display port Display port causes system to shutdown in some scenarios like gamming

[RageousRaccoon]

Hi all.

 

I have my system for last years with everything working fine until some weeks ago when my computer starts to shutdown without BSOD or any other kind of warnings like tearing, dead pixels, sound or visual glitches, or anything similar. After some weeks doing some minor tests, I've finally find out that the DP from my GPU is causing the problem.

 

By whatever reason, if I use the DP for literally anything that is not gaming, the system works fine. I'm using my PC last days to build a 3D portfolio and Blender and other applications works flawless even with pretty heavy loads like 170M of vertex in a sculpt model, but if I try to run a game like Banishers: Ghosts Of New Eden, the system just shutdown after some moments, the same with some other games, even indies with PS3 graphics, by any reason, some games don't shutdown the system, like The Callisto Protocol, half of campaign done and no problems, not even a tearing during game play.

 

Use another DP don't cause the issue, convert the problematic DP signal to HDMI with a active converter from DP to HDMI makes the system works fine (at least it seems like it), image and audio. Use another DP cable in the problematic DP cause the issue.

 

R9 3900X

32GB RAM Corsair CMW32GX4M4D3600C18

RADEON 7900XTX 24GB Founder's Edition

ASUS TUFF X570

WD BLACK

XPG Gaming, something 70, I don't remember the right model now.

DELL UPD2716D

 

No errors like BSOD, tearing, flickering, dead pixels or sound issues, or anything else.

Different configurations in Radeon Softwarre has no impact

There is no symptoms of any problem during PC usage
System only have issues with games and video renders in Blender, this one some times occur, and some times not, but I'm only rendering 360 frames turntables at this moment.

 

I've discarded the followings issues, at least until I'm writing this post, after some tests.

• RAM: any configuration of RAM clocks, auto with lower latency and low clock, 15 CAS with 2133 MHZ if I remember it right, XMP (DOCP in this mobo) with 18-22-22-42-93 with 3600 MHZ or 4000 MHZ with same CAS causes the problem, Windows MDSCHED reported no issues with RAM, even with tests at 4000 MHZ.
• I've done a CPU under volt to 1.25 since I've bought it, and never have any issue, change this config to auto or something else doesn't fix.
• Power Supply, Davinc Resove Studio (paid version that unlock GPU rendering) can run freely with a high power demand, for testing purpose, I've rendered a 4K 60 FPS 60.000 CBR with no problems even with 100% GPU usage over one hour, Alan Wake 2 last year uses something like 300 W in GPU during full game play, also a high demand on CPU and nothing happens, but I have no issues in December when I play it, it started some weeks ago.
• It's not a third party issue, like Process Lasso, or a driver install issue, or a windows tweak issue, since I've made a clean install twice to check it.
• It's not overheating, anything in system even in hot days still under 60º Celsius, only some 1~2 seconds peaks above it, bu fans bring it down very fast.
• The cables, I've switched my current cable (3 meters) to the older DP to Mini DP (1.8 meters) in the problematic DP and the problem persist.
• Problems like dusty or similar, I've cleaned up the entire PC with isopropyl alcohol, changed thermal paste, and thermal pads for the "maximum gaming performance" pads, also checked things out of place and poorly fitted parts.
• I've made some tweaks in Windows 11 like disable Copilot, Cortana, bing search, but nothing related to hardware, like the old "enable L3 cache for performance improvements" (urban legend from 2000's)

LITTLE UPDATE:

 

I've decided to try run some tests again, it seems that my second DP just takes much more time to trigger the error, something like four to five times, but both DP trigger it. After two hours playing on HDMI, no issues to report, even in 4K with maximum settings, and it seems that uses HDMI with DP trigger the error faster than use only one at time.

 

[emosun]

4 minutes ago, RageousRaccoon said:

Use another DP don't cause the issue

so just not using that port solves the problem?

[OddOod]

Sounds like yet another AMD driver bug

 

 

[RageousRaccoon]

38 minutes ago, emosun said:

so just not using that port solves the problem?

It's a turnaround, but I'm trying to figure out what is the problem, if this is a software issue I can try to report, if it is a hardware problem I may understand what cause it and avoid the same behavior in future.

 

27 minutes ago, OddOod said:

Sounds like yet another AMD driver bug

 

 

I though the same, but, if this a driver or software issue, the problem should happen in all DP instead of only one, or by any reason it's problem to a driver issue address only one DP?

[emosun]

4 minutes ago, RageousRaccoon said:

but I'm trying to figure out what is the problem

well you're trying to do firmware level diagnostics which you aren't equipped for.

if using a different port solves the problem , then the issue is either at the hardware level (such as a surface mount component being broken/damaged/manufactured wrong) , or it's a firmware issue which you of course will not have any access to the tools or code or knowledge it takes to modify the firmware.

anyone who works with enough pc hardware will tell you that if using a different port fixes it , then use a different port . because figuring out WHY it broke is a job for the manufacturer , not you.

[RageousRaccoon]

3 hours ago, emosun said:

well you're trying to do firmware level diagnostics which you aren't equipped for.

if using a different port solves the problem , then the issue is either at the hardware level (such as a surface mount component being broken/damaged/manufactured wrong) , or it's a firmware issue which you of course will not have any access to the tools or code or knowledge it takes to modify the firmware.

anyone who works with enough pc hardware will tell you that if using a different port fixes it , then use a different port . because figuring out WHY it broke is a job for the manufacturer , not you.

You're almost completely right, but there is minor issues caused by external events or by the user that can be avoided in future.

 

One external event, and pretty rare happens with regular RAM sticks for example, ECC have a error correction but regular DDR4 not, if you take some time to research, cosmic radiation can damage data, knowing this, you can prevent data loss choosing what kind of hardware you're gonna use, off course, this something for large data centers with sensitive data, regular users probably will never notice such problems, same with NTFS replacement for ReFS some point in future, but in my case, know if some kind of electrical failure like a city blackout or a sudden power loss can cause such thing, can help me to find out a way to avoid it, or a way to test the cables or something.

 

You're right, when the diagnostics point out that the failure is caused by software, off course only M$ can fix a code error that leads to a BSOD and users not, but users can avoid certain types of behavior, or try to identify minor hardware problems, like bad cables and etc. Also identify the software error and report it, may help to fix it, and that's what I'm trying to do. I'm not American so my natural language is not English, in case I can be understood like a rude person, my anticipated apologies, it's not my intention.

[emosun]

2 hours ago, RageousRaccoon said:

You're almost completely right, but there is minor issues caused by external events or by the user that can be avoided in future.

Your own diagnostics show that the variable is the port you use. So that's the variable.... the port being used.

so it's either the port (the hardware) or a software issue related to that port , which nobody here or you is going to diagnose.

your options for fixing it are , use the working port , or return/replace the hardware , or use totally different software/firmware that doesn't have the issue.

[RageousRaccoon]

After a long research due to problem persists in all ports, I've found a application called OCCT, used to stress tests for people that plays with overclock and etc.

 

During the stress tests I've noticed that the HOT SPOT (HS) was reaching 100º C or above, so after some point like 115º or 120º the the system shutdown to avoid permanent damage to the components.

 

In this video Jay-Z Two Cents have a issue where the thermal pads create a big volume of mass, when he put everything together, this volume creates a angle between the GPU processor and card metal plate, what causes a temperature increase.

 

In my case, the reference card uses thermal putty in the RAM, what gives a perfect contact between the RAM and the vapor chamber, GPU processor is also flat, when I replaced it with the thermal pad, it creates the same angle and also increase temperature, something I don't noticed when I put everything together, it can be trick even for experienced people like him.

 

After change the thermal paste to Thermal Grizzly Kryonaut Extreme and uses a thermal putty, 12.8 WMK like the old thermal pad, system has been repaired.

 

The delta temperature of the system (hot spot temp - gpu temp = delta) when I created the post, it would hit something like 40º C or above, now in Horizon Forbiden West in 4K with all settings in maximum in a open environment, the max peak was 15º C and usually 9º in closed environment or another games like Alone In The Dark 2024.

 

Some games, reach high HS temperatures pretty fast and some don't, by any reason, Dragons Dogma and Callisto Protocol, don't push the card to hard, so the system don't shutdown, same for Davinci Resolve, another titles like Alone In The Dark 2024 or the PS3 graphics indie game, does it.

 

By any weird reason, the first DP causes the system to crash faster than the second DP or HDMI, maybe just a nonsense coincidence, maybe not. I also found this video from Ancient Gameplays where he talks about a reddit post where some users report having problems with a bad DP cable, changethe bad cable for a good one, seems to improve overall temp, the video is not a test, so there is no technical answer, maybe a April's foll joke, maybe not.

 

Sorry my misunderstood about the situation and thanks fro trying to help.