Updated: BeeLink GTR7 7840/7940 Pro Random Reboot/Crashing Issue (The Reason and a possible fix for some)

Hello,

I just watched Level1’s Bellink GTR7 video (RIP GTR7) and I wanted to post this to help anyone else out that may have one and are having issues with random reboot/crashes.

I received my BeeLink GTR7 (7840) unit last week and I love the little guy as it is an astounding what Beelink was able to pack into this will package but like many of you out there that have also received their GTR7. It quickly went from awesome fun to bad times due to random issues or problems. One of the biggest issues most users are having is the random rebooting/crashing problems and I, like many of you have experienced that on my unit.

What has been discovered on this issue is that most of the common reboot/crashing that is occurring comes down to a problem with the power management of the APU/Memory SOC while the system is at idle. Specifically, the C6 power state is not working as it should and causing issues with core parking, memory parking (didn’t know this was a thing with DDR5) and hardware power control on the peripherals like the NIC’s and Bluetooth controller. It is a two-part problem with the current AMD drivers and some issues within the currnet BIOS.

BeeLink has acknowledged this issue and is currently working on BIOS update (no ETA given) that will address the broken power settings in the current BIOS (namely the TDP option for the APU that actually doesn’t work when setting to anything other 65w), other power related options that are either missing or are not working as intended as well as an update to the memory training process to help with memory detection and stability.

They have also stated in their forum that they are expecting an updated driver package from AMD sometime around July 26th for the 7840/7940 APU’s. Whether it is the official release or just an updated OEM beta driver, we don’t know.

In the meantime, I have put together a little something that hopefully will allow you to start enjoying your new little toy while we wait for all the proper fixes. I have provided steps and settings below for the GTR7 that have worked for many others to resolve the random rebooting/ crashing. These settings have worked for my unit which has now been reboot/crash free for 48 hours but please note, they are not an official fix nor has Beelink given any advice or suggestion for any of these provided changes. I found these through my own testing and forum searches and have put together this list in a single, central location where I believe will be the most helpful. These settings are only for Windows currently as I am not that versed with messing within the guts of Linux (nor do I have the time or brain power available).

BIOS:

  1. Advanced>ACPI Settings: Disable “Enable ACPI Auto Config” and set both options below that appear to Disable and Suspend Disable

  2. Advanced> AMD CBS>NBIO Common Options>GFX Configuration: Set iGPU to UMA_Specified. Set UMA Frame Buffer Size to a static size. You choose what you want or what you can afford to give to the GPU for memory. Suggested 2GB or 4GB. (This supposedly helps with keeping both DIMM’s awake).

Windows:

  1. Disable hibernate: Open CMD/Terminal and enter in command: powercfg hibernate off (Must run terminal/cmd as administrator)

  2. Disable hardware power savings: Easiest way to do this is open device manager, then you must open the affected hardware’s properties page by double clicking on them and then under the Power Management tab. Uncheck “Allow the computer to turn off this device to save power”. Do this for both of the NICs, the Intel Wi-Fi 6 AX200 NIC and Intel Wireless Bluetooth adaptor. Note if you have any of these devices disabled, then you can ignore them.

  3. Disable sleep and CPU low power/core parking. Go to Control Panel\All Control Panel Items\Power Options\Edit Plan Settings. Here change “Put the computer to sleep” to Never. Then just under that setting, click on the link “Change advanced power settings”. Click on the + to open Processor Power Management then say for Minimum Processor state and here enter in 10%.

After you have changed these settings, reboot using run prompt and type in shutdown -r which bypasses the Windows quick reboot and performs a full clean reboot. Once it comes back up, test it out and since this issue usually occurs at idle, I suggest letting the unit sit just at the desktop for a while and see if your system reboots anymore. After it has been sitting for a while with no reboots, push it for a bit with a benchmark, game or anything that is going to make it work, then let it go back to idle for a while.

Hopefully this will help some of you out and if anyone has any additional settings, they found to add to this list that also work. Please add them in.

4 Likes

Small update to this post as I may have found an issue that may or may not be connected to some of the other instability issues with GTR7 some of you are having.

After running a few games to test out the systems stability under load. I was monitoring all the temps with HWINFO and I noticed that while all the temps were doing great. One of the DRAM chips was showing really high temp hitting 94-95C which is right at the specified limit for DDR5 operational temps. The memory chip in questions was found to be the bottom most chip closest to the PCB. During the testing, I did start getting some strange stuttering as well as some graphical artifacting and that is what drew my attention to the temps and finding the memory situation.

I stopped the testing, let all temps return back to normal without rebooting, then started the game testing back up and the stuttering as well as the artifacts were gone. I did run the test till the memory temp hit 95C again and after about 10 mins of hitting the high temps again, it started showing the stuttering and artifacting in the game.

So, there may be an inherent design issue that is not allowing the bottom stick of memory to get enough airflow and causing some performance issues and may eventually even killing that stick.

I am going to run some test with the just the bottom plate off and then also removing the plastic shroud and fan off and see what the temps end of being in both situations running the same test.

I will update on what I find. I am also curious if Wendell ever found out what actually happened to his GTR7 unit?

Could it be a faulty memory cell ?

I wonder what would happen if you swapped the bottom SODIMM with the top module and test once more to see.

I am in the process of testing that as we speak. I will update when I have completed my results.

I also purchased a whole new memory kit G.Skill (5600S4040A16GX2-RS CL-40) to see if it performs any differently on temps as well if I get anymore performance with the lower timings.

Gotcha, was the default RAM kit a set from Crucial ?

Also, another update, AMD did release the WHQL driver package that fully supports 7840HS/7940HS and is available now (July 26th, 2023) directly from AMD website.

So far it appears to have resolved most of the C6 power state issues, but I am still testing it on my unit. I highly suggest to DDU the current BeeLink provided driver before installing the new WHQL version.

Yep, the same Crucial DIMMs that everyone is getting in the GTR7 along with the Crucial P3 Plus SSD.

Nice to know about the latest driver package.

I want to get the latest BIOS/Firmware for my ASUS mobo and have it do that long/extended training for my Gskill ram kit and see if it helps.

Then install the latest driver package.

I bet even desktop boards can benefit from the additional power management functions and fixes.

We should see AGESA 1007B updates in the next month or so but in your case. I would wait a little bit after ASUS release their BIOS to be safe and give the update a little time to cook out in the wild. Their track record on BIOS stability as of late has been less than stellar.

Yeah I get that but I couldn’t wait, I updated and then installed the latest AMD driver package and there was an update for Windows 11 too.

How is your system running now?

After installing the updated AMD Drivers, the system appears to be much more stable. I have reverted all power settings back to default, enabled secure boot finally and had only one odd reboot when I left it idle and that was it running for four days straight of use (Actually using as my full-time office PC to test it).

Still testing the cooling situation for the bottom and waiting on the G.Skill memory kit to show up but preliminary results show there is a slight cooling issue with the current setup on the GTR7 when the system is pushed for gaming and such. Looks like BeeLink already knew this as they have had a replacement bottom cover for the GTR7 and GTR6 that has a grill cut out for the bottom fan to pull more direct air in which has been available since July 18th. I have reached out to Beelink’s support to find out how to get one of the replacement bottoms and will post the process here when/if they respond.

What I have found so far is with the just the bottom cover off and with the unit sitting at the same height on a desk as if the bottom was on. There is 3C drop (from 95C to 92C) on the memory closest to the PCB. Then I raised the unit up 1in from the desk with just the bottom cover off and then I got around 5C (from 95C to 89-90C) running the same games. Still going to run more tests with the bottom plate along with the plastic cover with the fan off and see what I get as well as what temps the G.Skill memory also end up running at over the Crucial. Hopefully have it all completed next week and will post what I have found.

1 Like

Process to get replacement bottom with fan grill for GTR7.

This is the process for those who have GTR7 with the solid bottom cover and would like get the updated bottom cover with a fan grill cutout. (You really should get it)

It’s pretty simple process. Just send an email to their support email [email protected] with the subject (Replacement Bottom Cover for Beelink GTR7with the fan grill). In the body, just let them know you recently bought a GTR7 and would like to get the replacement bottom cover with a fan grill.

A support representative will reply back (I received a response in 24hours) asking for a photo of the full bottom of your GTR7 and clearly showing the sticker with the serial number on it. A copy of your receipt/invoice of purchase for the GTR7 and some information for shipping.

Once they receive that information and all checks out. They will ask you to confirm the color of your unit just to make sure they send the correct cover for your version.

Looks they are shipping out at no charge which is great but not sure how long it will take to arrive as I am sure they are sending as cheap as possible so may be awhile before you get it.

Just as a note, all newer production version of the GTR7 are supposed to come default with the new bottom cover so if you are looking to get one. Just note, you may or may not get the new version with the fan grill cutout.

Update on Memory temps and system stability.

Okay took a bit longer than expected but below are my results for the memory temps using two different kits and with and without the bottom cover.

I did all testing fresh once I received the new memory kit to keep all variables close to the same as possible. Testing environment was at ambient 74F, and all tests were performed with two Samsung 980 Pros installed (as this is the default config, I am going to be using the unit in) while on an open desk.

These results are just my findings during my simple testing and should not be taken as gospel in anyway. More testing from professionals with much more reliable tools should be performed to get more accurate data as well as use of more testing variations of memory before anyone should run out and purchase additional memory for this unit. While the G.Skill kit did make the system run overall cooler and did see a noticeable performance difference in OW2, the stability issues unfortunately are still present and while system ran perfect with the new memory during testing. The next day it started to reboot every hour or so. This is more likely due to the BIOS issues and memory training that Beelink is working on. What I have found to help with this is to full power drain the unit, clear CMOS and make it retrain the memory which is a pain but does work most of the time.

Testing:
System with cover on, sitting normally on the desk idle.
CPU: 45C
Crucial Memory: Slot1 47C, Slot2 (closet to PCB) 48C
G.Skill Memory: Slot1 45C, Slot2 (closest to PCB) 46C

System at load running Overwatch 2 with bottom cover on (Ran the best on the system) After 30mins of playing.
CPU: 85C
Crucial Memory: Slot1 87C, Slot2 (closest to PCB) 95C
G.Skill Memory: Slot1 85C, Slot2 (closest to PCB) 92C

System with just the bottom cover off sitting at same height as with cover on.
CPU: 45C
Crucial Memory: Slot1 44C, Slot2 (closest to PCB) 44C
G.Skill Memory: Slot1 43C, Slot2 (closest to PCB) 44C

System at load running Overwatch 2 with bottom cover off. After 30mins of playing.
CPU: 79C
Crucial Memory: Slot1 79C, Slot2 (closest to PCB) 82C
G.Skill Memory: Slot1 72C, Slot2 (closest to PCB) 74C

System with bottom cover off and sitting 1/2in off desk idle
CPU: 42C
Crucial Memory: Slot1 41C, Slot2 (closest to PCB) 42C
G.Skill Memory: Slot1 38C, Slot2 (closest to PCB) 39C

System at load running Overwatch 2 with bottom cover off and 1/2 off desk. After 30mins of playing.
CPU: 76C
Crucial Memory: Slot1 75C, Slot2 (closest to PCB) 78C
G.Skill Memory: Slot1 68C, Slot2 (closest to PCB) 69C

I did test with both the bottom cover and plastic fan cover off but wasn’t much change in the memory temps just the SSD’s ASIC’s ran a bit hotter.

The optimal temps are found when the unit is 1/2in off the desk without the bottom cover and shows why Beelink has produced the newer cover with the fan grill cutout. The solid bottom case appears to be trapping heat inside the unit increasing the overall operating temps of all components.

As far as the height off the desk affecting the cooling, I found that 1/2in is the optimal distance with the bottom cover off and anything higher didn’t make a positive difference.

Once I receive the new bottom cover with the Fan grill cutout. I will retest and see how the system performs with just a fan grill added and at different heights off the desk.

4 Likes

Thanks, nice info :slight_smile:

So I have to report a sad note here. My GTR7 unit has basically died as of last night. The system started to randomly rebooting at idle about every thirty minutes or so progressively getting worse and then it finally gave one good BSOD after which. The system has become completely unstable and is crashing within a few minutes of being in the OS at idle.

Pulled out both 980 Pro SSD’s. Put a simple OEM Samsung 256GB SSD in, install another known good DDR5 16GB kit of 4800Mhz memory and attempted to reinstall Windows 11 fresh and before I was able to install any drivers, it started rebooting itself.

My last ditch effort is to see if it may be an issue with the factory power supply. I had to go get a 100W+ USB-C power adaptor. So I will test tonight and see if makes any difference.

Also I am curious, for those that have a GTR7. What color is your power adaptor? I have seen many peoples review units and they all appear to be black. Mine however is white which I was puzzled by so interested if anyone else out there has a white power adapter for theirs?

1 Like

Sorry to hear this, please post any updates on this matter, thanks.

So this unit can be powered via the USB-c connector IF say your display has one of these ports that can provide at least 100 watts of power ?

That is nice to know, I wonder if my GT5 unit supports this, I just use the supplied AC power adapter which is also black in color.

So supposedly you can’t run the GTR7 at the 65W level while powering it over USB-C and needs to be lowered to 55W for USB-C power to work from what I have been told.

Though I just watched ETA Primes review and he just plugged his GTR7 right into his Monitors USB-C and it powered it and ran gaming test just fine at 65W. So I guess if you don’t bust the 100W limit you are fine.

Also good to know, so it maybe an upper limit of either 55 or 65 watts over USB-c

Well the limit for the GTR7 is supposed to be 100W power delivery over USB-C. From my own test and what other have seen, at the 65w setting. You can easily peak over 100W total power draw at the wall when pushing the GTR7 but if you are just using it as a normal office PC, then you should not have any issue with powering it over USB-C like from a monitor.

I just ordered a 140W USB-C power adaptor and will see how that does. Strange thing is the system has been sitting all day unplugged on my lab desk and when I came home, I just plugged it in and left again and when I got back. It is still on and has not rebooted once in the past 5 hours. So I have no idea what the heck is going on with this crazy thing.

I keep peaking in on the official Beelink forum for any new official updates and saw that some people have now traced the rebooting issue to possibly being something when you are using dual monitors with one in the HDMI port and one in the DisplayPort. The one user stated when they switched the one monitor in the HDMI port to using DisplayPort over USB-C, they did not have anymore rebooting issues.

Currently my GTR7 is sitting on my Lab desk plugged into a single monitor through the HDMI port, so I am going to let it run over night just doing some simple automated tasks and see if it is still on in the morning. If it is, I will then attempt to plug it back into my main 32in monitors and run the one through USB-C instead of HDMI and see how it behaves.