Reduced NVME drive performance following OS re-imaging

compgeek89

Limp Gawd
Joined
Nov 30, 2018
Messages
176
So, I have a dell XPS 9510 laptop (i7-11800H, 32GB DDR4-3200, RTX 3050, Win10 Enterprise, for reference). From the factory it came with a 512GB Micron 2300 PCIe 3.0x4 SSD. However, the laptop actually supports PCIe 4.0, so I got a Sabrent Rocket 4 Plus 1TB to replace the OEM Micron drive.

I had re-imaged the 9510 from scratch using a flash drive and manually installing all the Dell drivers from the XPS support page prior to the drive swap, so it wasn't running the factory image. Initially, I used the Sabrent Acronis software to simply clone the Micron over to the Sabrent. However, I have a, perhaps unfounded, skepticism of exactly what goes on in a clone operation and figured it wouldn't hurt to start from scratch, and perhaps it would even squeeze a little more performance out of the drive to do a fresh install on it.

Anyway, following the re-imaging of the Sabrent R4P, Crystal DiskMark actually shows significantly worse numbers on a number of benchmarks than it did when it was cloned from the Micron (though some are comparable, and one is actually higher). I am trying to come up with an explanation or troubleshooting process, but I haven't been able to diagnose it. My first thought is drivers, but as far as I know I followed the same process for both imaging events, so the drivers should be the same... but I don't have the old image, so I can't do a line by line check. Still, I had the same list of Dell drivers, so I have a hard time imagining there could be a difference. The only real potential reason for a difference would the drivers that Windows Update pulls down once I give it an internet connection, if any.

Micron 2300 Benchmarks:

16ff54a-CrystalDiskMark_20211226145658_XPS_OEM_SSD.png



Sabrent R4P after Acronis Cloning:

ark_20211226145658_XPS_Sabrent_Rocket_4_Plus_Clone.png

Sabrent R4P after re-imaging:
k_20220112155621_XPS_Sabrent_Rocket_4_Plus_Scratch.png

So, my write speeds on the SEQ tests are consistently slower by up to 12%, and then my RND4K Q32T16 reads are less than 50% of the clone numbers (not even matching the original Micron numbers), while the writes are around 2/3. These numbers are all consistent, so it's not a fluke. I've tried rebooting and ensuring there's not activity in the background. Besides, the SEQ read numbers are not showing any issues. As far as I know, nothing has changed. I didn't mess with the BIOS or anything like that. I just have no good explanation, and it's bugging the crap out of me. If anyone can provide any insight, I'd be very appreciative. Thanks!
 

daglesj

Supreme [H]ardness
Joined
May 7, 2005
Messages
5,609
Can you tell any difference?

Nope, didn't think so.

Some days you get good benches and some days you get bad.
 

pendragon1

Extremely [H]
Joined
Oct 7, 2000
Messages
41,769
are you reimaging, cloning or reinstalling? they are not the same. maybe theres a formatting difference between the cloned oem drive and whatever you are meaning by "re-imaged".
 

mwroobel

Supreme [H]ardness
Joined
Jul 24, 2008
Messages
5,093
Benchmarking an empty (minimal garbage collection going on in the background) SSD which has no other processes in contention will often bench a few percentage points off per run. Doing the test on your active OS drive can yield wildly different numbers.
 

cpufrost

Limp Gawd
Joined
Sep 28, 2020
Messages
403
Try AS-SSD, it will show if your alignment is bad. That will cause lower benchmarks.
 
Top