[SOLVED] Is my graphics card dying? Or might there be a different problem?

[throgoth]

Hello all

TL;DR: I am almost certain that my graphics card has had enough of this cruel world, just wanna rule out other possible problems before I throw myself into financial ruin by buying a new one.
as the title implies, I am having problems with my PC and I am almost certain that it is the graphics card that is slowly dying, but in order to be absolutely sure, I wanna ask you all what you think of it.
We all know the prices for GPUs are pretty ridiculous right now and I don't wanna buy a new graphics card unless I absolutely have to.
So here are some facts:
my system

• CPU: Intel XEON E3-1231 v3
• Graphics Card: Radeon R9 390
• Motherboard: Gigabyte H97-D3H
• RAM: 16 GB
• PSU: Cool Master Masterwatt Lite 600W

My problem is that while gaming or just watching YouTube videos random black screens occour, not even BSODs, the screen goes completely black and the system becomes unresponsive untill I force shut down the computer using the power button.The problem occured first a few months ago and it seemed to get worse with time. At first the crashes happened when I played, then watching YouTube joined the party later.
Currently I manage not to have crashes when underclock the GPU (core base clock 1040Mhz, now I run 850Mhz, memory base clock 1500 Mhz, now I use 1350 Mhz). With these setting I haven't had problems so far (when gaming, I only play Dark Souls 1-3 atm, DS3 under min settings), but I am afraid that this is just a matter of time.
All drivers for all hardware components are up to date and I recently clean installed Win10.
I hoped that only the PSU was dying, therefore I replaced it a few months ago but the problem still persists.

To give you an idea what my PC has been through over the last years (I built the PC early 2016):I wouldn't say that I put a lot of "heavy load" on the PC. Heavy gaming sessions were very rare, I've never overclocked any component, I usually turned off my PC every night, except for very rare occasions where I had a download running.

Also of note: at first the crashes occured very random, I could push the graphics card to limit for a while and nothing happened, while reducing the load then seemed to crash the system (for example playing a game like Witcher 3 for a while without problems, but as soon as I closed the game I went to YouTube, boom, black screen). So at first it seemed so random, that the crashes weren't reliably reproducible, but now on base clock rate it pretty much occurs after few minutes under load and YouTube is still more or less random, but it will happen for sure.

A word on temps: the crashes seem unrelated to temps. I never had abnormal high temps (speaking of going above 80°C). Also the system crashes when watching YouTube and the temps don't even scratch 50°C (also the temps for YT are the same on base clock speed and on reduced, however on reduced there are no crashes).

In conclusion: I think my graphics card is dying. The evidence seems to point in a clear direction. I just wanna ask the more knowledgeable among you what you think of this. Could there be another reason for this problem? Could it be the Motherboard, for example, and if so, how could I test this?

I am grateful for every reply and imput!

 

[Darkbreeze]

Get a real power supply. I'd almost certainly bet the house that the PSU is your problem. One of the absolute worst power supplies ever sold by a well known brand name.

See here:

Discussion - PSU recommendations and power supply discussion thread - Tom's hardware

Below you will find MY standard list of recommended power supplies and beyond that this thread is intended as a landing place for questions or discussions regarding specific units, platforms or related PSU tech, all of which are all welcome to be discussed here. If it's related to power supplies...

forums.tomshardware.com

Regarding this review, which is for YOUR exact power supply model.

Cooler Master MasterWatt Lite 600W (230V) Power Supply – Page 6 – JonnyGURU.com

www.jonnyguru.com

Well, looks like jonnyguru review is gone, as is the site, but I assure you, of the hundreds of reviews I've read on well name branded power supplies, that one was one of the worst ever. 99.999% likely that is your problem, not the card.

 

[Darkbreeze]

Fortunately, the Wayback machine has us covered. A 5.5 point score out of 10, is pretty abysmal. The use of third tier capacitors is where your problem likely lies however. Given that your unit is VERY LIKELY more than five years old, and third tier capacitors are extremely lucky to have made it this far, it's almost a certainty that that PSU is on the brink of giving up the ghost and I'd highly recommend you take it out of service and replace it with a quality unit before it has the chance to destroy the rest of your hardware. Because, it likely will when it goes.

Cooler Master MasterWatt Lite 600W (230V) Power Supply – Page 6 – JonnyGURU.com

web.archive.org

 

[throgoth]

Fortunately, the Wayback machine has us covered. A 5.5 point score out of 10, is pretty abysmal. The use of third tier capacitors is where your problem likely lies however. Given that your unit is VERY LIKELY more than five years old, and third tier capacitors are extremely lucky to have made it this far, it's almost a certainty that that PSU is on the brink of giving up the ghost and I'd highly recommend you take it out of service and replace it with a quality unit before it has the chance to destroy the rest of your hardware. Because, it likely will when it goes.

Cooler Master MasterWatt Lite 600W (230V) Power Supply – Page 6 – JonnyGURU.com

web.archive.org

Thanks a lot for your reply!
At first I suspected the PSU, that's why I replaced my old one (a Cooler Master G550M, that I had since 2016, no problems before earlier this year) with the Masterwatt Lite 600 (that was the only one they had in store where I went to that didn't have fewer Watts). And despite the new PSU, the same problem still occurs.

The fact that the problem still occurs, despite this new PSU, garbage as it might seem on further research, kinda suggests to me that it might be the card.

I unfortunately have no access to another system, where I could plug my card in to test the card independently from the other components in my system.

 

[Darkbreeze]

I won't discount the possibility that it's a card issue, but I WILL tell you that replacing a failing or faulty PSU with THAT particular PSU, new or not, doesn't tell us much since it's probably as weak in new condition as your G550m was all old and tired out. Besides which, given the known problems with that unit you would be VERY wise to return it if possible. If not, I would still highly recommend not using it except maybe in a very lightweight machine with extremely low power requirements.

In the future, any power supply with "lite" in the name should be a dead giveaway that it should be 100% avoided.

Having a reliable PSU that is capable of delivering it's full power, should be compulsory for anybody who values their hardware. And, the fact that you are having problems with your system, possibly with your card, could be directly stemming from the use of a rather old, followed by an extremely poor, power supply. Realize that poor quality or old tired power supplies tend to deliver higher than desirable levels of ripple, noise and in many cases, poor or out of spec voltage regulation, which can definitely wreak havoc on the caps on any graphics card or motherboard.

In simpler terms, a PSU can kill your other hardware, slowly, or even quickly, regardless of whether it seems to be "working" or not. BUT, if the damage is already done, you might actually need to replace both, and IF you DO buy a new graphics card, please, PLEASE, be wise enough not to run it with that power supply. If you have to order one in to get one that is worthwhile, do it, don't limit yourself to only what you can find on the shelf in your retailer if what they offer isn't worthwhile.

These are where to look for more information and recommendations on what models are ok, good or to be avoided.

Discussion - PSU recommendations and power supply discussion thread - Tom's hardware

Below you will find MY standard list of recommended power supplies and beyond that this thread is intended as a landing place for questions or discussions regarding specific units, platforms or related PSU tech, all of which are all welcome to be discussed here. If it's related to power supplies...

forums.tomshardware.com

Article - [psucultists] PSU tier list rev. 14.8 (Final Update Jul '21)

PSU Tier List 4.0 rev. 14.8 (outdated) Last Update: 12-07-2021 Legend : Gray - EoL/obsolete and/or otherwise not recommended for purchase. Green - small form-factor (gold and blue colors are disregarded due to scarcity of SFX PSUs) Gold - best units in the tier (includes requirements for blue...

forums.tomshardware.com

And since it's advisable to replace the power supply NO MATTER WHAT, if you buy a new, highly expensive graphics card, it would be VERY wise for you to do that FIRST, since there is the possibility still of it not being the card. If it turns out the card itself IS still bad, you lose nothing, because you needed to do it ANYWAY. But, you might find that with a high quality unit, you might not need to replace the card. Just, that's the best advice I think anybody could offer in this situation, or at least what I can offer you since you have apparently no way to test the card in another machine that does have a capable and reliable PSU installed.

If you choose to buy a new graphics card and use it with that PSU, you might just as well light your 8 to 15 hundred dollars bills on fire in the trash can. Honestly.

 

[SkyNetRising]

At first I suspected the PSU, that's why I replaced my old one (a Cooler Master G550M) with the Masterwatt Lite 600. And despite the new PSU, the same problem still occurs.

What make/model R9 390 ?
For some R9 390 overclocked models - minimum recommended PSU can be 750W.
You're trying to run it with not very good 550W and 600W models.

 

[Darkbreeze]

Right. Capacity could absolutely be an issue. Especially when trying to use an old 550w or a 600w that struggles to supply rated capacity with a card that even not highly overclocked models might want to see a good 650w unit for, given the system as a whole.

 

[throgoth]

And since it's advisable to replace the power supply NO MATTER WHAT, if you buy a new, highly expensive graphics card, it would be VERY wise for you to do that FIRST, since there is the possibility still of it not being the card. If it turns out the card itself IS still bad, you lose nothing, because you needed to do it ANYWAY. But, you might find that with a high quality unit, you might not need to replace the card. Just, that's the best advice I think anybody could offer in this situation, or at least what I can offer you since you have apparently no way to test the card in another machine that does have a capable and reliable PSU installed.

Thank you very much for this statement and your input. I am convinced now that I should replace my PSU no matter what asap, because I definetely don't want to risk wasting money on a new graphics card just to kill it soon after. Maybe I am lucky enough to still be able to run my system normally after that.
After reading your links while checking availability and prices of a trusty online vendor for hardware in my country, a Corsair RM850x 80+ Gold seems to be the one I go with.
850W seems to be more than enough for me I think, since I don't overclock and I do not fire up heavy games anyways.

I will purchase a new one today and will get back to you once I replaced the old one. Fingers crossed it's just the PSU.

Edit: Found and purchased a Seasonic Focus PX 850W.

 

[Darkbreeze]

Either of those units would have been good choices, and in fact a 750w would have been perfectly fine, however it's not a bad idea that you chose an 850w just in case you decide or are impelled to purchase one of the higher end RTX 3000 series cards that require a bit more power than most recent generations did.

 

[Dead_Fighter_Pilots]

i had similar problems with my r9 290, but i had a waterblock for it, i luckilly got a brand new old stock r9 290x off ebay for $150, dunno who shelved a r9 290x for 7 years and never used it but its finally working great for me. old one would soar straight to 95C and thermal throttle and get horrible performance, even under a water block with good thermal paste. card was broken. tried to thermal paste an aluminum heatsink on the cards backplate and that was a bad idea. but i got it all fixed up now and never get past 80C usually low 70s

 

[throgoth]

Either of those units would have been good choices, and in fact a 750w would have been perfectly fine, however it's not a bad idea that you chose an 850w just in case you decide or are impelled to purchase one of the higher end RTX 3000 series cards that require a bit more power than most recent generations did.

Exactly my thought process. Tomorrow the new PSU will arrive and I will install it. I hope that this will solve my problem. If not, and the card is actually dying, I am still grateful for your valuable input. I would have bought an (atm) overprized graphics card and just killed it with a bad PSU either slowly or rapidly.
Back in the day when I built the computer I really went with the cheapest thing recommended to me by my then colleague/friend, who also built his own PC and seemed to know what he was talking about.

I'll come back for an update after testing out the new PSU tomorrow night!

 

[throgoth]

i had similar problems with my r9 290, but i had a waterblock for it, i luckilly got a brand new old stock r9 290x off ebay for $150, dunno who shelved a r9 290x for 7 years and never used it but its finally working great for me. old one would soar straight to 95C and thermal throttle and get horrible performance, even under a water block with good thermal paste. card was broken. tried to thermal paste an aluminum heatsink on the cards backplate and that was a bad idea. but i got it all fixed up now and never get past 80C usually low 70s

how did you get it all fixed up in the end, tho?

 

[throgoth]

Installed the new PSU, but problem still persists.

 

[Darkbreeze]

Well, that sucks, but, at least now if you have to end up getting a new graphics card you can take comfort in the fact that the PSU you have will be a benefit rather than a thorn in your side.

I'm still not fully convinced that your problem is the graphics card, based on the symptoms, but honestly I'm not sure how short of replacing either the graphics card or the motherboard, you could verify either one without simply replacing them. I guess the positive side of things is, new PSU and if you go with a graphics card you'll definitely be a big further up to speed since your card has some age on it by now.

 

[throgoth]

I am also not really convinced that it is the card, because I cannot quite make sense of how it can work under clocked but not at normal clock speed.
I recently thought about the possibility of it being the VRAM temp maybe? The temp reading for the GPU are nominal, but I can't read VRAM temp, as the card doesn't have a sensor for it, but here I am really just taking very wild guesses, because I have read that other people have problems with their VRAM temp, but I don't even know how temp in VRAM works, how my card cools its VRAM and so on.
Also it wouldn't explain how even when just watching a YouTube vid it would crash, because it doesn't get very hot during that anyways.

When running Ubuntu I never had any issues with Vids, but never tested any heavier loads.
However I don't think it is a pure software issue either, because the problem just one day appeared, despite having nothing changed about my system and even a clean install and up-to-date drivers didn't help either.
But maybe Ubuntu is utilising the GPU differently and therefore it doesn't crash when watching vids?

I am completely lost here basically

 

[Third-Eye]

throgoth
Have you tried any new display cables, monitors or a different display out port on the GPU?

 

[dotas1]

I am also not really convinced that it is the card, because I cannot quite make sense of how it can work under clocked but not at normal clock speed.
I recently thought about the possibility of it being the VRAM temp maybe? The temp reading for the GPU are nominal, but I can't read VRAM temp, as the card doesn't have a sensor for it, but here I am really just taking very wild guesses, because I have read that other people have problems with their VRAM temp, but I don't even know how temp in VRAM works, how my card cools its VRAM and so on.
Also it wouldn't explain how even when just watching a YouTube vid it would crash, because it doesn't get very hot during that anyways.

When running Ubuntu I never had any issues with Vids, but never tested any heavier loads.
However I don't think it is a pure software issue either, because the problem just one day appeared, despite having nothing changed about my system and even a clean install and up-to-date drivers didn't help either.
But maybe Ubuntu is utilising the GPU differently and therefore it doesn't crash when watching vids?

I am completely lost here basically

@Darkbreeze was correct advising you to swap PSU. It was a terrible one and there is a high chance of contributing to the degradation of the GPU.

Your GPU is old enough to have degraded on its own and can't have stable advertised clocks. The PSU did not help at all with this. The way to confirm is exactly what you did. You lower clocks of core and memory, usually small increments and if it's stable... You save for a new GPU.

If you have the possibility to try the GPU in another system with a decent PSU, you can also verify it is the GPU.

 

[throgoth]

throgoth
Have you tried any new display cables, monitors or a different display out port on the GPU?

No, I haven't tried this yet, but I could.

 

[SkyNetRising]

No, I haven't tried this yet, but I could.

We still don't know, what make/model Radeon R9 390 you have.
Can you provide the necessary info?

 

[throgoth]

@Darkbreeze was correct advising you to swap PSU. It was a terrible one and there is a high chance of contributing to the degradation of the GPU.

Your GPU is old enough to have degraded on its own and can't have stable advertised clocks. The PSU did not help at all with this. The way to confirm is exactly what you did. You lower clocks of core and memory, usually small increments and if it's stable... You save for a new GPU.

If you have the possibility to try the GPU in another system with a decent PSU, you can also verify it is the GPU.

Yeah, I guess I gotta face the fact that it really is just degradation. At least I am lucky enough that my screen doesn't stay completely black (at least as for now).
Hopefully my card lasts long enough to be able to save up for a new one or maybe to see prices decrease at least a bit.

 

[throgoth]

We still don't know, what make/model Radeon R9 390 you have.
Can you provide the necessary info?

According to my invoice it was the "8GB Sapphire Radeon R9 390 Nitro with backplate aktive PCIe 3.0x16"

 

[Darkbreeze]

Sigh.