Samsung 970 EVO Plus NVME performance drop

[Grumpy ECMO]

I recently reran the PassMark 10.0 tests on a recently built computer and the results for the subject hard drive have dropped considerably. I don't remember the actual test numbers (should have kept a record) but it originally came in at around 98% versus the rest of the world, now it's at 82%, and showing very poorly against the same model in other computers. The temperature during testing never exceeds 110 F and the latest Samsung NVME drivers are installed.

I was thinking it may be BIOS related as ASUS has updated several times (ROG Crosshair Hero VIII, with a Ryzen 7 3800X). I don't see a way to install an older BIOS to check out that theory.

Any suggestions?

[David (PassMark)]

Maybe your machine stayed the same, but the world got faster?
Hard to really know without seeing the actual numbers.
Many SSD drives also get a bit slower as they fill up.

[Grumpy ECMO]

Is there any way to find the original baseline that was submitted when I built the computer?

[David (PassMark)]

If you have all the machine specs and the rough date (and the specs aren't super common) we can probably find it.

[Grumpy ECMO]

Thanks, I appreciate the help. It would have been around February 18 or 19, 2020.

Specs:
Processors Information
-------------------------------------------------------------------------

Manufacturer AuthenticAMD
Name AMD Ryzen 7 3800X
Codename Matisse
Specification AMD Ryzen 7 3800X 8-Core Processor
Package Socket AM4 (1331)
CPUID F.1.0
Extended CPUID 17.71
Core Stepping MTS-B0

Memory SPD
-------------------------------------------------------------------------

DIMM # 1
SMBus address 0x52
Memory type DDR4
Module format UDIMM
Module Manufacturer(ID) Corsair (7F7F9E0000000000000000)
SDRAM Manufacturer (ID) Samsung (CE00000000000000000000)
Size 16384 MBytes
Max bandwidth DDR4-2132 (1066 MHz)
Part number CMW32GX4M2C3200C16

DIMM # 2
SMBus address 0x53
Memory type DDR4
Module format UDIMM
Module Manufacturer(ID) Corsair (7F7F9E0000000000000000)
SDRAM Manufacturer (ID) Samsung (CE00000000000000000000)
Size 16384 MBytes
Max bandwidth DDR4-2132 (1066 MHz)
Part number CMW32GX4M2C3200C16

Mainboard Model ROG CROSSHAIR VIII HERO (0x00000655 - 0x8EAED7F0)

SMBIOS Version 3.2

DMI BIOS
vendor American Megatrends Inc.
version 2103
date 06/19/2020
ROM size 16384 KB

Storage
-------------------------------------------------------------------------

Drive 0
Name Samsung SSD 860 QVO 2TB
Revision RVQ02B6Q
Serial S4PHNF0M602716P
Capacity 1863.0 GB
Type Fixed, SSD


Drive 1
Name Samsung SSD 970 EVO Plus 1TB
Serial 2B2QEXM7414045T
Capacity 931.5 GB
Type Fixed, SSD

Display Adapters
-------------------------------------------------------------------------

Display adapter 0
ID 0x2160003
Name AMD Radeon RX 5700 XT
Board Manufacturer GIGABYTE Technology
Codename Navi 10
Technology 7 nm
Memory size 8176 MB
Memory type GDDR6

Monitor 0
Model U32J59x (Samsung)
ID SAM0F33
Size 31.5 inches
Max Resolution 3840 x 2160 @ 30 Hz


Software
-------------------------------------------------------------------------

Windows Version Microsoft Windows 10 (10.0) Professional 64-bit (Build 18363)
DirectX Version 12.0

[David (PassMark)]

We have 112 results from machine with that CPU and video card. The earliest submission was in March 2020. So none match.
Search for a match on the CPU, video card and HDD also returned zero matches.

[Grumpy ECMO]

The February 2020 test was when I upgraded the video card. So that run may not have been recorded. The initial build was September 16, 17 or 18 2019. Same specs except for an XFX RX580 video card. Would you please check for that baseline. BTW - decided to purchase the PassMark 10 software yesterday, now a registered user.

Looking at the data more closely, the issue seems to be sequential write performance. Over 3 successive runs the values range from 1610 down to 233. I've ordered a replacement drive, just to be safe.

[David (PassMark)]

You can use the Advanced search option in the baseline menu to search all the submitted results.

Might not be a hardware issue. Could be some other software running in the background causing trouble.

[Grumpy ECMO]

Looks like my baseline was never uploaded. I'll keep investigating and will let you know what I find.

[Grumpy ECMO]

A minor update. I had Process Explorer running when I did the last Disc Mark test. Lo and behold, MalwareBytes suddenly came to the forefront as consuming CPU cycles. Exiting that program and rerunning Disc Mark gave results more in line with other testers. I'm still experimenting with other programs that may impact results. My interest has been peaked, will keep you advised.

[David (PassMark)]

Bloatware strikes again

[Grumpy ECMO]

I replaced the NVME drive with a new one, same make and model. Used Samsung data migration software to populate the new drive. Disc Mark score jumped to 28221 from 20381.