Question Randomly gpu not recognized at boot

[Momo__owoW]

Hi!
I just built a pc...

The specs:
GPU: MSI 4070 Super Ventus 3x
CPU: Ryzen 5 7600x
MoBo: B650 Eagle
RAM: 2x16 DDR5 Corsair Vengenace ram 6000hz
SSD: nvme wd_black 1tb
PSU: MSI A750GF
CASE: MSI Gungnir 110R
SO: W11 Pro

I also have 2 monitors connected to my GPU, one via HDMI/HDMI cable, the other one via HDMI/DISPLAY PORT cable.

The issue:
In a total random way, at boot, sometimes (let's say 1 over 6-7 boots) the GPU is not recognized.
The rgb on my peripherals (mouse/keyboard) are on, fans are spinning (GPU fans as well), but nothing is shown on the monitors.
I can hear the windows sound from my speakers tho.
The red led on my mobo shows "VGA".
The issue goes away if I restart the pc via reset button on my case. Never happened to have a black screen if I reset the pc that way.

What I tried:
- updated MoBo BIOS (F3)
- cleaned drivers with DDU in safe mode and installed both older and latest drivers again
- reinstalled W11 from scratch
- checked cable connections on GPU (the GPU is connected using 2 8 pin cable -no daisy- to the adapter I found in the box)
- unplug (when black screen happens) one or both the HDMI cables
- plugging one cable to the MoBo directly (and in that case I can see everything but the GPU is then not recognized by any program I have like HWinfo/Monitor)
- install some drivers from the MoBo page like the AMD chipset one
- unplug all the USB peripheral when the issue appears

What I think:
I will exclude a power issue, since the PSU is new (all pieces are brand new from Amazon) and I tested a lot the GPU during those days not only running benchmarks like Cinebench 2024, but also by playing demanding games like Cyberpunk 2077 with Path Tracing on, Dragon's Dogma 2 maxed out, etc. I played for hours with no issues at all, not a single crash or weird temp value (always monitoring).
The only weird response I had from a program was from OBS. Sometimes the GPU usage, normally around 10% (I do use Nvidia Broadcast mostly for the "background blur" feature), goes up to 60-70% in IDLE for no reason.
It only goes away if I restart the pc, going back to a normal value, but then when I again shut it down and boot then there is a high chance I get the "black screen" issue. I don't really understand if those are related somehow.

I also noticed a weird thing. The first thing I do in every pc is to turn off fast boot from OS settings, because I had issues with my old pc. Even both the fast boot and hybernation are off sometimes the pc stays "on" even after shutting it down (it do it very rarely, but it happened). I also noticed that it kept (randomly) rebooting itself after shutting it down. I think I solved this issue by changing a value in the registry editor that was set on "0" (about power at shut down), maybe something changed after the latest W11 update, idk.

Anyway, the random black screen(s) at boot still persist. I can shut it down/reboot like 6-7 times without any issue, or I can have like it happing twice in a row.

Then I noticed something related to Razer Synapse: sometimes when the pc "rebooted" itself after a shut down, or after staying "on" even if I shut it down, the Synapse program had an update. Didn't notice if it was always the case, but surely it did it 2-3 times. Now I stopped the automatic updates tho.

Also, when the pc boots up it usually gives power to my peripherals first (mouse/keyboard/mouse charger/hub) and then to the first monitor where the BIOS splashpage should be visible. Like in a chain. If it doesn't turn on immediately after my keyboard then it won't at all.

The bad thing is that this issue doesn't happen every time, so I can't check properly if there is a conflict between hardware or not (like, idk, trying to boot it with only m/k). It seems totally random (the only thing that seems to have a "regularity" is the OBS issue with the GPU usage, that seems to appear close to the moment I will receive a black screen at boot).

I also had some BSOD (didn't have time to read the error...) before reinstalling W11. Like 2 of them, right after some of the "I turned off the pc but it is still powered" moments.

p.s. about the BIOS setting I didn't touch anything BUT the XMP profile. The first time I used that, then I updated the BIOS again (the first time I installed F2 since F3 was in beta) and since it reverted it to no XMP I then chose the other one (the AMD one, don't remember the name now... anyway, same result).

What I think could be the culprit (but I will be too upset to find...):
- faulty GPU: it works like a charm with every demanding task, it would be absurd to fail at boot...
- faulty PSU: should be powerful enough (also Gold+ certified...) so, idk...
- incompatibilty to XMP/theotherone profile: then I will have my RAM not at its best...
- faulty MoBo: I don't really know how to test it
- maaaaybe the DISPLAY PORT cable? Since the BIOS splashpage is shown on that monitor (I switched the cables, the BIOS switched as well ofc)... but isn't it weird that nothing is shown on the HDMI one (like the windows login screen, that was shown on the other monitor usually)?
- the one cable I didn't connect from the case: the case comes with some rgb fans, and has a board on the back to control them. The board also has a sata port to connecto to the PSU for the power and another cable that I left out, that should be connected to the MoBo in order to controll the rgb via software (and not only via the case front panel like I am doing now). I don't really think this could be the culprit but I prefer to be clear about what I have done with my build...

Thanks 🙁

Edit1:

Right now again it stayed “on” (lights on, fans spinning, etc) after shut down. I hade to reboot it and then shut down again.

This is getting ridicolous.

 

[Ralston18]

This motherboard?

https://download.gigabyte.com/FileL...1002_e.pdf?v=6374658e7c527373e6a080ff09835a5c

[Verify that I found the applicanble User Manual.]

= = = =

My thoughts and suggestions (no particular order):

1) The PSU, just because it is new, cannot be excluded, Likewise for any other component.

2) PSUs provide three different voltages (3, 5, and 12) to various system components. So some components/LED's etc. appearing to work is not a sign that the PSU is fully functional.

3) Drivers: manually download drivers directly from the applicable manufacturer's website. Reinstall and reconfigure. No third party tools or installers. Also verify that the source website is truly the manufacturer's website. Just because the manufacturer's name appears in the pathname/URL does not mean that the website is the manufacturer.

4) Intermittment/random problems tend to be a loose connection. May come and go with expansion/contraction/vibrations. Refer to the applicable User Guides/Manuals (starting with the motherboard) to doublecheck that all connections are correct and, more importantly, fully and firmly in place. Do the same for all components. Cards, RAM, Jumpers, and case connections. Case connections can be confusing and are easily mixed up.

5) Work/test with only one monitor for awhile. Determine if the problem stops or changes in some manner. Then swap monitors and test again.

6) Remember that during shutdown Windows may be doing some updates or other "housekeeping" activities. Meaning that Windows may not immediately shutdown. If you force a shutdown then that can lead to file corruption and data loss.

7) Stay out of the registry. Registry edits are a last resort and should be done (if at all) only when such edits are clearly known to resolve a specific problem. And editing should only be attempted after a full system backup that includes the Registry itself.

8) Take a look in Reliability History/Monitor for any error codes, warnings, or informational events being captured just before or at the time the monitor failures (random black screens) occur.

9) Set aside "past issues". Many things have changed so what may have been problematic before will work just fine now. Plus there are likely some dependencies on having various options and configurations enabled.

10) Reading back: "The board also has a sata port to connecto to the PSU for the power and another cable that I left out, that should be connected to the MoBo in order to controll the rgb via software (and not only via the case front panel like I am doing now). I don't really think this could be the culprit but I prefer to be clear about what I have done with my build..."

Install/connect the cable. May or may not be the culprit but troubleshooting includes eliminating possible issues.

= = = =

Reread all documentation. Pay attention to the fine print. Visit manufacturer websites for more up-to-date information, FAQs, and Forums.

Take your time, be methodical, keep notes, Change only one thing at a time and keep track of what was changed, where it was changed along with the original and new value.

 

[Momo__owoW]

This motherboard?

https://download.gigabyte.com/FileL...1002_e.pdf?v=6374658e7c527373e6a080ff09835a5c

[Verify that I found the applicanble User Manual.]

= = = =

My thoughts and suggestions (no particular order):

1) The PSU, just because it is new, cannot be excluded, Likewise for any other component.

2) PSUs provide three different voltages (3, 5, and 12) to various system components. So some components/LED's etc. appearing to work is not a sign that the PSU is fully functional.

3) Drivers: manually download drivers directly from the applicable manufacturer's website. Reinstall and reconfigure. No third party tools or installers. Also verify that the source website is truly the manufacturer's website. Just because the manufacturer's name appears in the pathname/URL does not mean that the website is the manufacturer.

4) Intermittment/random problems tend to be a loose connection. May come and go with expansion/contraction/vibrations. Refer to the applicable User Guides/Manuals (starting with the motherboard) to doublecheck that all connections are correct and, more importantly, fully and firmly in place. Do the same for all components. Cards, RAM, Jumpers, and case connections. Case connections can be confusing and are easily mixed up.

5) Work/test with only one monitor for awhile. Determine if the problem stops or changes in some manner. Then swap monitors and test again.

6) Remember that during shutdown Windows may be doing some updates or other "housekeeping" activities. Meaning that Windows may not immediately shutdown. If you force a shutdown then that can lead to file corruption and data loss.

7) Stay out of the registry. Registry edits are a last resort and should be done (if at all) only when such edits are clearly known to resolve a specific problem. And editing should only be attempted after a full system backup that includes the Registry itself.

8) Take a look in Reliability History/Monitor for any error codes, warnings, or informational events being captured just before or at the time the monitor failures (random black screens) occur.

9) Set aside "past issues". Many things have changed so what may have been problematic before will work just fine now. Plus there are likely some dependencies on having various options and configurations enabled.

10) Reading back: "The board also has a sata port to connecto to the PSU for the power and another cable that I left out, that should be connected to the MoBo in order to controll the rgb via software (and not only via the case front panel like I am doing now). I don't really think this could be the culprit but I prefer to be clear about what I have done with my build..."

Install/connect the cable. May or may not be the culprit but troubleshooting includes eliminating possible issues.

= = = =

Reread all documentation. Pay attention to the fine print. Visit manufacturer websites for more up-to-date information, FAQs, and Forums.

Take your time, be methodical, keep notes, Change only one thing at a time and keep track of what was changed, where it was changed along with the original and new value.

Hi.
Yes, this is the motherboard.

1-2) how do I know if the psu is funcional? Under huge stress everything works well, I don't think that the gpu need more power to boot than to run games like Cyberpunk with path tracing on.

3) I installed every driver directly by the official site (nvidia and the mobo ones).

4) I tried to reseat the gpu and looked like it was the issue until the black screen appeared again. Also it was pretty straigthforward to connect everything else tbh, even the front panel cables went ready to connection (no single cables, just one big chunk for audio and another one for the buttons/leds). I will re-check them if other software-related solutions will be non useful...

5) tomorrow I will try that

6) I know that but I think I waited a fair ammount of time. But I have to say that sometimes I got some BSOD (last one said "critical_process_died", but let me reboot and sing in as always). I also did a sfc /scannow and the health check as well, no results

7) the value I edited didn't seem to did anything tho. It was the "powerdownaftershutdown" value, but I got another random reboot-after-shutdown once after that... and that will lead me to another idea: what if is a MoBo issue? In one of the threads I have read about that registry edit a user said that, among other issues, the MoBo was the culprit for the reboots. Could it be my case? I am reading a lot of stuff about the B650 MoBo (not necesarry my model that it is pretty recent) that concerns me.

8) in reliability history there are some critical errors (even hardware ones) but those are not related to every black screen issue I had. What do you mean with "Monitor"?

10) tomorrow I will try to plug it in

-------------

Meanwhile I tried to disable XMP, no effects. Disable synapse on startup and deleting nvidia broadcast. Nothing happened.


I am starting to wonder if this is a true issue... I mean, I can still use my pc I just have to reboot when the screen are blacks... I just don't know if it can be "harmful" for the pc itself.

Maybe I could try to send back both the GPU and the MoBo but I am also strating to think that maybe those two are not compatible? (I am reading that the MSI version of the B650 has an issue with the BIOS with some 4070 cards... maybe it is the same for the Gigabyte ._.). And without knowing if could be the PSU I don't really know if it is a good move... (I bought the parts one week ago, I should have 3 more weeks to send them back to amazon).

 

[Ralston18]

"Monitor": Reliability History/Monitor are the same thing with different names appearing.

Another tool that can be used is Event Viewer. Requires more time and effort to navigate and understand.

How To - How to use Windows 10 Event Viewer | Tom's Hardware Forum (tomshardware.com)

Windows has many process dependencies. Some other process may be the issue. As a result any dependent processes can fail and lead to a BSOS/black screen.

Registry edits can solve problems or perhaps appear to do so. Then some update or other change finds an unexpected registry entry and generates an error, warning, or informational event followed by a crash. Or if no crash may simply change the registry value to the default or some other value relevant to the update in progress. Possibly undoing a real workable fix for some issue.

I would hope that the MSI A750GF PSU is enough to keep up with peak gaming demands. I will defer to the gaming community to comment accordingly, Wattage is important but so is quality.

 

[Momo__owoW]

"Monitor": Reliability History/Monitor are the same thing with different names appearing.

Another tool that can be used is Event Viewer. Requires more time and effort to navigate and understand.

How To - How to use Windows 10 Event Viewer | Tom's Hardware Forum (tomshardware.com)

Windows has many process dependencies. Some other process may be the issue. As a result any dependent processes can fail and lead to a BSOS/black screen.

Registry edits can solve problems or perhaps appear to do so. Then some update or other change finds an unexpected registry entry and generates an error, warning, or informational event followed by a crash. Or if no crash may simply change the registry value to the default or some other value relevant to the update in progress. Possibly undoing a real workable fix for some issue.

I would hope that the MSI A750GF PSU is enough to keep up with peak gaming demands. I will defer to the gaming community to comment accordingly, Wattage is important but so is quality.

The PSU should be a quality one as I see from reviews, the brand and the gold certification. At least I hope so lol

Anyway I did a test with the pc-monitor cables today.

I did 10 boot cycle for each test (until I got the error).

1) only HDMI cable to 1 monitor: 10 boot - 10 success, no black screen
2) 2 monitors, 2 cables, same HDMI one and the other one a HDMI/D.port cable (different from the one I was using before). At the 2nd boot I got the black screen and the pc remained "alive" at the shut down
3) 2 monitors, 2 cables, same cables I always used (so I again switched the D.port one) BUT I changed the port on the GPU (have 3 of them, took the middle one). 10 boots again, no black screen at all but I got 1 "pc alive at shut down" and 2 "auto reboot at shut down".

I did an integrity check with the command DSIM.exe (etc), took some seconds and said "the operation completed successfully. sfc /scannow found nothing.

As you told me I was methodic and this time I took note of the time for every boot so I can check in the even viewer. I just don't now where to look, if you can help me I will post the results.

Now, if I "solved" the black screen issue, I just need to get rid of those weird pc behavior when ti doesn't shut down properly when I say so.

 

[Momo__owoW]

Nope, it did it again. I am going crazy.

On another forum someone said I chose the worst 4070 super and that my monitors (C27F390FHU and S24D300H) are too bad for my gpu.

He also said something like "you MUST buy a 460 ti and two 144hz monitor instead" but I don't need them lol... a 4060 Ti can't run Cyberpunk with path tracing at 1080p, I don't understand why I should downgrade it.

Anyway, is it possible that an old monitor is causing my issues? Because I can just keep the 4070 and then change monitors on a second time.

 

[Ralston18]

"Anyway, is it possible that an old monitor is causing my issues?"

Continue being methodical. Figure out some testing process where everything remains the same except for the presence or absence of the old monitor.

If possible borrow another known working monitor that can be swapped in.

= = = =

As for "worst" - ask for some objective sources, citations, reviews, etc. to support "too bad" or "you MUST" type statements.

Especially if downgrading is being recommended or considered. Sometimes downgrading is a solution but only if there are known reasons for doing so. Maybe, for example, as a temporary fix until some buggy or corrupt code gets fixed....

= = = =

As for Event Viewer (run as Admin) I would start by looking in Windows Logs > System.

However, for the sake of remaining methodical and perhaps discovering more information, begin with Application and work down the listed logs.

Then do the same working through the Applications and Services Logs. Some of the logs may not be applicable - still look just for elimination purposes. Some logs could be empty.

Doing the above will 1) help with understanding the nature of the logs and 2) becoming familar with the information provided by each log.

Which is important because, if the logs are lengthy and nothing is found or noted, the next step will be to clear the logs and repeat the testing to date. Making newer events (of any sort) more readily apparent and discoverable.

 

[Momo__owoW]

"Anyway, is it possible that an old monitor is causing my issues?"

Continue being methodical. Figure out some testing process where everything remains the same except for the presence or absence of the old monitor.

If possible borrow another known working monitor that can be swapped in.

= = = =

As for "worst" - ask for some objective sources, citations, reviews, etc. to support "too bad" or "you MUST" type statements.

Especially if downgrading is being recommended or considered. Sometimes downgrading is a solution but only if there are known reasons for doing so. Maybe, for example, as a temporary fix until some buggy or corrupt code gets fixed....

= = = =

As for Event Viewer (run as Admin) I would start by looking in Windows Logs > System.

However, for the sake of remaining methodical and perhaps discovering more information, begin with Application and work down the listed logs.

Then do the same working through the Applications and Services Logs. Some of the logs may not be applicable - still look just for elimination purposes. Some logs could be empty.

Doing the above will 1) help with understanding the nature of the logs and 2) becoming familar with the information provided by each log.

Which is important because, if the logs are lengthy and nothing is found or noted, the next step will be to clear the logs and repeat the testing to date. Making newer events (of any sort) more readily apparent and discoverable.

The guy on the other forum keeps repeting that, since I have 60hz monitor, I am "dumb" (kinda) and I should downgrade or buy new montiors.
That I need gsync and that if the gpu gives an output of (like) "300" I am doing wrong with my monitors causinc a total desync or something similar.

But shouldn't this happen every time I boot the pc then?

Anwyay I don't have any good monitor to swap (not even a 4ktv, everything I have is 10 yers or more older...).

What I did now (reading infos online about similar issues) was to remove the "Resizable bar" in bios, reading that could cause instability issues. I think it is worth a try.

 

[Ralston18]

If the other guy in the other Forum is insulting you and making recommendations without any supporting rational then I will leave it to you determine or otherwise weigh in the value of those post(s).

= = = =

As for me I do not know what the problem is. Perhaps some error of omission or commission on my part.

And I would continue troubleshooting....

= = = =

What I do know is that that removing the Resizeable Bar may or may not be a solution. Others may post accordingly one way or another.

Try to find more information about removing Resizeable Bar. Post a couple of the links that you are reading.

I found this link:

https://www.nvidia.com/en-us/geforce/news/geforce-rtx-30-series-resizable-bar-support/

May or may not be applicable.

 

[Momo__owoW]

Resizable bar did nothing, another thing I am excluding. I would not dare to touch anything else in BIOS tho. I have read something about setting "legacy" status for PCIe slot but I don't really feel confortable to do that atm.

I also did a check on the reliability monitor at the timing I got the black screen: aside from some critical errors about "windows not shut down properly" I had nothing else (oh yeah, some issues about the Cyberpunk 2077 launcher lol...). Maybe because I had to reboot it.

In the windows event viewer under the "windows logs - system" tabs I see nothing "critical" (red) around those timings.

Do you suggest to check under other tabs?

p.s.
The guy didn't "really" insulted me, just said that what I did was kinda dumb (so maybe a bit of an insult... yeah).

Anyway he was totally sure about one thing: if I hear windows running (and it runs) then the issue is not the GPU but the cable/monitor 100%. Like the driver not recognizing the (horrible, horrible...) monitor(s) I have.

In fact is something I can understand... the fans are spinning, the SO is running... it looks like a LOT about a problem about the gpu "talking" with the monitor(s).

What do you thing about that? Like "if the issue was about your hardware the SO shouldn't even boot". Seems to good to be true (that means I just have to change my monitor to solve the issue lol).

And since I found nothing (at least where I looked...) in the event logs could it be possible that, since the issues are the monitors, that's why I'm not finding anything "pc" related?