID controller specs as well and how much buffer it has, not all RAID controller will give you the same performance.

 

Just in case someone wants to switch between them, this is the steps, and it's very safe:

Switch Windows 10 from RAID/IDE to AHCI
Some systems will have the Windows operating system installed using RAID drivers including the Intel Rapid Storage Technology. SSD drives typically perform better using AHCI drivers. There is in fact a way to switch operation from either IDE / RAID to AHCI within Windows 10 without having to reinstall. Here are the steps:

1. Click the Start Button and type cmd
2. Right-click the result and select Run as administrator
3. Type this command and press ENTER: bcdedit /set {current} safeboot minimal (ALT: bcdedit /set safeboot minimal)
4. Restart the computer and enter BIOS Setup
5. Change the SATA Operation mode to AHCI from either IDE or RAID
6. Save changes and exit Setup and Windows will automatically boot to Safe Mode.
7. Right-click the Windows Start Menu once more. Choose Command Prompt (Admin).
8. Type this command and press ENTER: bcdedit /deletevalue {current} safeboot (ALT: bcdedit /deletevalue safeboot)
9. Reboot once more and Windows will automatically start with AHCI drivers enabled.

 

The difference is pretty minor, though. I doubt anyone will feel that in daily use.

 

17% difference in 4K QD1 writes, thats pretty significant if you ask me...

 

Thing is, this benchmark is not very reliable so this percentage might be just indication of margin of error.
I just ran it several times, one after another. It produced different results for this particular section. Here's an example:

 

I always do that, saves the hassle of Clean install for a minor change. Not everyone is patient as Ultra male. Atleast, his wife understands and doesn't complain.

 

I haven't found any mere benchmark that is run for a few minutes/hours and a single 'score' to be reliable either.

That is why I test hardware and software with real-world workouts for the platforms I use and over an extended period (a week or more).

And why I don't chase 'scores' and BM's any longer either. I not only know what is best on my platforms and usage scenarios, but I also don't need to run benchmarks to validate it either.

How can I know? Because every platform change, upgrade or purchase is fully compared against the previous, fully tested platform(s) I am already running. If there is little to no benefit, that new (potential) platform stays out of my workloads and is never considered again.

 

I prefer consistent performance rather than short burst IO/CPU/GPU huge scores in benching. Higher reliability simply overtakes more bench scores.

 

You can have both, bruh And avoid put hot running NVMe drives into thin and flimsy or in notebooks with not adequate cooling.

 

Yes I know but nowadays people really want size 0 laptops and a small bulge or thick laptop is a major eye sore.