- Aug 1, 2020
- 7
- 0
- 4,520
Good afternoon everyone,
I'm hoping someone out there will be able to help me with this frustrating problem, as I'm running out of ideas (apologies, long post ahead).
The issue almost always occurs during gaming or when running GPU benchmark tests. My PC simply restarts itself with no warning; no BSOD; nothing obvious in the event viewer (at least not obvious to me); no obvious impending issues from monitoring the temperatures/voltages/loads before it happens.
I'll get into the many steps I've gone through to try to resolve this below, but it might be useful to know a bit of background first. I'm based in the UK. I purchased the components during 2024 and finally ended up building the machine over the 2024 Christmas break. This was the first time in about 20-years that I've built a rig from scratch, so it's entirely possible that I've made some sort of mistake somewhere. It has worked faultlessly since December until around 2-weeks ago (mid-April 25 is where I first noticed any issues).
* All components were bought brand new with the exception of those marked * which were carried over from my previous machine (two HDD, PSU and Monitor).
** The GPU was purchased as an ex-display Refurbished unit from Ebuyer. The sticker on the box states it was refurbished on 17 Apr 2024 and I purchased it in October 2024. I'd have to double check, but I believe it came with a 1-year warranty, so should still be covered.
CPU: AMD Ryzen 7 7700X
CPU cooler: Corsair H100x RGB ELITE
Motherboard: Gigabyte X670 AORUS ELITE AX Rev 1.3 (current BIOS - FB4 - 12 Mar 25)
Ram: 64GB DDR5 (2x 32GB - G.Skill F5-6000J3040G32GX2-FX5, Flare X5, AMD EXPO kit)
SSD/HDD: 1x Crucial P3 PLUS M.2 NVME PCIe Gen 4 1TB SSD, 1x Western Digital 2TB HDD*, 1x Toshiba 3TB HDD*
GPU: MSI Radeon RX 6750 XT Gaming X Trio 12gb**
PSU: Corsair RM850x* (5-years old)
Chassis: Cooler Master HAF 700 Full Tower
OS: Windows 11 Home 10.0.26100 (24H2) fully updated
Monitor: AOC 27B1H 27" IPS Full HD Monitor* (1920x1080) - also linked to main TV using HDMI splitter
+ I have not fiddled with any of the CPU settings or done any overclocking - it has always stayed at factory settings
+ Also worth mentioning is that my setup is a little unusual in its layout. My PC is located in my front room next to the main TV (to which it connected via a HD splitter) though the desk where I work is in the next room. I use a 10m HDMI cable (with booster) and a 10m USB cable with hub to reach the desk next door. I don't think this should make any difference, but thought I would mention it just in case
-
I do a reasonable amount of gaming but nothing outrageously demanding. Fallout 4, Skyrim and Sea Power probably being the hardest on the GPU. Then a few weeks ago I got into Borderlands 3 which I'd not played on PC before. This is where the issues began. Hours of playing this on Ultra settings passed without a hitch, including online co-op multiplayer, until one day I had the first of many restarts. This was about 4-hours in to a 6-hour online multiplayer session. After the restart, I went back into the game and there were no further issues for the rest of the session, albeit with my graphic settings turned down to Medium just in case.
Since then, I think things have -possibly- become worse. The problem is that I can't 'provoke' the issue reliably; sometimes it'll work for hours in various games without restarting, sometimes it'll restart after 10-mins. Sometimes it will make it through GPU benchmark tests, and sometimes it won't. As you'll see from the steps I've taken below, it almost always happens when the GPU is under some form of load, but the amount of load doesn't seem to matter too much as long as its above sitting on the desktop sort of levels.
-
Below is a kind of timeline of the steps I've taken to try to fix the problem. It is (as far as I can recall) roughly in the order I tried each fix.
- Monitor temperatures using HWINFO and AMD Adrenaline overlay. CPU averages 36c at rest and around 45c while gaming with maximum spikes up to around 80c. GPU averages around 28c at rest and around 50c during gaming/benchmarks with highs of up to around 58c.
- Experiment with fan profiles, including playing with all fans running 100% at all times - makes no noticeable difference to restarting issue.
- Checked Windows fully up to date
- Updated BIOS as I noticed it was not on the most current version available (now on FB4 released 12 Mar 25)
- Updated all chipset and GPU drivers to newest versions available
- Ran SFC /Scannow and DISM image-cleanup in Command Prompt
None of this managed to solve the issue. At this point I thought the RAM might be the main suspect, so I did the following checks on the RAM:
- Double checked my RAM was on the QVL list for my motherboard (it is)
- Removed and re-inserted sticks to check seated properly
- Swapped the sticks around
- Tried running with one stick at a time (for both sticks)
- Ran Windows Memory Diagnostic - no issues found
- Ran MemTest86 - 14 tests x 3 passes (it took a while!) - no issues found
- Checked RAM in BIOS (having not changed any settings from factory up to this point) and noticed it was running at 6000MT on EXPO profile 1. I decided to turn off EXPO profile 1 and put everything on the bog-standard settings at 4800MT to see if it would improve stability.
None of this seemed to make any difference. After my RAM passed all the checks I thew at it, I've pretty much ruled it out as a suspect. Unfortunately, I don't have any 'spare' DDR5 sticks to swap mine out with or another machine to test mine in currently. If some of the answers to this post come back to the RAM as being a possible cause, then I'll try to get some.
After this, I moved on to the GPU as my prime suspect.
- Removed GPU, visually inspected (no issues noticed), and re-seated
- Checked cabling to GPU - noticed that there were 2x 8-pin connectors but that I had used a 'daisy chained' cable to both from one output on the PSU. Replaced the single 'daisy chained' cable with two separate ones direct from the PSU.
- Checked/refreshed drivers
Again, none of this seemed to make any difference. So I then moved on to looking at other random stuff and doing general PC health checks on the offchance there was some conflict or background issue somewhere.
- Ran CHKDSK checks on all my drives - no issues found
- Ran disk optimiser/defrag - the 3TB HDD was 1% fragmented and the 2TB was 10%, so fixed those.
- Ran the System File Checker (SFC /scannow) in command prompt - no issues found
- Ran the DISM cleanup-image thing in command prompt - no issues found
- Used the Norton removal tool to (at last) completely obliterate the expired Norton antivirus that came installed with something (Windows?)
- Download Malwarebytes and run full check - no issues found
- Checked/removed/updated drivers for sound devices and network adapters. Disabled WiFi adapter (my machine is plugged directly into router)
- Cleared browser cache (Chrome); stopped using Chrome and started using Opera GX instead.
- Checked that 'Fast Boot' was off in the BIOS (it was)
- Checked setting in BIOS relating to loss of power - set to 'stay off' when there was a loss of power (still get restarts, doesn't stay off)
- Changed systemwide power profile to Balanced (was previously on High Performance)
- Changed the virtual memory settings (in system performance settings) to 1.5x initial and 3x maximum RAM amount (something I read on a Windows forum)
Once again, none of that seemed to make any difference to my restarting issue. At this point I was getting a little frustrated, so decided to go a bit more drastic:
- Disassembled my PC, making sure to visually inspect all components and connectors - no obvious issues identified.
- Checked PSU function by bridging the two pins to get the fan to spin - working.
- Re-assembled my PC, using fresh thermal paste etc. The only change I made was in the routing of some cables and how some of the fans were grouped together to make it more logical (two big fans on SYSFANS2, fans in the base on SYSFANS3, etc.).
- Removed GPU drivers using DDU (in Safe Mode) and re-installed fresh drivers direct from manufacturers website
- Ran SFC/scannow and DISM image-cleanup again - no issued found.
At this point it was the weekend again and time for my next online Borderlands 3 session. I played on medium graphics settings, monitoring temps, FPS, GPU and CPU load, system memory usage, etc. etc. - I got a restart after about 2-hours, nothing from the monitoring showed any potential issues - after the restart, carried on playing for the next 3-hours with no restarts.
I then decided to see if I could 'provoke' a restart to try to figure out what the underlying cause was - I was thinking along the lines of the GPU drawing too much power from the PSU (or a failing PSU) or some other internal fault with the GPU at this point.
- Ran various benchmark tests within Borderlands 3, getting good results as expected (98 FPS on 'badass', 104 FPS on 'ultra', 120 FPS on 'high', 149 FPS on 'medium') with no restarting issues and temperatures all low.
- Ran FurMark GL benchmark test (1080p benchmark 1); 2 times out of 3 it passed with flying colours, scoring 11747 and 11552. The other time, I got a restart.
- Ran Unigine Valley v1 benchmark on various different settings; it only completed the test without restarting on one occasion (on high settings, even though it had been consistently failing on medium).
- Asked a friend to run the Unigine Valley test on his machine (7-ish year-old pre-built i7-9700k with a GTX 1080) at medium settings for comparison; no issues for him running the test repeatedly on various different modes.
- Ran the CPU Burner thing in FurMark just to try to rule out CPU issues - CPU got hot (topped out at around 93c) but no issues or restarts.
- Went back into Borderlands 3, put the settings on 'badass' mode and left my character standing in a relatively busy spot (NPCs walking past and background activity) for around 8-hours (overnight) - no restart, all fine.
I then started to look at the graphics card settings in the AMD Adrenalin software to see if detuning the card a bit to reduce the power draw would make any difference:
- Changed overall settings from 'high performance' to 'HYPR-RX Eco' with Radeon Chill enabled; aiming to draw as little power as possible while maintaining 60FPS. This did reduce the wattage by quite a bit. While I had seen over 200w drawn before, I could no longer get it anywhere near that. I did some more benchmark tests in Borderlands 3 and noted the highest wattage (approximately) I noticed on each setting. Throughout, the GPU temp never got above 48c.
Badass - 160w
Ultra - 150w
High - 130w
Medium - 90w
Low - 80w
However, even on these settings, I was still getting restarts in game from time to time. no real improvement.
I did some further testing on other games, playing for an hour each on Fallout 4, Skyrim, FTL, Plague Inc, and Sea Power, to try to get a mix of different games in there. I played all on 'high' graphics settings or equivalent. Fallout 4 and Skyrim both worked well with draw around 70-90w with GPU utilisation topping out at 48% and max GPU temp at 51c. FLT barely cracked 10w and 5% usage at 33c, while Plague Inc was not much higher than that. I was getting up to 120w in Sea Power at something like 75% GPU utilisation and 50c-ish temps, which is when I eventually got a restart.
I then decided to look at undervolting the GPU to see if I could reduce things further.
- Enabled the standard undervolt option in AMD Adrenalin, which knocked it down from 1200mV to 1175mV, but I did not notice any meaningful difference when playing the same games above.
- Tried some settings I found online somewhere (Reddit possibly?) undervolting to 1100mV, changing clock speed to 2600, upping the VRAM frequency and enabling VRAM fast mode. This seemed a bit better at first, but then I got a BSOD pretty much straight away in Borderlands 3.
- Changed GPU settings back to the standard 'eco mode' ones.
- Tried running the FurMark and Valley benchmark tests again - it kept failing to complete these resulting in a restart.
I was now running out of ideas, so looked at power delivery:
- Changed the power cable between the wall socket at PSU (twice, so using three different cables in total, including a brand new one)
- Changed which wall socket the main power cable was plugged into. (though still on the same circuit - all my sockets are on one circuit (except for the kitchen) as it is a small older house).
After all that, I'm still no closer to figuring out what is causing these restarts. If anything, it seems to take less stress to cause one now that I'm 2-weeks down the line from the first occurrence.
My main suspects are still the PSU and GPU though, as far as everything I know/have read leads me in that direction.
My next steps were going to be:
- Swap out PSU (a friend of mine has a brand new Cooler Master MWE 650 v2 (MPE-6501-ACABW) that he's willing to let me try)
- Swap out GPU (I have my old Gigabyte 1050ti OC to try)
- Reinstall Windows?
I would greatly appreciate it if anyone here has any other ideas for things to check, or past experiences to share on how to resolve this. Thanks in Advance.
I'm hoping someone out there will be able to help me with this frustrating problem, as I'm running out of ideas (apologies, long post ahead).
The issue almost always occurs during gaming or when running GPU benchmark tests. My PC simply restarts itself with no warning; no BSOD; nothing obvious in the event viewer (at least not obvious to me); no obvious impending issues from monitoring the temperatures/voltages/loads before it happens.
I'll get into the many steps I've gone through to try to resolve this below, but it might be useful to know a bit of background first. I'm based in the UK. I purchased the components during 2024 and finally ended up building the machine over the 2024 Christmas break. This was the first time in about 20-years that I've built a rig from scratch, so it's entirely possible that I've made some sort of mistake somewhere. It has worked faultlessly since December until around 2-weeks ago (mid-April 25 is where I first noticed any issues).
* All components were bought brand new with the exception of those marked * which were carried over from my previous machine (two HDD, PSU and Monitor).
** The GPU was purchased as an ex-display Refurbished unit from Ebuyer. The sticker on the box states it was refurbished on 17 Apr 2024 and I purchased it in October 2024. I'd have to double check, but I believe it came with a 1-year warranty, so should still be covered.
CPU: AMD Ryzen 7 7700X
CPU cooler: Corsair H100x RGB ELITE
Motherboard: Gigabyte X670 AORUS ELITE AX Rev 1.3 (current BIOS - FB4 - 12 Mar 25)
Ram: 64GB DDR5 (2x 32GB - G.Skill F5-6000J3040G32GX2-FX5, Flare X5, AMD EXPO kit)
SSD/HDD: 1x Crucial P3 PLUS M.2 NVME PCIe Gen 4 1TB SSD, 1x Western Digital 2TB HDD*, 1x Toshiba 3TB HDD*
GPU: MSI Radeon RX 6750 XT Gaming X Trio 12gb**
PSU: Corsair RM850x* (5-years old)
Chassis: Cooler Master HAF 700 Full Tower
OS: Windows 11 Home 10.0.26100 (24H2) fully updated
Monitor: AOC 27B1H 27" IPS Full HD Monitor* (1920x1080) - also linked to main TV using HDMI splitter
+ I have not fiddled with any of the CPU settings or done any overclocking - it has always stayed at factory settings
+ Also worth mentioning is that my setup is a little unusual in its layout. My PC is located in my front room next to the main TV (to which it connected via a HD splitter) though the desk where I work is in the next room. I use a 10m HDMI cable (with booster) and a 10m USB cable with hub to reach the desk next door. I don't think this should make any difference, but thought I would mention it just in case
-
I do a reasonable amount of gaming but nothing outrageously demanding. Fallout 4, Skyrim and Sea Power probably being the hardest on the GPU. Then a few weeks ago I got into Borderlands 3 which I'd not played on PC before. This is where the issues began. Hours of playing this on Ultra settings passed without a hitch, including online co-op multiplayer, until one day I had the first of many restarts. This was about 4-hours in to a 6-hour online multiplayer session. After the restart, I went back into the game and there were no further issues for the rest of the session, albeit with my graphic settings turned down to Medium just in case.
Since then, I think things have -possibly- become worse. The problem is that I can't 'provoke' the issue reliably; sometimes it'll work for hours in various games without restarting, sometimes it'll restart after 10-mins. Sometimes it will make it through GPU benchmark tests, and sometimes it won't. As you'll see from the steps I've taken below, it almost always happens when the GPU is under some form of load, but the amount of load doesn't seem to matter too much as long as its above sitting on the desktop sort of levels.
-
Below is a kind of timeline of the steps I've taken to try to fix the problem. It is (as far as I can recall) roughly in the order I tried each fix.
- Monitor temperatures using HWINFO and AMD Adrenaline overlay. CPU averages 36c at rest and around 45c while gaming with maximum spikes up to around 80c. GPU averages around 28c at rest and around 50c during gaming/benchmarks with highs of up to around 58c.
- Experiment with fan profiles, including playing with all fans running 100% at all times - makes no noticeable difference to restarting issue.
- Checked Windows fully up to date
- Updated BIOS as I noticed it was not on the most current version available (now on FB4 released 12 Mar 25)
- Updated all chipset and GPU drivers to newest versions available
- Ran SFC /Scannow and DISM image-cleanup in Command Prompt
None of this managed to solve the issue. At this point I thought the RAM might be the main suspect, so I did the following checks on the RAM:
- Double checked my RAM was on the QVL list for my motherboard (it is)
- Removed and re-inserted sticks to check seated properly
- Swapped the sticks around
- Tried running with one stick at a time (for both sticks)
- Ran Windows Memory Diagnostic - no issues found
- Ran MemTest86 - 14 tests x 3 passes (it took a while!) - no issues found
- Checked RAM in BIOS (having not changed any settings from factory up to this point) and noticed it was running at 6000MT on EXPO profile 1. I decided to turn off EXPO profile 1 and put everything on the bog-standard settings at 4800MT to see if it would improve stability.
None of this seemed to make any difference. After my RAM passed all the checks I thew at it, I've pretty much ruled it out as a suspect. Unfortunately, I don't have any 'spare' DDR5 sticks to swap mine out with or another machine to test mine in currently. If some of the answers to this post come back to the RAM as being a possible cause, then I'll try to get some.
After this, I moved on to the GPU as my prime suspect.
- Removed GPU, visually inspected (no issues noticed), and re-seated
- Checked cabling to GPU - noticed that there were 2x 8-pin connectors but that I had used a 'daisy chained' cable to both from one output on the PSU. Replaced the single 'daisy chained' cable with two separate ones direct from the PSU.
- Checked/refreshed drivers
Again, none of this seemed to make any difference. So I then moved on to looking at other random stuff and doing general PC health checks on the offchance there was some conflict or background issue somewhere.
- Ran CHKDSK checks on all my drives - no issues found
- Ran disk optimiser/defrag - the 3TB HDD was 1% fragmented and the 2TB was 10%, so fixed those.
- Ran the System File Checker (SFC /scannow) in command prompt - no issues found
- Ran the DISM cleanup-image thing in command prompt - no issues found
- Used the Norton removal tool to (at last) completely obliterate the expired Norton antivirus that came installed with something (Windows?)
- Download Malwarebytes and run full check - no issues found
- Checked/removed/updated drivers for sound devices and network adapters. Disabled WiFi adapter (my machine is plugged directly into router)
- Cleared browser cache (Chrome); stopped using Chrome and started using Opera GX instead.
- Checked that 'Fast Boot' was off in the BIOS (it was)
- Checked setting in BIOS relating to loss of power - set to 'stay off' when there was a loss of power (still get restarts, doesn't stay off)
- Changed systemwide power profile to Balanced (was previously on High Performance)
- Changed the virtual memory settings (in system performance settings) to 1.5x initial and 3x maximum RAM amount (something I read on a Windows forum)
Once again, none of that seemed to make any difference to my restarting issue. At this point I was getting a little frustrated, so decided to go a bit more drastic:
- Disassembled my PC, making sure to visually inspect all components and connectors - no obvious issues identified.
- Checked PSU function by bridging the two pins to get the fan to spin - working.
- Re-assembled my PC, using fresh thermal paste etc. The only change I made was in the routing of some cables and how some of the fans were grouped together to make it more logical (two big fans on SYSFANS2, fans in the base on SYSFANS3, etc.).
- Removed GPU drivers using DDU (in Safe Mode) and re-installed fresh drivers direct from manufacturers website
- Ran SFC/scannow and DISM image-cleanup again - no issued found.
At this point it was the weekend again and time for my next online Borderlands 3 session. I played on medium graphics settings, monitoring temps, FPS, GPU and CPU load, system memory usage, etc. etc. - I got a restart after about 2-hours, nothing from the monitoring showed any potential issues - after the restart, carried on playing for the next 3-hours with no restarts.
I then decided to see if I could 'provoke' a restart to try to figure out what the underlying cause was - I was thinking along the lines of the GPU drawing too much power from the PSU (or a failing PSU) or some other internal fault with the GPU at this point.
- Ran various benchmark tests within Borderlands 3, getting good results as expected (98 FPS on 'badass', 104 FPS on 'ultra', 120 FPS on 'high', 149 FPS on 'medium') with no restarting issues and temperatures all low.
- Ran FurMark GL benchmark test (1080p benchmark 1); 2 times out of 3 it passed with flying colours, scoring 11747 and 11552. The other time, I got a restart.
- Ran Unigine Valley v1 benchmark on various different settings; it only completed the test without restarting on one occasion (on high settings, even though it had been consistently failing on medium).
- Asked a friend to run the Unigine Valley test on his machine (7-ish year-old pre-built i7-9700k with a GTX 1080) at medium settings for comparison; no issues for him running the test repeatedly on various different modes.
- Ran the CPU Burner thing in FurMark just to try to rule out CPU issues - CPU got hot (topped out at around 93c) but no issues or restarts.
- Went back into Borderlands 3, put the settings on 'badass' mode and left my character standing in a relatively busy spot (NPCs walking past and background activity) for around 8-hours (overnight) - no restart, all fine.
I then started to look at the graphics card settings in the AMD Adrenalin software to see if detuning the card a bit to reduce the power draw would make any difference:
- Changed overall settings from 'high performance' to 'HYPR-RX Eco' with Radeon Chill enabled; aiming to draw as little power as possible while maintaining 60FPS. This did reduce the wattage by quite a bit. While I had seen over 200w drawn before, I could no longer get it anywhere near that. I did some more benchmark tests in Borderlands 3 and noted the highest wattage (approximately) I noticed on each setting. Throughout, the GPU temp never got above 48c.
Badass - 160w
Ultra - 150w
High - 130w
Medium - 90w
Low - 80w
However, even on these settings, I was still getting restarts in game from time to time. no real improvement.
I did some further testing on other games, playing for an hour each on Fallout 4, Skyrim, FTL, Plague Inc, and Sea Power, to try to get a mix of different games in there. I played all on 'high' graphics settings or equivalent. Fallout 4 and Skyrim both worked well with draw around 70-90w with GPU utilisation topping out at 48% and max GPU temp at 51c. FLT barely cracked 10w and 5% usage at 33c, while Plague Inc was not much higher than that. I was getting up to 120w in Sea Power at something like 75% GPU utilisation and 50c-ish temps, which is when I eventually got a restart.
I then decided to look at undervolting the GPU to see if I could reduce things further.
- Enabled the standard undervolt option in AMD Adrenalin, which knocked it down from 1200mV to 1175mV, but I did not notice any meaningful difference when playing the same games above.
- Tried some settings I found online somewhere (Reddit possibly?) undervolting to 1100mV, changing clock speed to 2600, upping the VRAM frequency and enabling VRAM fast mode. This seemed a bit better at first, but then I got a BSOD pretty much straight away in Borderlands 3.
- Changed GPU settings back to the standard 'eco mode' ones.
- Tried running the FurMark and Valley benchmark tests again - it kept failing to complete these resulting in a restart.
I was now running out of ideas, so looked at power delivery:
- Changed the power cable between the wall socket at PSU (twice, so using three different cables in total, including a brand new one)
- Changed which wall socket the main power cable was plugged into. (though still on the same circuit - all my sockets are on one circuit (except for the kitchen) as it is a small older house).
After all that, I'm still no closer to figuring out what is causing these restarts. If anything, it seems to take less stress to cause one now that I'm 2-weeks down the line from the first occurrence.
My main suspects are still the PSU and GPU though, as far as everything I know/have read leads me in that direction.
My next steps were going to be:
- Swap out PSU (a friend of mine has a brand new Cooler Master MWE 650 v2 (MPE-6501-ACABW) that he's willing to let me try)
- Swap out GPU (I have my old Gigabyte 1050ti OC to try)
- Reinstall Windows?
I would greatly appreciate it if anyone here has any other ideas for things to check, or past experiences to share on how to resolve this. Thanks in Advance.
Facebook
Twitter
Instagram