Question GPU intermittently not detected on boot - Multiple restarts to fix

[biolizade]

Hey,

I've been having this issue for a while that I have been failing to diagnose myself. My GPU on a boot will often not be detected at all (No video output).

Usually on the first boot of the day I'll get the motherboard splash screen, and then lose video signal before windows boots. Subsequent restarts produce no video signal at all until seemingly randomly the GPU is detected properly and PC works fine.

The PC is booting successfully, just without detecting the GPU and giving a video output (I can hear windows starting up chimes).

It takes multiple restarts (Holding power button to turn off, starting up again) before I can finally get the GPU detected and a video output. It feels like waiting ~1m between reboots has a higher success rate, but that may just be coincidence.

Once the PC has booted with the GPU correctly working, it works fine and can run at full capacity with demanding games without issue until the next restart. This issue only occurs on a boot, or awaking after sleep (if the PC has been asleep overnight for example).

This issue first appeared around 18 months ago, and would occur on every boot without fail (with me spending every morning restarting to finally get a successful boot).
After a windows update, the issue disappeared completely and the PC would boot successfully every single time, leading me to believe it was a software issue. This may be a coincidence?

In the past few weeks, the issue has crept back again (without changing anything internally with the PC). First only occurring every few days, but now occurring every single boot.

Plugging the DisplayPort cable directly into the motherboard and booting does work successfully, and I can see that the PC works fine but does not detect a GPU installed at all.

My full specs (All purchased in 2019, excluding the RAM/970 SSD which were 2021)
CPU: Intel i7 9700k (Not overclocked)
CPU cooler: NZXT Kraken X52
Motherboard: ASRock Z390 Phantom Gaming-ITX/ac Mini ITX LGA1151 Motherboard
Ram: 32GB Crucial Ballistix 3000MHz CL15 BL2K16G30C15U4W
SSD 1: Samsung 970 EVO Plus 1 TB PCIe NVMe M.2
SSD 2: Western Digital Blue 1 TB M.2-2280 SATA
GPU: EVGA Black RTX 2080 Ti
GPU Cooler: Corsair Hydro Series H55 mounted with NZXT G12 bracket
PSU: EVGA SuperNOVA 750W G3 80+ Gold
PSU Cables: PHANTEKS Extension Cable Set (Daisy chained w original PSU cables)
Chassis: NZXT H200 Mini ITX
OS: Windows 11
Monitor: Acer Predator X34P

Things I've tried:

• Reseating GPU, RAM, SSD's
• Swapping displayport cables
• Booting with all USB devices unplugged
• Reinstalling GPU drivers using DDU
• Updating all drivers / windows updates
• Reseating GPU power cables
• Removing braided PSU cable extensions
• Trying different VGA power sockets on the PSU for both 8-pin cables (VGA3&4 rather than 1&2)
• Removing kettle lead power cable and holding power button to drain before boot
• Updating BIOS
• Resetting CMOS
• Fresh install of Windows

Things I've not tried:

• Testing the GPU in another system (My GPU is cooled with an AIO waterblock using NZXT G12, so this isn't simple, although I expect it to be the only next step?)
• Testing another PSU / Motherboard
• ????
  

Seems to be a hardware issue at this point, but in that case how did the issue fix itself 18 months ago and disappear only to start becoming an issue again the last few weeks? When I've not changed anything internally or even opened the case other than for cleaning.
If it is a hardware issue then why, when finally the PC boots with a detected GPU, does everything work completely fine and runs without issue until the next boot?

 

[Aeacus]

but in that case how did the issue fix itself 18 months ago and disappear only to start becoming an issue again the last few weeks? When I've not changed anything internally or even opened the case other than for cleaning.
If it is a hardware issue then why, when finally the PC boots with a detected GPU, does everything work completely fine and runs without issue until the next boot?

A long shot, but one possibility would be too high mounting pressure on GPU die.

You run custom water block on it. That alone differs from the original cooling method GPU received from factory. Could be that the waterblock adds too much pressure on GPU die (or memory controllers), whereby GPU won't work. Once the pressure shifts a little (or is released a bit), GPU works fine.

Too high mounting pressure on CPU can cause all sorts of issues. I see no reason why GPU would be any different.

So, i suggest that you remove NZXT Kraken G12 completely and put the original heatsink back.

 

[biolizade]

A long shot, but one possibility would be too high mounting pressure on GPU die.

You run custom water block on it. That alone differs from the original cooling method GPU received from factory. Could be that the waterblock adds too much pressure on GPU die (or memory controllers), whereby GPU won't work. Once the pressure shifts a little (or is released a bit), GPU works fine.

Too high mounting pressure on CPU can cause all sorts of issues. I see no reason why GPU would be any different.

So, i suggest that you remove NZXT Kraken G12 completely and put the original heatsink back.

Good shout. The stock cooler is completely choked in the case so is non viable for use as it hits 88c in about 2 minutes lol, but I'll give it a go for testing the issue in the next few days.