Unsolved
8 Professor
•
7.1K Posts
•
29.7K Points
0
1251
April 21st, 2021 15:00
Aurora R10, retail SSD upgrade
I am looking for some feedback from other users who have an Aurora R10 with a Ryzen 3000 and installed a second SSD using a retail brand, preferably a WD Digital Blue 1TB.
To give some feedback on what I am looking for:
- My Aurora R10 came factory default with the SATA ports configured in raid mode in the BIOS.
- My Aurora R10 also came with an NVMe 512GB SSD factory installed. (Samsung OEM, M2 slot)
- I installed a second SSD, a 1TB WD Blue retail using a regular SSD port.
The problem is twofold:
1. If you leave the bios in raid mode, you cannot read SMART attributes from the WD. You can still read them from the NVMe, because that's an M2 port and not part of the raid mode. It's normal not be able to read SMART attributes for individual drives of a raid mode.
2. To fix issue 1, I set the bios to AHCI mode instead of raid mode since it makes no sense to use raid mode on a single drive configuration anyways. That allows me to read the SMART attributes, and they are all good. However, that gives me the problem below.
When I run in AHCI mode, once a day or so I get a timeout on the WD SSD, followed by a retired block on the device. These are recorded in the event log. Disk scan shows no issues, SMART shows no issues and the WD manufacturers dashboard shows no issue. There's also no bad or retired blocks on the SSD. (It's a smart attribute)
When I run in raid mode, the issue goes away and I never get a timeout or a retired block.
At this point I am thinking this is either a compatibility issue with between the SATA controller and the SSD, or a firmware issue with the SSD. I am leaning towards the controller, because it goes away in raid mode but is persistent in AHCI mode. I am also using the latest AMD drivers for the chipset. It did the same with the factory shipset drivers, whatever version was shipped with it.
I already played with timeouts and some other things, no luck so far.
So if someone else has seen this before and maybe has some input, that would be great.


Tesla1856
10 Wizard
•
18K Posts
•
71.5K Points
1
April 21st, 2021 20:00
That's a fairly specific situation you are looking for, but hopefully ... someone will answer.
In the meantime, lets see if this helps.
In general, Dell ships it's systems in RAID-Mode.
When I'm running in RAID-Mode, I have Intel-RST installed in Windows-10.
When running in AHCI-Mode (usually takes a clean-install to truly get rid of it) ... I never install or use Intel-RST. The bundled Microsoft SATA-drivers works just fine.
And, AFAIK, Intel-RST is really only for SATA-drives. It's not really used for any installed NVMe-SSD drives.
More info is here:
https://www.dell.com/community/Alienware-Desktops/Aurora-R7-M-2-NVMe-bootable-options/td-p/6073037
Now, all that said ... My Aurora-R6 is still in RAID-Mode (because that is how it came from Dell) and I'm using Intel-RST. The reason I mention it is because Crystal-DiskInfo has no problems reading SMART from the C-drive M.2-NVMe SSD or the D-drive SATA-HDD.
Additionally, (on any machine) I've never seen AHCI/RAID setting in BIOS have a bearing on whether or not Crystal-DiskInfo (or Passmark DiskCheckup) can read SMART from ANY connected drive (NVMe or SATA).
If you must use Intel-RST, be sure you are running the latest version.
However, if you were to set it to AHCI and clean-install Windows-10, (and never install Intel-RST) ... both drives should work just fine. If either drive is still acting up:
- Be sure it is running it's latest firmware
- Be sure it has been "DiskPart Cleaned" . Initialized as GPT and reformatted as NTFS.
If either drive is still acting-up (and you are sure it's not a motherboard or cable problem) you might have a bad drive. Most drives makers have Diagnostic programs.
Finally, remember that any SSD should really be Over-Provisioned by at least 5% (and 10% is better). That way, if it needs to "swap out a block" there is plenty available at end of drive.
Vanadiel
8 Professor
•
7.1K Posts
•
29.7K Points
0
April 22nd, 2021 03:00
I have an R10, so it's AMD chipset based. No Intel RST drivers on it for sure. It's using the default Microsoft drivers for the storage devices and I am using the manufacturers software to read the smart attributes. (It's called WD dash board). That software only reads SMART attributes in AHCI mode, it cannot read them in RAID mode. All the other functions of that software work in both modes. Just not the SMART function.
Because it's only this drive that has this behavior, and only when using AHCI mode, I am thinking it's drive firmware related or an oddball issue between the SSD controller and the drive.
It's not really then end of the world since it has no effect on the drive or it's performance, but I am one of those people who likes to get to the bottom of it.
It's also interesting because this specific drive has some controversy around it. When I purchased it retail in the WD Blue packaging and the WD Blue manufacturers label, the drive actually was identified as a Sandisk and not a WD. (WD is the OEM provider for SANDISK). I was not the only one with that issue. Eventually WD changed the software to make it identify it was a WD Blue, but to this day it's still not 100% correct because the ID string is still Sandisk specific and some other utilities still insist it's a Sandisk.
So I found it very interesting this drive has this behavior. I should add it never had any issues in my previous system under AHCI, just on this system.
Tesla1856
10 Wizard
•
18K Posts
•
71.5K Points
0
April 23rd, 2021 22:00
Aurora-R10 is AMD and no Intel or Intel-RST. Gotcha.
Well, if you re-read my post ... there are still some good tips and suggestions ... focused on Windows and how it handles SATA and NVMe drives.
Also, Crystal Disk-Info is very popular and my go-to SMART reading program on ANY Windows system.
Yeah, I stopped buying SanDisk anything after I had some of their "high-end" USB flash-drives go bad in a spectacularly weird way.
Stopped buying WD-drives after some new drives that failed early, some older drives that didn't make it 3 years, and my brother and clients lost some as well. Got to be a hassle with all the returns and RMAs. Anyway, now-days only really use spinning-platter drives in USB-3.0 externals for local backups and (or course) Synology NAS.
Kingston SSDs used to be good budget pick back in the days of SandForce controllers. Also Intels were OK. Now-days I just buy Samsung SSD (NVMe or SATA) ... EVOs for budget, Pros for mission-critical work and/or systems.
Vanadiel
8 Professor
•
7.1K Posts
•
29.7K Points
0
April 24th, 2021 11:00
Looks like I have it fixed, have not encountered this issue for some days now.
I ended up disabling Dynamic ticks in windows 10, and so far so good.
I now get the regular warning that a reset was send to the drive, but no issues with retired blocks right after that reset command.
Going to monitor for another week, but looks like it's under control.