Aurora-R6 : Hard-Lockup and crash while gaming (SOLVED)

Tesla,

You've helped this community a lot, sorry to see no one's offered you some back.

I've also been bitten by that darn boot process issue. Similar to as others have posted, I had grief when trying to migrate over to a 960Pro, the boot order changed on me, and that caused a few blood pressure points and 15 minutes of wasted F12 commands... Luckily after changing that back, it hasn't seemed to caused any more grief, I assume same for you.

As for the actual crashes, no doubt you're knowledgeable enough to troubleshoot the basics yourself, but I thought i'd see how this is progressing at least.

From the crash dumps, Windows 10 has that "reliability monitor" tool, does that parse out any info that helps diagnose this issue?

Otherwise, I might suggest Blue screen of death (STOP error) information in dump files.  Nirsoft is a decently well known company in this space, and their software is usually pretty helpful for running through the logs.

I wouldn't put it past the SSD being trouble though. These last 6 months of the 40+ computers I've ordered for work, 20 of them have had PM9X1 failures/corruptions. If it wasn't late on a Friday, I'd remember the bloody Event ID it keeps tripping, but it's pretty obvious and has something to do with the SSD controller.

I've also never really had great luck with that ePSA diag. It really only triggers for full/obvious hardware failures. Event viewer catches so much more for me...

Good luck!

There was an bios update. Have you installed  ?

Yes, I did (as I posted). It didn't seem to help.
 
Hopefully, BIOS v1.0.3 at least fixed the Windows Boot Manager - Boot Priorities bug that CometusCrimsonion and I have experienced. I haven't seen that happen again lately. Sure would be nice if we could get a proper list of fixes, features, and issues about new Dell BIOSes when they are published.

CometusCrimsonion wrote:

 ePSA diag. It really only triggers for full/obvious hardware failures. Event viewer catches so much more for me...

 

That is my experience as well. Same with UEFI Diags on other machines. It's important, but really seems more like just the first hardware test or hurdle you need to pass.

Thanks for posting. Yes, I spent thousands of dollars on this machine and I'm getting a little worried about now.

Yes, I can use all the help I can get. I've been updating drivers, uninstalling Dell stuff, and turning processes off pretty much daily trying to get it stable enough to play a game for a couple of hours without crashing. Productivity work and everything else seems fine. It just seems to be when machine is pushed hard.

For Critical Events in Reliability Monitor I was about to report that it only shows "Windows wasn't properly shut-down" (after each lockup) but further closer inspection now reveals other Criticals like "Hardware Error" . This warrants further analysis for sure. I embarrassed to say I don't use this tool enough when I should. It's definitely easier to find the errors than with Event Viewer. 

Yes, I tried BlueScreenView (and another similar one). Best I can tell, Crash Dumps are working but they never get written during this event. It's like the machine hard-locks before the OS can even BSOD or write a dump file. 

CometusCrimsonion wrote:

... 

I wouldn't put it past the SSD being trouble though. These last 6 months of the 40+ computers I've ordered for work, 20 of them have had PM9X1 failures/corruptions. If it wasn't late on a Friday, I'd remember the bloody Event ID it keeps tripping, but it's pretty obvious and has something to do with the SSD controller.

 

Well, that’s dis-heartening to hear. Overall, I’ve had excellent performance from my Samsung SSDs over the years. Maybe I’ve just been lucky. Also, I’m pretty sure Apple still uses custom Samsung SSDs in their machines. I think the Controller in 960/961 is all new and maybe even a new memory type.

I thought I had done good in the Aurora-R6 SSD-lottery with my 512gb pm961 (at least the price was right). I read the posts about NVMe SSD setup challenges with these Dells. I thought it would be nice to finally order a machine that doesn’t have to be upgraded as soon as I open the box (Now-days, I refuse to use a machine without a bootable SSD).

I see a lot of SSD firmware upgrades (but no Samsung) for Aurora-R5 but none for the Aurora-R6 yet. Not sure if that is good or bad. If there was a required firmware upgrade for pm961, you would think Dell would follow-suit and not only have access to it, but to actually post it.

Personally, I’m more leery of the (tricked out) Intel-RST (Rapid Storage Technology) driver-suite . While historically unstable (not what you want in a HDD/SSD driver) maybe Intel finally has it right with this Dell pre-installed v15.2.2.1030. Honestly, I’m still a little unsure how all the NVMe stuff works, so I’m not confident enough to try to remove Intel-RST (if it even can be). I’ve got some other M.2-SATA machines here, but this is my first with M.2-NVMe.

Please let me know if you hear anything more about this subject. Thanks again for taking the time to help me.

Intel RST does not support MVMe drives. you can uninstall the driver.

NVme drives have their own drivers. see here:

Tool & Software | Download | Samsung V-NAND SSD 

the only problem is, you have an OEM SSd and i don't know if the Samsung driver for consumer SSDs works. if not, try to find another one or, you know, contact Dell tech-support.

as for hard lockup: try with your own advice. remove the SSD, use another hard drive, install windows and see what happens.

So, you think Intel-RST is only installed for the secondary 1tb spinner HDD? I did notice the BIOS is set to RAID-Mode, but again, like you said ... that's just for SATA drives? I used to think that but recently came across some conflicting info.

Well, either Windows-10 supports NVMe natively, or you need a "F6 Driver", right? It must exist or it would not be working now. 

Yeah, I was waiting for someone to write my "words of wisdom" back to me ... "you should just Nuke-and-Pave that thing Tesla and get back to us."

Haha, Yes, your system is the br0kenz, I'll send you an address where you can return it.

This reddit thread I keep trying to post everywhere, cause it's decently useful information IMHO. Basically, to get the information out there as much as possible, there's a few take-a-ways

1) it really doesn't matter, based solely on performance mindset, if your computer is AHCI vs RAID in the bios. If you're going for simple windows usage, just use whichever freggin works.

2) AHCI mode will be the only mode where you can install Samsungs drivers - RAID mode uses generic stuff

3) I THINK (as in, my HYPOTHESIS, but don't really know) the biggest issue in general for people is secure boot ON with AHCI, causing no one to be able to F6 their way into installing drivers...

But basically don't worry too much about it, just find whichever works (and theoretically the generic windows should be "easiest" under RAID)

After 30 years in this industry, when I was going through this process for my new-out-of-box Samsung 960 Pro, I felt absolutely clueless and immensely frustrated. In the end, after 45 minutes of various stages of failed installs/boot, I just cloned the original dell M.2 and called it a day. Luckily I had a laptop with 2x m.2 slots. This is waaaay too hard of a process for 2017 & current tech...

If it helps, i followed the insane set of hoops to jump through for downloading and extracting the samsung drivers here:

I don't entirely remember what happened (round 9 of 30 attempts or some such) but I was able to get win 10 installed, it just wouldn't boot. (this might be the Secure boot:on issue, but TBH I really don't remember)

I do at least remember getting slightly excited that win10 install would at least see my drive finally... Ultimately, Aomei backupper clone of the original OEM drive is how I got my computer to work tho.

Side note:While we're troubleshooting the overall issue, and this is SSD is an absolute time drain, do you have just a "traditional" Sata SSD or yikes-Spindle-based hard drive to try on, if even for a 24 hour basis just to confirm the motherboard/PSU/Memory side of things is in tact?

Both of those links are interesting reads (I also PDFed them both).

Yeah, Macrium Reflect guys figured out similar, but I think slightly different and shorter. It should also be noted that Macrium Reflect bootable-flash works fine on my system as it was shipped from Dell (even in UEFI/SecureBoot mode).

Your cloning solution was ... well, like you said ... whatever works.

As for my problem, I don't think it has anything to do with the SSD. It works fine for heavy productivity work, stress tests, SSD benchmarks, etc. Never had a "disk error" or even a need to "soft repair" anything ... I do a disk-check on both drives after a lockup (it never finds anything wrong). Oh, yeah, it can do this:

is the problem still not fixed?

is Intel RST still installed? could you please open the Device Manager and then go to Storage Controllers. what do you see there?

Yes, Intel-RST (Rapid Storage Technology) v15.2.2.1030 is still installed (I haven't messed with it). Additionally, it does not seem to be causing a problem with the spinning 1gb HDD or interfering with the PCIe/NVMe SSD.

It has two listed:

Intel Chipset SATA RAID Controller
Microsoft Storage Spaces Controller

I updated my NVidia.com drivers again. This time with a Custom Clean-Install of latest WHQL driver and with only the base/minimum "components".

It seems to be running a little better now. However, I haven't had much time to game recently, so I think its still to early to tell. Please recall that it only hard-locked while gaming (and it was fairly intermittent). I am getting less Critical-Events in Reliability Monitor report lately.

Thanks for checking-in on me.

Thats strange. no NVMe Controller listed. instead you have Intel Chipset SATA RAID Controller. not even Intel NVMe RAID or something like that.

it could be the NVIDIA Driver causing problems. every now and then Nvidia releases faulty Drivers. last year it was pretty much hit or miss. they released one hotfix after another....

come back if the System still crushes.

Yes, it seems the Nvidia v375.63 driver that Dell Pre-installed is one of those particularly bad sets. It was just the luck-of-the-draw that it ended up in their current Aurora-R6 BootDisk pre-load Image.

They would cause an actual Windows Hardware Failure. If it happened during a Full-Screen Direct-X game, the computer would hard-lockup (no BSOD, no crash-dump files). It's been a long time since I've seen something like this, that wasn't truly malfunctioning hardware that needed to be replaced.

I'm reminded that this Aurora-R6 is one of my only truly dual-GPU systems (because of its Intel i7-7700k). All my other systems (even my laptops) are either one-or-the-other (Intel/AMD Internal GPU or dedicated external Nvidia/AMD GPU). I'm wondering if this has something to do with it.

The only way I know to insure a system with an intermittent problem is "Fixed" is ... hours of stress-testing without another single re-occurrence of the failure. That's what I'm doing now.

Carbon Based Lifeform wrote:

it could be the NVIDIA Driver causing problems. every now and then Nvidia releases faulty Drivers. last year it was pretty much hit or miss. they released one hotfix after another....