Hello

I installed a new brand M.2 NVMe SSD with a PCI-E adapter. My motherboard is a MSI 970-G43 PLUS.
It boots well with Windows 10. All runs fine.
What drivers I have to install? Since the chipset is AMD, I think the best option is AMD NVMe Drivers but mine is not a X399 chipset....
Any clue?
Thanks

@Fernando

PCI\VEN_126F&DEV_2263&SUBSYS_2263126F&REV_03

This is the info you need? Thanks

Would you please sign the OFA driver 1.5 for windows7 x64？
I only found the signed OFA driver for windows10 in your onedrive.
Native driver(KB2990941) doesn't support SMART while the original OFA driver requires test mode on.

Thanks

Do you mean the detail of an OFA driver indicates that it was compiled by Microsoft so it is unnecessary to be mod+signed?
I guess that the details might be generated by WinDDK as a default value(possibly equals to target OS version).
The nvme driver provided by apacer also shows similar details.

Actually,anyone can change the value by modifying the source code.
Since the file was removed from OFA's website now and I didn't download the source code,I'll take firedisk's opensource driver for example.
http://reboot.pro/topic/8804-firadisk-latest-00130/
sourcecode-->firadisk.rc

The string is completely equal to the detail page of the compiled firadisk.sys.

BEGIN
BLOCK "StringFileInfo"
BEGIN
BLOCK "040904b0"
BEGIN
VALUE "CompanyName", "Karyonix\0"
VALUE "FileDescription", "FiraDisk Virtual Disk Driver\0"
VALUE "FileVersion", PROGRAM_VERSION_STRING "\0"
VALUE "InternalName", "firadisk\0"
VALUE "LegalCopyright", "Copyright 2009-2011 Panot Joonkhiaw\0"
VALUE "OriginalFilename", "firadisk.sys\0"
VALUE "ProductName", "FiraDisk\0"
VALUE "ProductVersion", PROGRAM_VERSION_STRING "\0"
END
END
BLOCK "VarFileInfo"
BEGIN
VALUE "Translation", 0x409, 1200
END
END

if these lines were deleted,the detail of the compiled sys file might contains the information like an OFA driver.

If you are not wiling to make the OFA driver mod+signed.
Could you give me some advice about what I can do so that the driver would be loaded by Windows kernel without test mode?
I feel that disabling DSE and Patchguard may make the system unstable,but win7's native nvme driver stornvme.sys won't shows any SMART info including the temperature.
I have to monitor the temerature since the ssd is installed on a laptop where nothing could be installed to protect the ssd from overheating.

Thanks for your explanation.

Currently,I have no choice but to use OFA driver and enable test mode since western digital doesn't offer their specific driver.

I plan to demand a refund and buy a PM981 or SM961 which has official win7 driver. o(╥﹏╥)o

Hello Dieter/Fernando:

First, I just wanted to thank you for both past and future help. I last encountered your work when I needed BIOS updates for older Athlon 64 machines, and you had them when the industry had passed them by. I believe I'm still running that update on a server down in the basement. So, thanks for that. And for the education in this and related threads.

Now I have picked up a new XPS 13 Dell laptop -- a 9360 -- the one generation old form factor, but with the new 8550 cpu. New form factor made the old model too cheap to pass up. This has the Toshiba NVMe Drive at 256gb, running the expected (now that I've been reading here) microsoft controller driver. I've updated to the lastest BIOS, updated Intel RST, and updated the drive firmware -- all without incident, and all without making much difference in performance (presumably this is mostly about write caching).

I also used the bcdedit procedure to swap to AHCI mode, and after using it again to force windows into a mode where it would allow me to install unsigned drivers (which is how it saw both your update of the OCZ drivers for the Toshiba and the OpenFabrics drivers -- not sure why; Windows as nanny state perhaps?). The results were good with both, with the OCZ driver having a bit of an advantage in the 4k low Queue depth department -- performance about doubled on the standard crystal diskmark 4KiB Q1T1 test segment. Have some results at home I can post if folks are interested. Presently running RD400 OCZ driver.

But, the drive is 256Gig, and I want to keep the machine a while, so space is an issue. Now I'm seeing very good prices on the HP EX920 1tb. Maybe too good to pass up. It appears that there's no proprietary driver for these . . . at least no obvious one. Which, now that I've worked along to this point with the Toshiba, leads to:

First question, is there and alternate driver for the Silicon Motion SM2262 8-channel controller in the HP? It does look like Intel's 760p uses the same controller, so we might have an analog of the Toshiba/OCZ situation that would allow testing alternative drivers???

Second question. It looks like you removed the OpenFabrics drivers, but they appeared to work for me. Particularly if there's no specific driver, are they likely to be back, so that we can enable the write cache on the HP drive? Even if the OFA team has some issues at present (I think I saw a note that you removed the drivers for the time being) the fact that they might come later would still be useful to know if they're likely to work with this drive.

If not, I may go for this anyway. Need the space, and the speed looks pretty good too (size really seems to help performance on the NVMe SSDs).

Sale ends shortly (of course!), so I may have to decide before I hear back. But any thoughts would be appreciated.

Best Regards,
Mike

Annandale, VA (US)

Dieter:

Most helpful. Thanks. I'll look a bit, but even with the Microsoft Driver, this drive is so much bigger/faster and so cheap that I may have to try it. Even without the improved drivers, it looks to be much faster -- which is really just a bonus, as space on the drive in the primary issue.

If I were to keep poking around, can you describe the "hardware IDs" you have in mind? Presumably that means that "Silicon Motion SM2262" is insufficient -- perhaps a family of controllers? So are we down to reading the silk screened IDs on the chip . . . and what do we need? If I knew what to look for, this might be an interesting quick project.

Either way, thanks very much for the quick response and for all you help.

Mike

Ah . . . THOSE hardware IDs. OK. Having read reviews during a late lunch, it looks as if Intel and HP have their own custom firmware, so this may not work. But we'll see.

Now I just need to see if the double sided board will fit in the XPS. Thanks much for the help. Even if I do this, it will help with the Toshiba drive, which will soon find a new home, I'm sure.

Mike