"New" PC crashes: No video signal,no Bluescreen, no Event Viewer entry

OK fine.

If you activate driver verifier follow scrupulously the tutorial you have indicated.

No, it's not something to leave overnight or during the day. It is necessary to use the computer during its activation.
It's supposed to create bsods and therefore a dump file which should be posted in the bsod part of this forum, in which moreover there is a description before posting. Then someone who decrypts the dump files will have to answer you, it can be long, if not ever.

Whereas if you uninstall the wifi driver from the new Windows installation and do restarts and your reproduction tests it will go faster I think.

Apologies if this post is a bit long, and the formatting a bit wonky. Posting a bit in a hurry.


Okay, so things outside my control happened last week that hindered my productivity.

Didn't manage to get the PC ready for send-off, told the One.de Support as much, expecting to be told "nope, you're too late", but I guess I got some more leeway. Not completely clear how much, I guess sending it in this week is still fine, maybe even later? Their mail left that vague, I clearly slid into a gray area.


But maybe I can finally figure this out myself now though with the information I've gathered, so hopefully I won't have to send the PC in at all.

I currently suspect either the GPU or the Power Supply is to blame.



Again, to reiterate: The only real important question is, if this is a physical hardware defect that may get worse over time.

If not, I don't need to fix it, under normal conditions there are only 3 - 4 crashes per month, usually on bootup, I can live with how it is now. This is not bad "yet", the only question is "will this get worse". Then I'd need to send my PC off for repair and risk transport damage, to have it looked at preemptively within the "warranty period" that technically already expired on 22.01.23.

If this is just a matter of "find the right settings to change and then it'll work", "nothing is actually physically damaged", then this is fine, no rush to find those exact settings right now, no need to send the PC off to a repair-shop.

If this is just a matter of "your power supply isn't strong enough / your Custom PC's GPU isn't fully compatible, but this isn't going to break your PC, just occasionally crash sometimes", then I can live with that.

^{(I mean, I'd still want to figure out how to fix it, but then there's no rush.)}




One important realization is that installing certain drivers, changing some settings, and unplugging one device has not made the crashes go away - but it has modified their behaviour, affected how and under what circumstances crashes happen.

This to me seems to point away from the "hardware defect" hypothesis, which would be very, very good. :)




Alright, let's go over this one by one.



____________________________________________________________________________________________________ __________________________

itsme1 said:

if you uninstall the wifi driver from the new Windows installation and do restarts and your reproduction tests it will go faster I think.

Sorry it took so long to try that. Yeah... turns out it's not that, but I learned some stuff along the way. Here's what I did:

First of, I tested the new Windows installation with Fritz!-WLAN drivers uninstalled, and the WLAN-stick removed.

In fact I uninstalled (almost) everything that wasn't there by default - programs that I'd previously installed either for testing/analysis purposes or to provide a "minimum work environment".

I also removed the few things I'd put on autostart ("open this folder", "start screenshot hotkey tool", "open Hardware Monitor tool", "play music so I can listen for audio glitches").

On a hunch, I also unplugged everything except the monitor, keyboard, and mouse. (This turned out to be important!)

Then I booted into the new Windows installation, waited 10 minutes, shut it down, waited 10 minutes, turned it back on, waited 10 minutes, shut it down, repeat.


I only did work that was as resource-low as possible - writing in a *.txt file, open some folders with a handful of files each, maybe look at some screenshots.


So ~8 hours later, after ~18 such restarts, the PC had not crashed.



Then, on a hunch, I plugged in the (several meter long) HDMI cable that connects to my TV. That was the only thing I changed.

(My actual monitor is connected with a DVI-cable. Windows apparently thinks "HDMI = main screen".)

After 3 more restarts, my PC crashed 6 minutes after startup. TV wasn't even on when it did.

I tried this setup again the next day, still with HDMI cable connected to my TV. Crashed on the 7th startup.

Somehow I doubt it's a coincidence, I'm guessing that might point to the GPU as the culprit, which would also explain the "no video signal", albeit not why there aren't Event Viewer entries to log the error.


VERDICT: "The fresh Windows installation, with no additional programs installed, absolutely minimal workload, and only mouse/keyboard/monitor connected does not crash on bootup, unless I also plug in the HDMI cable that connects my TV as a screen mirror."



I had tested "what happens if I disconnect all devices" (except monitor and keyboard) early on (see video in the first post) and had still gotten a crash - however this was on my old Windows installation, AND this was before I'd realized that the MSI / AMD drivers for motherboard / CPU / GPU weren't installed yet.

Which brings me to something I wanted to post last week but simply didn't manage to get around to.


____________________________________________________________________________________________________ __________________________

Early on, one of my reproducible "crash tests" used to be this:

Autostart Crash-Test

Put the following programs on autostart, so they'll open in the first 2 minutes when the PC usually would be likely to crash:
- Chrome (600+ tabs)
- Waterfox Classic (9200+ tabs)
- A poorly optimized 3D game ("Event[0]") (Though loading past the title screen often takes some extra minutes anyway.)
- (Plus the Epic Games store app which is already on autostart.)

^{(Running Waterfox Classic with 9200+ tabs is not unusual for me, it's way better optimized than Firefox itself, makes little difference if you have 200, 2000, or 10.000 tabs open.)}

Running these programs isn't even a stress test, it's just "do it immediately on bootup, and all of them within seconds of each other" instead of "wait a few minutes before starting those programs" that causes problems.

I tested this on my regular old Windows installation*, three times each, both before and after installing the MSI & AMD drivers for motherboard / CPU / GPU.

Before, it crashed 3 times out of 3 tests.

After, this test case did NOT crash anymore on 3 different attempts. On the third attempt, I left the programs open for about 5 extra minutes. And suddenly I got graphical glitches instead of a crash:








These graphical glitches continued to persist even after I had closed Chrome/Waterfox/Event[0], though I admit I only left the system on for about 15 minutes.


Noticeably, while opening new instances of MS Paint was fine, opening a second instance of "SpeedFan", a hardware monitoring tool, did not load properly, just displaying a too small gray Window with no content. The first thing SpeedFan normally does is check for installed hardware:

"Win9x:NO 64Bit:YES GiveIO:NO SpeedFan:YES
I/O properly initialized
Linked ISA BUS at $0290
Scanning ISA BUS at $0290...
Found WDC WD40EZRZ-00GXCB0 on AdvSMART
Found SCSI drive Samsung SSD 970 (500,1GB)
Found Radeon RX 590 Series
End of detection"

I'm guessing that failed.


^{* (The Autostart Crash-Test was not reproducible on the fresh Windows installation because it took too long to start the programs to count as a valid crash test. I'd installed the new Windows on my hard drive, while my regular old Windows is booting from a much faster M2.SSD.)}



____________________________________________________________________________________________________ __________________________







_____________________________________________________________________________________________

NTN said:

I know in my bones that it has something to do with the GPU and the settings there.

Yeah I suspect you may be right. Either the GPU or the Power Supply.

NTN said:

The Power Plan in the PC should also be a "High Performance" plan.

Today, for testing purposes, I did the following:

• I set the power plan from "AMD Ryzen (Balanced)" to "High Performance".
  
  <sup>("Ultimate Performance" isn't really sensible for a home PC from what I've heard, and probably wouldn't add anything other than needless strain on the harddrive.)
  
  (Note: I do not have a "AMD Ryzen High Performance" power plan, just the preinstalled "High Performance" that comes with Windows.)</sup>
  
  
• In addition, I disconnected the HDMI cable that connects my TV as a screen-mirror, since that had caused problems before, see above text. Left my Fritz! WLAN stick plugged in this time though.
  
  
• I also plugged my PC directly in the wall outlet instead of its usual power strip. (The only other things in that power strip are a lamp, and a printer that's rarely on - but hey, just in case.)
  
  
  [...] check if plugging the system directly to a power socket (not to an extension cable or power strip) solves the problem. [source]

My test once again was "turn PC on, wait 10 min, shut it down, wait 10 min, turn it on, repeat".

This time I tested this on my regular old Windows, with all its usual autostart processes still on, so just regular normal load to see how that would fair.

^{(I already tested the more "minimalist" case with the new Windows installation after all, see earlier in this comment.)}



On the 3rd bootup, my regular old Windows crashed.


BUT!!! This time it DID NOT go "No video signal", instead Windows actually restarted on its own, BIOS bootup screen and everything! That is the FIRST TIME this has ever happened!!

Wow!! One more step, and I might actually get a bluescreen!

...I'm being a bit sarcastic, but yeah, Windows rebooting on a crash instead of the PC hanging in a limbostate with no video signal is actually a step forward to being "more normal".

Still no Event Viewer entry that could give any hint though - but something else:



Popup after bootup: "Default Radeon WattMan settings have been restored due to an unexpected system failure."

Not sure if it's a false-positive error message though?

Might just be related to me poking around in "AMD Software Adrenaline Edition > Performance > Tuning Control > Manual Tuning > Custom" and temporarily enabling Custom "Power Tuning", then disabling it again and switching back to "Automatic Tuning > Default" afterward without applying any changes.



I did this just to check the available options without saving anything, as adjusting that setting was one of the suggestions in the article linked by NTN (see section "Increase the Power Limit of the Card").

_____________________________________________________________________________________________

NTN said:

I think the most relevant is in relation to your situation:
Increase the Power Limit of the Card
and
Edit the System’s Registry
(TdrDelay)

I'd like to try the following suggestions from the article, and to save time I'd change all of these settings at once and then run a "turn PC on, wait 10min, turn PC off, wait 10 min, repeat" test.

^{(If this works out I guess I can go through each one-by-one afterwards to figure out which one did it.)}

- Edit the System’s Registry -> TdrDelay / TdrDiDelay
- Increase the Virtual Memory of Your System
- Increase the Power Limit of the Card
- Change the Power Plan -> Well, I already did that.



Before I can try these, I have several pressing questions to make sure I choose the right parameters.






_____________________________________________________________________________________________

From the first article you linked:

Edit the System’s Registry

[...]

Then, in the right pane, right-click on the GraphicsDrivers and select New>> Dword (32-bit) Value.
Now name the key as TdrDelay and set its value to 8 (or 30).
Then, again right-click on the GraphicsDrivers and select New>> Qword (64-bit) Value.
Now name the key as TdrDelay and set its value to 8 (or 30).

This is what the section HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\GraphicsDrivers looks like for me currently, in my regular Windows:



My TdrDelay is already at 1E (30), that's even higher than in your screenshot, is that already enough or should I set it even higher? To what value?

Should I also add the 64bit Qword-version of the key? (You don't seem to have that, so I assume that isn't required.)

I don't have a TdrDiDelay key - that also isn't mentioned in the article. Should I add that and set that to 1E (30) as well, or set it another value? (And is it enough to add the TdrDiDelay-key as a DWORD, or should I add it as both DWORD and QWORD just in case?)


_____________________________________________________________________________________________

Increase the Power Limit of the Card

[...]

Launch the AMD Radeon software and head to its Performance tab.
Now navigate to the Tuning tab and select Manual.
Then enable Power Tuning and increase the Power Limit by 50%. Then apply your changes.
Increase Power Limit of AMD by 50%



Now reboot your PC and upon reboot, check if the driver timeout issue is cleared.

Dumb question: The power slider only goes from -10% to +10%. In the picture it also is set to 5%. The default for me is 0%. Not sure where that "50%" comes from or if they meant to say "+5%" instead.

So... what value should I set this to? +5%? Or the maximum value of +10%?




_____________________________________________________________________________________________

Increase the Virtual Memory of Your System

[...]

Then, in the section of Virtual Memory, click on the Change button.

Image A:

Click on Change in Virtual Memory Section
Now uncheck the Automatically Manage Paging File Size checkbox for All Drives and select the system drive (in the Drive section).
Then select Custom Size and in the Initial Size box, enter the Recommended value from the Total Paging File section e.g., 1908 MB.
Now, in the Maximum Size box, enter the Currently Allocated Value from the Total Paging File section e.g., 6144 MB.

Image B:

Set Custom Size of Virtual Memory of Your System
Then click on the Set button and apply your changes.

Since the RAM did not turn out to be faulty, maybe the problem lies with Virtual Memory instead?


This is what that looks like for me currently:



...I'm guessing in the article, "Image A" and "Image B" were taken at different times, hence the difference between "total paging file size for all drives" (8199 MB) in Image A, and the value of "currently allocated" (6144 MB) in Image B? Since for me that is the same value.

What I don't get is the bit illustrated with green arrows in Image B, about manually entering the same values that Windows already chose automatically. That's... not "increasing" the Virtual Memory, it's just setting it at a fixed value.

Shouldn't I set higher values than whatever the value happens to be at that random moment? What values should I set this to?


Also, bonus question - I'm going out on a limb here and assume that the Virtual Memory for a Windows OS that is installed to M2.SSD, with an hard drive also present in the system ... is still going to place that virtual memory file on the M2.SSD, not the hard drive, right?

I'm asking, because...
_____________________________________________________________________________________________




Lastly - and I'm running a bit late so I cannot go into too much detail right now - I get a hunch that the new Windows installation may actually be a bit more crash prone under certain circumstances than the old Windows. Might be because of the hard drive.

On the newly installed Windows, I've had two crashes outside of any tests, and not on bootup either, that simply happened shortly after I started a process or program that accessed the hard drive. (Not related to the GPU in this case, and certainly no stress test.)

I think it might be because I installed the new Windows to the hard drive, whereas the old Windows is installed to a much faster M2.SSD. (Or possibly because I only gave the new Windows a 70 GB partition... though it's still got at least 20 GB free.) Maybe this messed with the new Windows' Virtual Memory?


-------------------------------------------------------------
It's also noteworthy that on my regular old M2.SSD Windows, the hard drive for some reason is at 100% very frequently for the first minutes after bootup. Not sure what's up with that, processes involved seem to be System-related. (Would have to look up exact names.) But, ya know, the hard drive is NOT where the OS installed for that Windows, so that seems odd to me.

Only thing of note is that the hard drive is listed as "Disk 0", the M2.SSD as "Disk 1".
-------------------------------------------------------------


On the new Windows installed to the hard drive, I could imagine crashes happening if the hard drive suddenly stopped responding in time, obviously.

Not sure how the M2.SSD Windows would react if the internal hard drive suddenly hung up for a moment. ...I mean I don't know if that's what's happening. Might explain the audio glitches forewarning a crash? The music seems to be "streaming" from a file on the hard drive... the music player I'm using is very old (~2002), it doesn't load the music into memory, just read it straight from disk, you can literally edit a file and a few seconds later the music will change.
_____________________________________________________________________________________________

NTN said:

The positive here is it is better with a bottleneck in the GPU instead of the CPU...

Okay that sounds like good news! :)

If this really is just a case of "one component isn't perfectly compatible with the rest of the PC", then it isn't physically damaged, and that means the crashes won't get worse with time, and no "send PC off to repair shop" is needed.

I mean, naturally I still want to be more certain before I send One.de-Support a mail and cancel my repair request.



Just a quick addendum;

I can boil down most - not all, but most - of the above questions regarding "What should I test?" to the following:

"Would it be bad if I set the values 'too high' ?"
^{"...I could still finetune and set them lower afterwards."}

(Asides from overclocking the GPU, of course. There, "too high" most definitely would be bad, of course.)


"TdrDelay":
E.g., if I set "TdrDelay = 3C", "TdrDiDelay = 3C" , which is double what it is right now (1E, which is already higher than your value), would there be any obvious downside?

Anything that could do any kind of permanent damage? Or that I may be able to test, but then should reset the value again afterwards?


Same question goes for "Increase the Virtual Memory of Your System" and "Increase the Power Limit of the Card".


"Increase the Virtual Memory of Your System":
If I increase the Virtual Memory and set, say, "Initial Size = 8 GB, Maximum Size = 12 GB", instead of the ~5 GB both these values were last I checked, is there any downside to doing that, asides from it taking up some small amount of space on the M2.SSD drive?


"Increase the Power Limit of the Card":

Screenshot from article:


The article says I should increase the AMD graphic card's Power Limit by 50%, but the settings only allow a maximum of +10%, and even the screenshot in the article only shows +5% being set, so the author may have mistyped ...three times.

Own screenshot:


I guess there wouldn't be harm in trying out the "+5%" setting? It's either 10 times lower than what I'm supposed to try, or exactly what the author meant to type.

What would happen if I tried the "+10%" setting, would there be a danger here somehow?




Overclocking the GPU:

NTN said:

So you don't need to overclock the CPU, but rather your GPU slightly.(not to much!).....and give it also some extra power.

As for "slightly overclocking the GPU", I admit I have zero experience with Overclocking a PC, so I don't know what settings to adjust nor by how much. :/

Is it the setting "GPU Tuning > Clockspeed", or...?




AMD Adrenaline Software apparently also has an auto-overclock function, but I don't know if that will "slightly" overclock the GPU or "too much".



Let's just say "overclock the GPU" isn't something I feel super-confident trying as my first test, from what I've heard doing that wrong could damage the GPU.

Overclocking the GPU: and "Increase the Power Limit of the Card":
So I have no experience myself with the Radeon RX 590, and one almost has to consider my considerations about this card as amateurish, and there are probably people here on the forum who can contribute more valuable than what I can about this. Anyway, luckily there are sites on the web that are serious and that don't try to set a world record every time.

Radeon RX 590 (PowerColor Red Devil) review - Overclocking
We ended up with the following settings:
Core: 1651 MHz
Mem: 2250 (x4) = 9000 MHz
Power +50%
Voltage +100
Manually tweaked we gained 10% seen from reference performance with this tweak. (...my math tell me 8.5%)

This gives me a small hint as to what the upper limit is when staying within the "safe overclocking" category of this card.

"Increase the Virtual Memory of Your System":
Personally I don't think it will matter and would have stayed at 5GB. (But it should make no harm with higher values...)

"TdrDelay":
"(1E, which is already higher than your value)"...................I would have kept it there.

NTN said:

"Manually tweaked we gained 10% seen from reference performance with this tweak." (...my math tell me 8.5%)

This gives me a small hint as to what the upper limit is when staying within the "safe overclocking" category of this card.

I'm guessing msi Afterburner (used in the article you linked) may let me set higher values, but is probably less safe to use than the software that came with the GPU driver-DVD, AMD Software Adrenaline Edition. :/

I've read somewhere (unless I misunderstood!) that AMD Software Adrenaline Edition might have its maximum values capped differently depending on the card or the manufacturer... or something? Which would explain why I can't go higher than 10%, but this random person I came across apparently can go to 50% on their card, according to their screenshot... though they have a RX 570, not RX 590.


...I still find myself very confused that the first article you had linked provides screenshots of AMD Software Adrenaline Edition for "Increase the Power Limit of the Card", instructs readers to "set value to +50%", but then only sets it to "+5%" in the screenshot.



On top of that, it also turns out there are multiple different versions of the ATI Radeon RX 590, "Power Color Red Devil" being a different version than mine. The one it says I got on my receipt is "8 GB AMD Radeon RX 590 Sapphire Pulse".


. . .



Okay, let me ask it this way: "What's the worst that could happen?", if I set the Power Limit to, say, +10% in AMD Software Adrenaline Edition.

Is the worst thing just a crash, or actual danger of causing physical damage?

...Yeah, that's the level of (lack of) knowledge regarding Overclocking I'm working with here. :/

idgat said:

Absolutely ridiculous thread to be going on with, for a new computer. Posts from the OP are totally unnecessary and actions taken run the risk of voiding warranty.

Troubleshooting and problem solving are all well and good, and can give a sense of satisfaction IF solved, But this is a new computer, and the processes should NOT be the responsibility of the owner/purchaser.

Especially when the poster stated when delivered, the delivery service literally dropped it onto the concrete. I would have immediately opened a damage claim against said delivery company.

HedgeToaster said:

^{(One of which is that I don't trust the postal service not to damage the PC ... any further. ...When delivering, they dropped it off right at my door step. As in: literally dropped the package, half a meter to the stone floor. Right at my door step. ...Styrofoam packaging and "inflated foam cushion" inside the PC and all, but still. Aargh! -_-# )}

HedgeToaster said:

I'm guessing msi Afterburner (used in the article you linked) may let me set higher values, but is probably less safe to use than the software that came with the GPU driver-DVD, AMD Software Adrenaline Edition. :/

I've read somewhere (unless I misunderstood!) that AMD Software Adrenaline Edition might have its maximum values capped differently depending on the card or the manufacturer... or something? Which would explain why I can't go higher than 10%, but this random person I came across apparently can go to 50% on their card, according to their screenshot... though they have a RX 570, not RX 590.


...I still find myself very confused that the first article you had linked provides screenshots of AMD Software Adrenaline Edition for "Increase the Power Limit of the Card", instructs readers to "set value to +50%", but then only sets it to "+5%" in the screenshot.



On top of that, it also turns out there are multiple different versions of the ATI Radeon RX 590, "Power Color Red Devil" being a different version than mine. The one it says I got on my receipt is "8 GB AMD Radeon RX 590 Sapphire Pulse".


. . .



Okay, let me ask it this way: "What's the worst that could happen?", if I set the Power Limit to, say, +10% in AMD Software Adrenaline Edition.

Is the worst thing just a crash, or actual danger of causing physical damage?

...Yeah, that's the level of (lack of) knowledge regarding Overclocking I'm working with here. :/

If it is still under warranty, do not mess with voltages, do not open the case, do not pass go or you will VOID THAT WARRANTY.

#1 biggest lesson you need to learn:
Many things can cause a bluescreen.

Your pc has multiple issues, dropping it was the first.
your wlan driver caused one, i put the debug info in the post, calling out the wlan driver, its old 2017 date, and that windows said it crashed because of it. If you dont want to look for an update, and dink around with your hdmi cable that you so insist is the problem (but hey, it crashed without it on the old installation too, you said), thats entirely your choice.

Dont gamble on how much "leeway" the company is gonna give you. A rep can say you can send it in next year, but that rep is not the boss. Your pc would get sent back with just a simple note: sorry, out of warranty.

Hurry up, box it up and send it back if you are going to.

There will always be arguments for and against when there are countless combinations of values.
Don't worry too much about the appearance of these sliders, but look at the values ​​in terms of Volt, Power, Memory and Core. Usually, a GPU should be able to withstand being overclocked by 10%, but if you are not at all comfortable with that, set a threshold of 5%, and do it step by step.
If you're not comfortable with that either, you better leave it be.

It's also good to have a plan in case everything goes completely wrong. If it turns out that the PC freezes completely and you get a BSOD and the PC won't boot again, it may be that you have to start in safe mode and reinstall Afterburner (or whatever you use...) so that the values ​​are reset. And how to do this you should be aware of in advance.

Okay, you guys are right.

I would've wanted to do "one more test" to feel more secure in my decision, but I'm not taking the risk of messing up with overclocking when I don't understand how to do that properly.

Of course it would've been ideal if we could've found a way to make the crashes go away completely, instead of merely "changing the behavior of when and how the crashes occur", but that isn't actually necessary.

From the start, the only question was: "Is this a hardware defect, i.e., will this problem get worse over time?"

If yes, then I should send it into repair under warranty. If no, then not.

Right now, with all the tests I've done (mostly during this last week) thanks to your suggestions and advice, it looks like that is NOT the case.

Right now it looks more like this is just an odd quirk this custom-build PC has, or some incorrect setup or minor incompatibility between components, that hasn't gotten any worse in the last 3 years (at least not noticeably so). And that means it getting worse than it currently is is unlikely. I can live with 3 or 4 crashes on bootup a month - I don't even lose anything if that happens in the first 2 minutes.

Therefor it won't require being sent into repair.



There is of course the risk I'm wrong, but there also is a risk in sending the PC in for repairs, especially in sending it via mail, and also the "leeway" I've been given quite possibly running out before it actually gets there.

Right now I judge the risk of sending the PC into repair as greater than not doing so, especially if "repair" turns out to be unnecessary.

This is honestly a relief, actually. It's been a very hectic, chaotic, and stressful last month and a half trying to sort all of this out.


Thanks so much you guys for keeping up with my nonsense, I hope I didn't try your patience too much.



Even if technically the crashes didn't get fixed completely, I declare this thread as "solved".

Maybe if something else presets itself at a later time, I'll post to this thread, depends on whatever the circumstances may be.

But no rush.



Cheers!

- HedgeToaster.


PS: Opening the case is actually covered by One.de's warranty, and I was in fact explicitly instructed by their support to, among other things, reseat the RAM. But yeah... don't remember if the warranty covers overclocking, didn't even think of that. Good call.