New M15 2070 max-q Video_TDR_Failure

Hi,

At long last a time has come to upgrade from the good old 15 R2, so I took advantage of a sale an ordered a M15 2070maxq. Only a few days late - fine. The machine would blue screen a few times on first use possibly with an external 4k monitor connected via HDMI it seemed. Everything else seems as advertised, the machine is blazing fast compared to AW 15 R2.

Still optimistic at that point, I updated bios from 1.3.3 to 1.5.2 and installed all the latest drivers from Dell support site. Now it blue screens like clockwork (Video_TDR_Failure in an Nvidia driver) with external monitor attached. Tried the latest Nvidia drivers, no joy.

Has anyone seen this? Any advice, or have I been had by Dell again (they had pushed a 4GB 980M rather than the 8GB ordered before - at least that machine was stable and usable!).

Many thanks...

 

What triggers the blue screen when you plug in the external display?

 

Good question. It doesn't happen immediately after plugging the external monitor in. Usually (but not always) the screen would go black briefly at some point, recover, then the machine would freeze after a while. Come to think of it, the way I remember it the freezes could have been happening after a mouse click. I have a Logitech G403 connected via a KVM and a Lenovo USB-C hub. Guess I can try unplugging all that to see if it makes any difference if I just use laptop's keyboard and pad.

BTW PSA and memcheck diagnostics pass.

 

Full OS reload. Do not use the recovery partition, download a fresh 1809 installer. Search for win10 media creation tool. The system will possibly be wiped so be sure to back up all important data. If that doesn't work. Warranty RMA with dell.

 

A

Thanks, I guess that's a nuclear option worth trying but just curious what underlying issue might it resolve? Why not use the recovery partition? Any Windows activation issues I could run into that way?

 

Just guessing that there is something wrong with your os and using the built in recovery may not fix the issue. First thing I do with any new laptop is nuke the factory os. Fresh clean OS with updated drivers from Dell and use the Nvidia driver from their website.

 

I've been having endless issues with the factory image myself so I formatted and installed a vanilla copy of Windows 10 then installed the latest drivers as per Dell's recommended order and now everything is perfect.
NBR Windows 10 Clean Installation Guide
Windows 10 Tweaks and Fixes (Index post #1)

 

Thanks guys, will strongly consider this before return / RMA. Amazing walk-through Ultra!

 

Dude, it's like you're sitting in my head. I was tearing my head apart with the Dell Factory Image, I even loaded it back again but the issue I was facing was I wasn't even able to install the Alienware Command Center. Then another format, I was able to install it but the FX tab was not selectable (like greyed out) so I couldn't change the lights. Then the next time I launched it, it gave me a message that I need to connect an Alienware keyboard before I can use it, WTF! I was about to return the laptop not for hardware issues but it really drove me nuts each time formatting, configuring all my staff and copying back all my data, then when it comes to installing the drivers, something here or there happened!

I will create a Drivers thread soon for the Alienware m15 as I did steal some of the newer chipset drivers from the Area-51m and they work perfectly.

Stay tuned bro

 

@etern4l

To be more specific than the above Dell installation order, rather than just saying install Chipset Drivers, here is my perfect driver install order:

 

Cool, looking forward to it. My outstanding question is how to avoid fking up Windows activation in the process. Do I need to deactivate the outgoing build before wiping it?

 

@Ultra Male Awesome, thank you. Looking at your list reminded me that Alienware Update worked once, reported no updates when about 12 were needed, then stopped working altogether. Good old Dell

 

Negative, firstly, the windows product key is injected into your BIOS so when you're installing Windows, the Windows installer will not even ask you for a key as it will detect the product key in your motherboard and install the appropriate edition accordingly (ie. Home or Pro)

Secondly, the moment you get online, give it a few minutes and it will auto activate

Finally, when installing the LAN/WLAN Drivers, I ensured to select CUSTOM and unchecked the box which says Killer Software. Don't need any of that Killer garbage near my system. Just install the drivers that way.

PS: When the system asks you to reboot, do that, do not take any shortcuts

 

That's why I have no number next to it because I consider it optional and don't install it anymore because it sucks big time! Just like the Dell SupportAssist does!

 

Thanks for all your help. Just to update: the problem has nothing to do with Lenovo USB hub (caused minor trouble with 15R2), clearly 100% to do with connecting an external monitor via HDMI. Otherwise the GPU works as expected with the laptop panel.
Bluescreens within minutes of connecting external monitor like clockwork. Worse yet, one of the crashes occurred during a cumulative Windows update and now the build is corrupted, the update fails to complete, can't create recovery media due to missing files etc - have no choice but to reinstall.

Now, Windows Media Creator tool doesn't work for some reason (doesn't allow me to choose Win 10 Pro), but the Dell Recovery media tool does - so will have to use that. There are two options: SupportAssist and pure Windows 10 Pro. Need to read up on which one is better to use (obv know your recommendation Ultra.. ).

Noticed one more issue though: there are these weird lags when typing, like it buffers up what I type and spits it all up after a few secs. Fairly rare though.

One more thing, I'm wondering whether it would be possible to roll the 1.5.2 firmware back as they mention some RTX "optimisations" and crashes seem to have intensified after upgrading from 1.3.3. I imagine it won't work without some serious hacking though.

 

BIOS rollback is disabled by default on m15/17/a51M and older 17r5 as well.
Do you use any Anti virus?

 

@Vasudev McAfee that comes bundled by Dell. Will replace with Bit Defender.

In general, many thanks for the advice to reinstall Windows! Haven't tested too extensively but apart from a single TDR blue screen (without an external monitor this time) before drivers were installed, it's looking stable so far on latest Dell drivers.

 

Nuke them both and stick with Win defender until your return period. BD blocks or causes BSOD for win update when it classifies as Trojans and crash Windows in unbootable state!

 

Bad news. So, to play it safe I restored to factory build first. Again, getting one or two BSODs right off the bat. After installing all the dell drivers I was able to play around with the machine on the external screen, however, I'm still getting the same BSOD like clockwork if I try to run Cinebench R15. It also crashed on attempt to run Dropbox installer. I did not get to try the latest Nvidia drivers because I popped in Dell's manual Win 10 Pro USB and it wiped off the HDD seemingly without warning.. .

That "manual" build has way less Dell stuff on it but still crashes even with all Dell drivers installed the same way. I still cannot create a pure Win 10 Pro install USB using the Microsoft tool.

The last thing to try I guess is newer Nvidia drivers and failing that I'm returning. Having bought a second Alienware from Dell on sale with issues, I suspect the sale means they try to push out returned units. Mine came with some protective foil peeling off suggesting it's a returned lemon.

I like the machine when it works though. If only there was some hope of this being just a RTX Max-Q driver issue. Could someone please check whether their m15 with RTX Max-Q crashes on Cinebench R15 with an external monitor connected directly to laptop's HDMI?

 

When you get BSOD with CLOCK related it seems that RTC/BCLK might be faulty and it was indeed an Orange or a Lemon you got. Return it and look at other options.

 

Why do you think it's clock related?

Yeah, I was looking at Gigabyte Aero. Similar price and specs, just no GA port and no clue about Gigabyte support - apparently it's even worse than Dell's.

 

Difficult to speculate w/o seeing the BSOD in person.
Gigabyte, I don't know. For now, I've seen +ve reviews from many people including @OWNORDISOWN and @B0B of all trades youtubers. i7 8750H and RTX 2060M has enough horsepower and has proper cooling even with factory paste.

 

Exactly. The dump message is as per the title plus there is reference to a nvidia dll. I might have a go at analysing the dumps. The fact that it almost always crashes in Cinebench but no issues whatsoever so far in DirectX games suggests something is wrong with OpenGL.

The reason I went with 2070 is 8MB and 240W PSU.

 

Hmm, so I actually managed to find a workaround, which is to change Power Management Mode in Nvidia Control Panel from Optimal Power to Prefer Maximum Performance. This is in the latest 419.67 driver. I can't think of a downside of the solution besides maybe a bit of a hit to battery life when using 3D apps on the RTX on battery which seems of no importance to me. When plugged in the GPU clock still sits at 1080Mhz when idle, no change vs "Optimal Power".

Will do some battery life testing.

 

You will get only max of 2-4hrs battery on dGPU unless you use traypwrdwn app that locks clocks to lowest level possible.
I'd recommend returning and order m15 2060 model. Constant TDRs is bad news.

 

I did not have to force dGPU, the setting just says, if you happen to use dGPU, don't use this Optimal Power plan, which is not even listed as an option in the help text below, prefer Max Peformance (for completeness setting Adaptive Power also leads to crashes) . Again, the setting doesn't actually force max GPU clock so I'm not worried about battery life yet, given that I don't expect to see much dGPU use on the go anyway. Seems like some Max-Q teething issue. Not a single bluescreen since I changed the setting and everything works perfectly and so much better than in my old 15R2. Hibernate works, sleep doesn't crash, no keyboard repeat rate bug, boots in what seems like 3s.
Again, not really keen on 180W power supply and the experience of Dell Hybrid Power and want the 8GB so would rather stick with 2070.

 

To be safe though, I will report this to Nvidia and see what they have to say for themselves.

 

I doubt it, they are busy selling RTX models ASAP.
I think its a hardware issue since Optimal mode is default during driver install.

 

You may be right, however, with the workaround in place there are not further issues (apart from a brief monitor flicker when closing and opening the lid) and performance is perfect. And, man, Shadow of the Tomb Rider at 4K HDR@60FPS Ultra and DLSS. It's unbelievable, like a gigantic pre-rendered cutscene. DLSS is not a marketing gimmick. I'm sure you'd return it, I'm not strong enough

 

Unfortunately, it seems you were right @Vasudev, while my workaround held water when plugged in, the laptop is less stable on battery (one random bluescreen and XCOM2 failing every time). People speculate an NVIDIA factory suffered a contamination issue affecting RTX chips.

https://forums.geforce.com/default/...-rtx-20-series/explanation-about-dying-rtx/1/

ANother theory is that's it's Micron GDDR6 (as in my unit), supposedly RTXs using Samsung memory fare better. Some anecdotal evidence regarding RTX desktop card failure rates reaching 35%!

https://forums.geforce.com/default/...post-personal-first-hand-experiences-here/12/

If true, how can Dell release products based on that junk?!? Perhaps that's why they aren't offering RTX in AW 15 & 17?

The only question left now is whether to RMA or return and look at something else, perhaps even stick to Pascal. Such a waste of time. FU Nvidia&Dell. Perhaps it's time to switch to a Mac lol.

 

Nvidia fixed that VRAM issues via vBIOS and most vendors have them deployed.
I did tell you that is signs of dead GPU or mobos that will die the next day after warranty expires.

 

@Vasudev explain this: I turned cooling up to max and all the issues went away, even in Optimal Power mode (there never have been issues once the fans are at max mid game). Could it be the Dell Thermal Toothpaste syndrome? Worth finding out.

 

I'm unsure but its a hunch. It might be because of inferior cooling and max cooling trick allows more airflow that prevents overheating or baking of VRAM, chokes, caps etc...
@Papusan does have all the info here Alienware m15 / i9-8950HK Review by Ultra Male

 

Yes, I'm hoping repaste will sort it out as suggested in the m15 user thread. I think there is also an issue with the "balanced" fan profile, e. g. CPU temps hit 100C in balanced mode, but just 90C in high perf. Idle temps are 10C lower in max perf mode.

 

You say just 90C. Have you tested with the new released benchmark test Cinebench R20 - MAXON | 3D FOR THE REAL WORLD

Then compare your results with equal setup... http://forum.notebookreview.com/thr...iew-by-ultra-male.828258/page-2#post-10892840

 

Yes, I tested with CB20. There is variability with the results - 2684, temps around 90 on the first run, 2550, temps up to 98 on second run (8750H, I don't think there is a significant case for putting a hotter CPU in that machine as clearly we are struggling to keep 8750H cool enough in the first place).

 

If you looked at the link I posted you can see same i7-8750H as yours performed above 3000cb. Or better say +-15% faster. He tested both the i9-8950Hk and the locked down i7-8750H. I expect even higher temp for yours if you was able to run it faster. It's clearly that your don't perform as adverticed.

 

Missed that, sorry. It is a surprising result. I wonder if his laptop was repasted and/or undervolted. My CB15 results are in line with results posted in mainstream reviews.

 

Yeah, it was re-pasted and Undervolted. But the owner was still not happy with it (performance and temp). He did the only correct... RMA for refund.

 

The theory that a repaste might fix this took a blow today. While the laptop has been stable on power with improved fan profile, it fell into pieces on battery. XCOM2 crashed every time, once with random small rectangles artifacting, Shadow of the Tomb Raider froze the PC for good.

Returning. No idea what to get instead.

 

Wise decision. So far I can suggest Legion y740 with 2060 mRTX

 

OK, so I contacted Premium/Pro support on Thursday and they arranged for a technician to come over and replace the motherboard on Monday. They said that's the quickest option and should it fail they will replace or refund. A friendly and experienced guy called and came over soon after, replaced the mobo, problem solved, no issues. 7450 TimeSpy on stock paste undervolted. Happy in the end, however, prob best to always upgrade warranty/support (local super-responsive people, all done by email - no wasting time on the phone) and just contact them straight away (although plenty of good advice here, such as regarding clean install of Windows, repasting etc. - many thanks). My previous poor Dell support experience was with basic support I think.

 

Well, let's hope they fix it.

 

TLDR my friend? It's all done, no probs. Premium support.

 

Mobo replaced?

 

Yep.

 

Check TDP values for GPU during 4k gameplay, it seems Dell adjusted to lower values for all GPUs after BIOS update.

 

Nope, still 90W while most competing laptops offer only 80W on the 2070 Max-Q.

 

Try using AIDA64 GPU and CPU tests and you'll see that dell set TDP according to temperatures. At 80W you might see a dip in GPU performance or sudden spikes to 90C on GPU for no reason.

 

Don't have AIDA64 and frankly don't want to spend any more time on this. The machine is working very well indeed now (still on stock paste). The suggested "TDP throttling" doesn't apply to my machine because GPU temps are far below 80C.



That was taken in TimeSpy (razer blade 15 advanced stock manages 6350 in this test according to notebookcheck)



CASE CLOSED.