brand New build - Fresh Windows Install - Hard Crashing, non events

Hello Folks,

While I've built my computers since I was a teen and this latest one is kicking my ass.

Here is the part list

Attached is my DXDialog.

Synopsys
This system randomly crashes with a black screen which produces zero useful events in the Windows event viewer. On reboot, half the time memory training takes so long or doesn't work that I restart the system a couple of times to get it back into Windows.

Situation so far
Just today I reinstalled Windows the clean out anything software-related that would be causing me this grief, and oddly my crashing has become WORSE.
For example, I'm patching in the games I typically play, 1 or 2 downloads at a time. CPU temp is fine, network seems fine and fast (it is an Intel Killer network card) BAMM black screen...

I don't have an extra AM5 CPU, MOBO, or DDR5 to swap in and out for testing purposes, though I can say that this PSU and GPU lived in my previous build, a 1200k@5.1GHz which didn't have this grief, so I'd like to think that the PSU and GPU are in good working order.

Would someone happen to have suggestions on what software I can leverage to test certain components or generate error event logs so I can hopefully resolve these issues?

Thanks!

Hello @_kaurus and welcome to ElevenForum.

You need to take a screen shot of your parts list... we can't see it.

To post a screenshot on this forum...​

1. Take the screenshot, edit it as desired, save to your desktop as .png or .jpg (max. size = 2MB)
2. Left click in your post, where you want to place the screenshot.
3. Click the Attach Files button at lower left, browse to your pic and click Insert.
4. After clicking Insert, I usually choose the Thumbnail (smaller) option, or the Full Image (larger) option.

I see in DxDiag.txt:

NVIDIA GeForce RTX 4090:
Current Mode: 3840 x 2160 (32 bit) (120Hz)
Generic PnP Monitor :
Native Mode: 3840 x 2160(p) (60.000Hz)

Try specifying in NVIDIA Control Panel
2160p, 60Hz

abactuon said:

I see in DxDiag.txt:

NVIDIA GeForce RTX 4090:
Current Mode: 3840 x 2160 (32 bit) (120Hz)
Generic PnP Monitor :
Native Mode: 3840 x 2160(p) (60.000Hz)

Try specifying in NVIDIA Control Panel
2160p, 60Hz

Click to expand...

@_kaurus, don't know if it makes a difference, but could your "Current Mode: 3840 x 2160 (32 bit) (120 Hz" being 32 bit be a problem? I know pretty much everything in today's computers is 64 bit. Just a WAG . . . I don't know all that much about the hardware.

@_kaurus
According to dxdiag there is problem with gcc.exe producing "access denied" errors, probably has something to do with precompiled headers.
Are you using gcc when this problem happens?

_kaurus said:

For example, I'm patching in the games

Click to expand...

It's possible that your patcher uses gcc but is having some issues creating precompiled headers.

To quickly test if this is true, try stop using the patcher to confirm it's causing the problem.

Wynona said:

@_kaurus, don't know if it makes a difference, but could your "Current Mode: 3840 x 2160 (32 bit) (120 Hz" being 32 bit be a problem? I know pretty much everything in today's computers is 64 bit. Just a WAG . . . I don't know all that much about the hardware.

Click to expand...

"32" bit is normal for graphics. It's 24 bit RGB color plus an 8 bit alpha channel (not that I have any idea what an "alpha channel" might be).

If the OP is running GCC, I suggest getting rid of that. I have found it to be buggy.

bobkn said:

not that I have any idea what an "alpha channel" might be

Click to expand...

Alpha channel is used to specify level of transparency

zebal said:

Alpha channel is used to specify level of transparency

Click to expand...

I've seen that before, but my monitor has never seemed to become transparent.

bobkn said:

"32" bit is normal for graphics. It's 24 bit RGB color plus an 8 bit alpha channel (not that I have any idea what an "alpha channel" might be).

If the OP is running GCC, I suggest getting rid of that. I have found it to be buggy.

Click to expand...

Thanks for the clarification, @bobkn.

Ghot said:

Hello @_kaurus and welcome to ElevenForum.

You need to take a screen shot of your parts list... we can't see it.

To post a screenshot on this forum...​

1. Take the screenshot, edit it as desired, save to your desktop as .png or .jpg (max. size = 2MB)
2. Left click in your post, where you want to place the screenshot.
3. Click the Attach Files button at lower left, browse to your pic and click Insert.
4. After clicking Insert, I usually choose the Thumbnail (smaller) option, or the Full Image (larger) option.

Click to expand...

Thank you for the welcome.

I don't tend to find myself on these types of community forums since try to be self-sufficient, but enough is enough. Do you know? I need help. lol

zebal said:

It's possible that your patcher uses gcc but is having some issues creating precompiled headers.

To quickly test if this is true, try stop using the patcher to confirm it's causing the problem.

Click to expand...

Pardon me, can you elaborate on what is a GCC patcher?

Googling it, is a GNU C Compiler. Since i just installed windows an hour before posting here, I'm not sure how this would be on my system or why the system would be trying to access it. Do you have an idea?

Problem signature:
P1: GCC.exe
P2: 23.9.28.1
P3: bc7a15b3
P4: StackHash_9083
P5: 0.0.0.0
P6: 00000000
P7: c0000005
P8: PCH_AE_FROM_KERNELBASE+0x0000000000064FFC


Since I have a Windows account, could this be showing up since I logged into my Windows account?

After I installed Windows, and it had a chance to download its stuff again, there were a bunch of icons on my taskbar that were not linked to any programs, so they each had to be re-downloaded.

Wynona said:

Thanks for the clarification, @bobkn.

Click to expand...

With HDR enabled, it should be 10bits per channel so 30 bit? I'm not sure what the extra 2 is for. , which the TV and GPU support once you enable HDR in windows settings.

Another interesting thing is PC part Picker suggests I'm bumping up to my Wattage, pretty hard, though the same hx850 supply was used in a nearly identical build, be it a 12700k which "should" be more power hungry than this 7800x3d.



Possibly I should replace the PSU...as i understand, if it's not keeping up you'll just hard crash.

So far my crashes have been rather strange, like Chrome or Brave browser will hang for a bit, doing nothing, but my mouse will still move around just fine.
Playing games, they will just freeze and the system kicks off to a restart (mentioned restarting takes a bunch of time).

• Yesterday I ran ADA65 total systems tress for 20 minutes, with no issues and zero errors.
• Overnight I ran MEMTest with Default JEDEC, with zero errors.
• while running default JECED and disabling C state and power down mode, the system doesn't even make it into Windows. It almost did once, but the screen had weird glitched-out lines, and back to failing to repost.
• I cleared CMOS, applied only XMP1 profile, and reloaded MEMtest, did that for 3.5 hours, with zero errors.

As far as I understand, Disabling CState for the system and power-down mode for the memory will decrease your efficiency, but it should be doable and increase your stability. While I'm not a memory OC guy, i typically just set XMP, and move on, this seems strange to m e.

FYI, if someone here was quite capable and able to do Zoom screen share with me, I'd be more than willing to pay for their time provided there was some sort of positive outcome.

My 5 years old has inherited my 12700k, and since the hx850 worked fine for it with all my drives and 4090, I bought an HM850 for her computer, so I do happen to have another PSU I can swap in just to be sure that on coincidence my hx850 isn't kicking the bucket (would be so sad, it's such a nice one).

_kaurus said:

Thank you for the welcome.

I don't tend to find myself on these types of community forums since try to be self-sufficient, but enough is enough. Do you know? I need help. lol
View attachment 81997

Click to expand...

IF you re-seat the RAM and still have the problem... check if the RAM is using XMP or DOCP. Might want to manually set the manufacturer's frequency, timings and voltage, instead.
Cause if the RAM isn't the cause, then...

I would guess that your power supply is not up to the job.
You're running quite a bit of hardware there, and most are recommending 850W minimum for the RTX 4090.

Ghot said:

IF you re-seat the RAM and still have the problem... check if the RAM is using XMP or DOCP. Might want to manually set the manufacturer's frequency, timings and voltage, instead.
Cause if the RAM isn't the cause, then...

I would guess that your power supply is not up to the job.
You're running quite a bit of hardware there, and most are recommending 850W minimum for the RTX 4090.

Click to expand...

Thanks for your suggestion. I have not been OCing the 4090, and I mostly run it frame capped at 120, and I mostly play WoW Classic and World of tanks, which neither are extremely difficult. for example, I get 270FPS at 4k, and I lock it to 120, so less than 50% GPU going on here.

I was thinking, it's quite possible that while this CPU is less power-hungry than a 12700k@5.1Ghz, it is likely letting more FPS get through, which means the GPU can generate more FPS, which means additional or bigger power spikes, and I bet you know where I'm going with that....

My timings are pretty generic, with like 89 latency..pretty bad, I was playing with it to get it down under 60, but the system was never stable so, now I'm here.



So, about the power supply, you'd think that after running AIDA full system "stability test" which is every drive (I have 3 NVME and 3 NAS), CPU, GPU, and memory to 100%, would show that the PSU is bad, though I did that for 20 minutes without issue...Then on the flip side, downloading one or two games at once onto an NVMe crashes me...

You're right, the symptoms seem to suggest that something is bad,

• be it the PSU which worked fine with the 4090 and the 12700k just 3 weeks ago
• the mobo (first-gen mobos don't seem so great across the board so it seems),
• the CPU
  • This one will post with IF set to 2200 which tubers say is not common and suggests it's a good sample.
  • I'm running stock ATM until I can figure out the instability.
• The RAM, Which so far seems to be okay at factor OC settings. I did apply my manual OC which is just manual settings from RealOverlocking Tuber, and then a memtest, and it blackscreened on me...
• Also, when I enable PBO, instability goes way up, and I was under the impression that PBO was AMD factory OC, how could that make instability go up...so strange.
  • Temps while running the AIDA full system stress test put the CPU at 75 degrees with PBO 85/-20mv set.

I will swap the PSU tomorrow, and game on it for Friday, without manual OC or PBO, etc. likely i should just measure what power draw I'm getting out of the wall, and that should guide me if i need a larger one.. hmmm.

While I've written quite a bit, i could use help with a troubleshooting game plan here, so i can check stuff off a list and get it done. Right now i'm sort of just shooting into the wind with running this, setting that, etc...

@_kaurus


If PCPartPicker has you at 889W with all your parts... you need a bigger power supply.
PCPartPicker's wattage total is fairly... conservative.

Also, AIDA64 Extreme's stress test is fairly mellow. A test like Prime95 is much, much more strenuous.

Lastly... black screens are usually caused by RAM or power supply, assuming the vid card, the cables and the monitor are functioning properly.

your PSU is to small recommended power supply for your 4090 is 850W https://www.gigabyte.com/Graphics-Card/GV-N4090WF3-24GD/sp#sp
now add everything else cpu motherboard you need at least 1000watts

_kaurus said:

Pardon me, can you elaborate on what is a GCC patcher?

Click to expand...

You said you're patching your games:

_kaurus said:

For example, I'm patching in the games I typically play

Click to expand...

So I assume your game patcher uses hot-patching, for which it needs a compiler to modify games code while the game is running.
Most likely your patcher installed gcc as part of it's installation.

_kaurus said:

After I installed Windows, and it had a chance to download its stuff again

Click to expand...

You see, you again confirm you've downloaded a bunch of software to your PC after OS reinstalling.

What's the name of this game patcher you're using? and are you using it?

Ghot said:

@_kaurus


If PCPartPicker has you at 889W with all your parts... you need a bigger power supply.
PCPartPicker's wattage total is fairly... conservative.

Also, AIDA64 Extreme's stress test is fairly mellow. A test like Prime95 is much, much more strenuous.

Lastly... black screens are usually caused by RAM or power supply, assuming the vid card, the cables and the monitor are functioning properly.

Click to expand...

Okay, Thank you. Ya, I'll buy a PSU at boxing day, regardless. No need to press this one so hard in the rare scenarios that it will get pressed. This also explains why when I was downloading two games from two game clients to two different NVMEs my computer would black screen.

zebal said:

You said you're patching your games:

So I assume your game patcher uses hot-patching, for which it needs a compiler to modify games code while the game is running.
Most likely your patcher installed gcc as part of it's installation.


You see, you again confirm you've downloaded a bunch of software to your PC after OS reinstalling.

What's the name of this game patcher you're using? and are you using it?

Click to expand...

Pardon me, I'm being transparent. I just didn't know what GCC was, the first I heard about it.

I think this is linked to war gaming's game client since I also see war gaming is on the report for errors.