Alienware 15 R3 shuts down...Resolved: Integrated Intel graphics driver

[Vester]

I have a great-nephew's Alienware 15 R3 that would shutdown after a few minutes of running Windows 11. He bought it almost two years ago after graduating from the Naval Academy, and tried to use if for about a month. Due to his intensive military training, he never got it running before deploying to Japan a few months ago. I told my neice, "I can fix that." She got it to me yesterday (Saturday).

Upon inspection, it booted and ran enough to get some Windows updates. Then it started shuting off as if running hot. I installed HWMonitor and the CPU temperature was about 70 degrees without a load and was bumping 100C when the CPU was loaded. I found a setting in BIOS to run the fans in Performance mode. The fans ran fast when first started and slowed as CPU activity abated. After getting all Windows updates, I shut it down for the night.

On Sunday morning, I opened the Alienware update utility on the computer, and it indicated that a CRITICAL BIOS update and an update for the Alienware Comand Center (AWCC) were available. The BIOS update was to fix a CPU vulnerability and to fix the overheating and subsequent shutdowns that the computer had experienced. I was elated.

I proceeded to download and update BIOS with the Alienware program (AWCC). I did not see a BIOS is being updated screen or any progress for BIOS installation. It went to a black screen. The computer does not respond to the power button being depressed. I tried pressing CTRL + ESC without the power adapter connected and then connected the adapter while still depressing CTRL + ESC (looking for the keyboard to light up). I did that several times. No joy. My next step is to remove the back, disconnect the battery, and hold the Power button for about 30 seconds to clear CMOS.

I have found with searches at my desktop that this is a common problem that is well documented. I am hopeful that clearing CMOS will get the computer to start. However, many Alienware owners have found the the BIOS is bricked. One owner consulted Dell and bought a new motherboard to fix the problem. The new motherboard came with the new BIOS that fixes the overheating problem.

What frosts me is that Dell should have recalled these computers for repair. I am posting here for any experience any of you or anyone you know has had regarding this issue with Alienware. Do the owners have no recourse but to buy a motherboard for following the AWCC software's instructions?

Edit on Monday: I determined that the board does not have a CMOS battery, disconnected the 86 Wh battery, and shorted CLRP 1. I also shorted CLRP 2 for good measure because some boards use CLRP 2. It would not power up.

I called Dell Support and the Indian technician was good. The trick for restarting after an overtemperature shutdown like I am now experiencing is to: 1. Get a timer and press the Power button for 35 seconds, then release. 2. Wait 90 more seconds. 3. Press the Stat button normally. 4. Wait for it to start (which it does after a few seconds).

Dell support claims that there have been no reported problems with this particular computer (wrong) and cost of Dell repairs would be at the owner's expense.

The computer is still shutting down on high temperature although the CPU and GPU fans are running in the performance mode. While looking for the location of the non-extant CMOS battery on YouTube, I found a tear-down of an overheating m17 R4 (verisimilar) which had an improperly installed CPU. This is a link to the point in the video showing the production error.

My plan of action is to remove the motherboard, reseat the GPU and CPU and look for any surprises. I have plenty of time and may take a break since my niece is not taking it to Japan until this summer. That will give me lots of time to stress test it on BOINC projects. I believe it is an Intel I7-10750 and know that it has an RTX 2060.

 

[Nick Name]

Did it overheat while updating the BIOS? That's crazy.

Dell support claims that there have been no reported problems with this particular computer (wrong) and cost of Dell repairs would be at the owner's expense.

Ridiculous, but sadly a typical response and not just from computer companies. My sister heard this exact line from a Samsung fridge rep when she was trying to get the defrost problem on her Dutch door fridge fixed. Complaints for that issue go back to at least 2014 and have resulted in a class action suit.

I was going to suggest a number of things but I just saw a it's a laptop. The only good suggestion I have is to put something under it to make sure it's getting good airflow. That worked for a laptop I had years ago although it was more than a couple years old at the time. Good luck!

 

[supdood]

I run BOINC on old laptops and here's what I find has helped with heat:

• As Nick Name suggested, getting the bottom off of the surface an inch or two is critical to get proper airflow.
• I added small USB travel fans, blowing across the laptop toward the fan exhaust. I point them so that the stream breaks evenly over the top and bottom, which helps dissipate heat on the keyboard side while also providing fresh cool intake air. This wouldn't be ideal for your situation, but maybe better than buying a new motherboard?
• Saved the best for last: There should be a setting in BIOS to disable CPU boost. I believe it is usually under the Intel settings. If it is CPU temps that are causing the shutdown, this might solve your problems on its own. There is a huge temp difference between normal full load and boost full load.

But at the root, I agree that it is absurd for Dell to put out a high-performance gaming laptop that can't handle the heat, and then try to pretend like nothing is wrong. Hope you find a solution!

 

[Vester]

After watching the video showing the production error, I have decided that I will not send it back to Dell unless I cannot properly seat the CPU. I have my magnetic mat and tools laid out to being disassembly of this nightmare laptop design later this week. If you watch the video, you will see that the motherboard and one daughter board have to be removed in order to get to the heatsink assembly. The case, not counting the monitor, is in just three pieces (one of those is a bezel over the rear ports) and the keyboard is not removable from the top side.

I wish my great-nephew had told me about this computer problem during the warranty period which ended last July. He bought it two years ago and had been unable to use it. Like most young people, he may have figured that his old, 80-year-old uncle wouldn't know what to do.

Edit: Thanks, supdood, I plan to change the BIOS settings because it does run at nearly 5GHz. It would be nice to see how much the updated BIOS could help, and your suggestion could allow it to run long enough to do the BIOS update. You all know me and this laptop won't be considered fixed until it can run hard and long. I'll try the easiest BOINC project, Numberfields, first; and then I'll try Folding@home which stresses computers the most.

Edit 2: Tuesday morning update. I proped the laptop on my toolkit to improve airflow, opened the laptop, and it started when the lid was opened (normal setting in BIOS). I went into BIOS and disabled SpeedStep and set the fans on Performance mode. It booted into Windows and ran at about 28 to 40C without a load, but had peaked 70C early during startup. When CPU activity decreased, CPU temperatures were 28C. I downloaded and installed BIOS version 1.22.2 dated 14 March 2023 (can't get much newer than that). After a period of running Windows satisfactorily at slow CPU speeds (2.6 vs 4.9GHz), I enabled the CPU SpeedStep. It ran for about two minutes before shutting off. I am going to run it a 2.6GHz for a while but I plan to reseat the CPU and GPU.

Edit 3: Tuesday afternoon. That disassembly/assembly wasn't so bad. There was a gross amount of dried themal compound present, and the black membrane, about the consistency of electrical tape, surrounding the GPU and CPU appeared to have been interfering a bit on the CPU. I trimmed it back about 3-4 mm on two sides with my exacto knife. It is still shutting down without heavy loading, but HWMonitor did not indicate any temperatures over 60C. More will be revealed... I'm going to boot Pop!_OS on a thumb drive just for fun. Follow-up: The shutdowns were software related. It now crunches Milkyway@home at 2.6GHz. The GPU runs 53C when running one task without increasing fan speed in MSI Afterburner. The Intel onboard GPU runs 52C. The CPU, however, is running at 80-90C when loaded 100% at 2.6GHz. I tried it SpeedStep enabled and Speed Shift disabled. That causes to run about 3.2GHz with thermal throttling. Unstat still. I'll keep tweaking for cures. The CPU fan does not run at full speed although bumping 100C.

Edit 4: Late evening. I found a setting in the Alienware Control Center for fans. It was set on quiet. No more! Crunching CPU temps are about 75C @2.6GHz. I'm getting somewhere. I'm letting it cool overnight. I'm thinking I should run MemTest86 although it passed diagnostics earlier.

 

[Vester]

The laptop ran about five hours yesterday and ran Milkyway@home some of the time. Now it won't run two minutes. The laptop shuts down as if someone had pulled the power cord on a desktop. I've disabled fast start and am trying to remove bloatware. I would do a Windows repair or reinstallation if I could trust it to run long enough! Coud this be a battery issue? I am using the 240 W Alienware power adapter and the laptop shuts down with or without the adapter.

 

[supdood]

Great to hear all the progress you are making! One thing I've noticed running BOINC on non-performance laptops is that they simply can't dissipate the heat when running high loads for long periods. Dropping the CPU frequency like you've done is the one thing I've found that keeps them in a thermal sweet spot for long periods. Might be worth seeing if you can crunch for a few days at low frequency without heat build up. I'm not positive, but I would assume that gaming won't put nearly as much load on the CPU, and with your resetting, disabling boost, and getting the fans running properly, that might be enough to keep it from spiking and shutting down when your nephew uses it!

EDIT: Just saw your new post as I was writing this. Yes, it could be a battery issue! Not sure why I didn't remember this, but I had a laptop once that kept shutting down because it was having battery errors due to my never running it on battery. You should go into the Event Viewer and limit the results by time to right before and right after the shutdown event. This may give you some leads as to what is going on. Here's what I was seeing that confirmed it was a battery issue (there are a lot more events, but these are the important ones):
Source: Kernel-Power; Message: Critical Battery Trigger Met
Source: Kernel-Power; Message: The system is entering sleep. Sleep Reason: Battery
(Manual restart)
Source: Kernel-Power; Message: The system has resumed from sleep.
Source: Power-Troubleshooter; Message: The system has returned from a low power state. Sleep Time: ‎2022‎-‎11‎-‎17T15:46:36.557950700Z Wake Time: ‎2022‎-‎11‎-‎17T15:51:37.637604000Z Wake Source: Unknown

 

[Vester]

I found the resolution. I was sitting and thinking about the laptop. I said to myself, "This is the point where the hardware guys say it is a software problem and the software guys say you still haven't found the hardware problem." Then I realized that I've done everything that I can to the hardware without a full electronics shop and more knowlege. I had already found four viruses with the Microsoft Online Virus Scan, and who knows how many had been removed in the past. Chkdsk and surface scan had found nothing. I could not find a driver conflict. There had been over a hundred "Kernel Power" code 41 shutdowns that could have caused damage to files.

The resolution: I did a clean installation of Windows 10 and it ran fine. I then upgraded it to Windows 11. It is crunching Milkyway at home at 2.6GHz, fan speed set on Performance in BIOS. CPU max temperatures are 80C. The RTX 2060 is crunching at 53C. You can welcome new team member Wyatt until I return his laptop.

If the heatsink is performing normally, I'd say these m15 R3 laptops cannot play games with Intel Speed Shift Technology enabled (CPU about 4.9GHz), but they probably can run games with Intel Speedstep only. (CPU runs up to about 3.7GHz). Crunching at 3.7GHz throttles the CPU at 100C frequently.

The Alienware bloatware, AWCC, is messing with my mind by offering to overclock the CPU. That is outrageous.

Get an ASUS gaming laptop.

 

[Jason Jung]

The newer Intel CPUs are designed to boost as far as they possibly can under load. 100C is something one should expect to hit under load. It's not like in years prior where it was a temperature to avoid at all cost. If the CPU isn't dropping below base clock at 100C it isn't really "throttling" it just doesn't have the thermal headroom to boost higher. Base clock for the i7-10750H is 2.6 GHz.

 

[Vester]

Thank you, Jason, for the helpful information.

I feel like Columbo. Just one more thing. I had some more abrupt shutdowns after thinking that all problems had been fixed. It is now running BOINC with normal, not performance, fan speeds set in BIOS. The CPU cores are running at about 3.9GHz with three GPU tasks and no CPU tasks. With CPU tasks and Intel GPU tasks, the CPU runs at 2.9GHz with fan speed set for performance.

The shutdowns happened after installing Nvidia drivers (531.29), but I cannot be sure that this driver was causing the problem. What I did: I started in Safe Mode, navigated to Startup Settings and entered 7 for "Disable driver signature enforcement."

Here's hoping the laptop has been repaired entirely.

 

[Vester]

The laptop has crunched Milkyway@home for more than 24 hours. I quit running CPU and Intel GPU tasks to better simulate running games. This computer is ready to be sent to Japan after it finishes work in queue. Fan speed is set to Balanced in the AWCC.

---

.

 

[Vester]

This is how it runs with fans set for full speed.

 

[Vester]

Not much ego in the computer name!

 

[Vester]

There's a long story about why the laptop is not in Japan. I sent it via USPS with packaging at the UPS Store in Seneca, SC, for $94.07. USPS returned the package to me from Chicago. They declined to ship it because of the installed Lithium battery. Neither USPS or UPS Store will return the money. It turns out that NEW laptops can be shipped to FPO addresses, but used ones cannot.

When I got it back, I wanted to ensure that it was not damaged in shipment. It ran fine. Then I used Windows Update...

I now believe the Intel Graphics driver was the problem. After a major Windows Update, the computer began to randomly quit again. After a couple hours trying varios settings including "Disable driver signature enforcement," I installed the latest driver from Intel and no longer have a problem. It is happily crunching Asteroids.

 

[Nick Name]

I've dealt with the local UPS store for years and to my knowledge they don't ship via anything but UPS. They will receive from other shippers, e.g. FedEx and USPS if you have a box at the store, which I do. They told me once I could drop USPS letters off if I wanted to but I have never tried it.

I can't recall shipping anything like that but I have shipped a number of things USPS Priority Mail over the years and you always have to declare the contents of the package. I expect the same is true at the UPS store, someone should have caught it if it couldn't be delivered.

I'm always worried when I come across problems like this, is it really fixed or just acting nicely for the moment.

 

[Vester]

By the way, the laptop could have been shipped anywhere in the lower 48 without a problem by ground. FedEx and UPS do not deliver to FPO (military) addresses.

 

[Jason Jung]

My local UPS Store offers a whole list of USPS services. I have zero reasons to go to the post office anymore.

Seems the restriction on shipping non-new laptops with lithium batteries to and from APO/FPO/DPO address came out last year.