@Fernando Tyvm buddy...

Zitat

---

Question: How did you get the "Intel AHCI & RAID drivers v17.8.0.1065" installed? Do you have created an Intel RAID array?

---

i just go to device manager, right click update driver ide/atapi control then browser to local folder.
after that rebooted PC go to device manager - storage controllers - standard nvme then force have disk to mod+signed OFA NVMe.

will try Samsung NVMe driver v3.3.0.1910 later then, thanks again for help...

Correction by Fernando: Meanwhile I have corrected the exact version of the driver within my source post. I meant the recently released Samsung NVMe driver v3.3.0.2003

My M.2 PCie OS drive is showing itself as "Ejectable" under "Safely Remove Hardware."

I just found your forum. I updated my MS NVMe Storage Controller from the OEM Windows 10 to your OFA. (Win10 64 1909.) Am running a WD Blue M.2 PCie.
The M.2 used to be Drive 0. Then I tried making a mirror of two HD's. (By accident I ended up making a spanned set instead.) After that, Windows moved the M.2 to Drive 4.
So I disabled all the drives in the BIOS (EUFI) except for the M.2 boot drive. That moved the M.2 back to Drive 0. But now I always have a "Safely Remove" icon in the Notification area of m Taskbar.

I assume this is, in part, why Windows doesn't like to hibernate with the OFA driver? (Because Windows kind of thinks that the OS drive is not internal?)

Danke sehr for the quick reply.

Sorry to have wasted your time with a known "bug."

I saw your tests with the Samsung driver, but I wasn't sure that it would play with a non-Samsung NVMe. (Just now I see that you said it would on your drive page. But I didn't even bother reading anything about the Samsung driver, because I didn't think it would play with my WD.)

I will plan to try it... later. (I am currently attempting to recover files from my failed attempt at mirroring drives. And I foolishly deleted my only backup of that drive thinking I could make a new one of the spanned volume.)

Here are my results with the MS driver and the 3 Samsung ones: 2.0.8.1611 mod, 3.3.0.2003 mod, 3.3.0.2003 pure:


MS


2.0.8.1611 mod


3.3.0.2003 mod


3.3.0.2003 pure


Basically all the 3 Samsung drivers are the same, and the default MS one is inferior in writes for the 4K parts.
And just ignore the higher 4K QD4 in the 3.3.0.2003 pure read test, it happened only then, a fluke...

sorry how to download nvme drivers please?

Hello, a search has brought me to this great article, may be someone has an idea how I fix my issue. Coincidently after installing a new GraphicCard (I assume has nothing to do with the problem, other than triggering a major driver update) my 2 Sabrent Rocket 4.0 now perform really bad (below 3%tile bad) in small file performance - large sequential files are still fast. My Older Samsung 960Pro still performes blasingly fast. I assume the Sabrent drives somehow switched to a slow driver - but how do I identify what driver they are running on given the Samsung Driver does reside on my system happily to support the 960Pro ? Attached the StorageController section from Device manager on whats shown there. Thanks ahead of time to anyone who can help.
Cheers

Hi,
Thanks for the fast response - some quick info on the system: Ryzen 3900X on Gigabyte Xtreme X570 with 16GB Memory
The lane Bandwith should not be the problem - both Rockets are direct to CPU - one in primary M2 one on extension card in PCIE Slot 2 - the way the bord is documented that should drive the primary slot for the GPU to 8X - the secondary also to 8X with only 4X used by the Rocket and finally the M2A having its own 4x lanes - the now much faster Samsung is in M2B Slot which is supported by the Chipset and hence should be the slow one.
The problem with identifying drivers is that only the Smasung shows as an NVMe storage controller, the others show as AMD-RAID Bottom Device and AMD-RAID Controller (storport) - however RAID has never been enabled in BIOS - not for SATA and not for NVME - another strange thing is that in HWINFO the Samsung shows as NVME with a green Checkmark the rockets show as "ATA AMD-RAID Sabrent Rocket 4 " without checkmark.
Something strange has happened here - I can not specifically pin down when it first appeared like this since it was the new Graphics Card triggering running benchmark - system as such behaved normally under light load use the last 2 weeks. But 3 weeks ago on the same set-up (other than GPU) the Rockets had a Userbenchscore setting a new 100%tile and now they are more or less on SATA level other than sequential write is far beyond max SATA bandwith. Somewhat lost what to try.
Cheers & also Happy Easter & Stay Healthy

EDIT:
PS: another strange symptom is that I can not activate Write Cache for the Rockets, while it works fine for the Samsung

All good now - I just forced the recomended Samsung Driver on the Rockets (and the today newly installed 905P 380 GB - which also configured first as an AMD - RAID even though in BIOS it is not even shown under NVMe Drives burt has its own storage class section) - full performance on all drives now - HWInfo also shows the drives as they should be - still no idea why Windows installed these crippling drivers - if it happens again I know where to look. Thanks a lot for this thread here, spared many frustrating days ! Pictures attached Cheers

Hi,
As two the strange install: even the new Optane 905p autoinstalled in windows with the AMD RAID driver without any contribution from my side. So no idea how, but not an active oversight.
I used the Samsung driver linked to in the thread in the Win 10 64bit version.
I have not installed the Rocket toolbox since I am not sure it is intended for the NVMe drives but rather for some of their other storage solutions. I did not find new firmware on their site - but since the optane drive installed the same funny AMD drivers, I rather assume it is something strange the way Gigabyte shows the drives to windows and not a firmware issue of the drives. The Samsung also did need an active install of the right driver but that was more straight forward since Samsung has an Exe install file for download. BTW the board has the latest BIOS and chipset drivers already before the issue first happened.