[SOLVED] Is my GPU dying?? Radeon RX 580

[Merzak]

Hello,

I've been having small 'issues' for months now, but I actually thought it was my monitor. First, a few details of my rig:
b450m Max Mortar M-ATX M-Board
Radeon RX 580 GPU (got it near launch and used since - fairly intense gaming)
Ryzen 5 3600 CPU
16 GB DDR4 RAM
Windows 10 OS (latest
Graphics drivers up-to-date with latest

Initially, I was getting, every now and then (maybe once every other hour if that) multi-coloured lines flicker on my screen, like lightning forking in places. Very thin lines, not strips, and not all the way across the monitor. These would appear in random locations, seemingly never in the same place. I thought it was just my monitor, and it was so infrequent it didn't seem immediate. This would occur during gaming, or even Netflix.

Then it seemed to get a little worse, by my screen sort of turning 'on and off' (black screen, for literally a tenth of a second, then flick back on right as rain). Again, this made me think my monitor.

Over the last couple of days now, maybe a week at most, if I'm playing a more intense game (specifically CoD: Warzone, but not CoD: Bl-Ops C-War) my screen completely locks up for about 5 seconds. During this time I cannot do anything (alt-tab, etc) and then after 5 seconds the game/screen catches up like it was running fine the entire time. Now as said, this only happens on Warzone, so it made me think that it was just the latest update that had implemented this bug (that's ENTIRELY plausible) and so I thought nothing of it once again.

But this morning I was watching YouTube, and my screen blacked out, like a mini-reset, and then came back after about 1-2 seconds. This occured two or three times in about 30 minutes, which made me think monitor again. But then I launched CoD: Warzone and I got Dev Error 6070, the game black screened and then crashed. Now when I launch the game, it just crashes to desktop (pretty cleanly at least) almost immediately. Dev Error 6070 seems to suggest a graphics issue, which is why I'm all of a sudden considering all of the above issues being GPU related, and not my monitor.

But before I go ahead and fork out £300+ simply to replace my perfectly capable GPU (aside perhaps beyond it's imminent death), I'm hoping I can simply get some feedback from someone who knows a bit more about this stuff to confirm if the above does indeed sound like my GPU is on it's way out, or not. I'm not exactly the richest guy to just go flinging money around regardless, though the more initial issue is that there seem to be no GPUs available for purchase anywhere on the planet either, so if I do fork out money, I want to make sure I do it for good reason.

Thank you in advance for any help!

 

[RARRAF]

well it is a fairly power hungry GPU and you fail to mention the make and model of your PSU which I would suspect is the problem and a lot cheaper than a new GPU.

 

[Merzak]

well it is a fairly power hungry GPU and you fail to mention the make and model of your PSU which I would suspect is the problem and a lot cheaper than a new GPU.

Thanks for the response, Rafraf. I'm not entirely sure how in any way an underspecced PSU would cause the issues I'm having, certainly only occuring 3+ years after the initial build. In any case, I've opened the side panel and it is a Berlin 630W 80+. I feel that should be capable of powering my rig okay.

 

[David0ne86]

The berlin psu (the ones produced post 2015) operates on the terrible and infamous cwt platform (it used to be a sirtec based psu but they changed for god knows what reason, sirtec is a great psu OEM). It's really not a good psu my friend, so yeah, that could DEFINITELY be the cause. Wattages in psu means basically nothing. There are 550w psu that could erogate more power than your current one and be more stable while doing so. They have a 3 year warranty for a reason lmao. On PSU 3 years are nothing. A good psu should ATLEAST work flawlessly for 5 years.

What you can do is put the 580 on a different system and check. But i would for sure look into upgrading that psu before declaring the gpu the problem. One it will be more cost efficient seeing the time we're currently in, two it's an upgrade you'd desperately need anyway.

 

[RARRAF]

here is the most up to date tier list for PSU's it might not be the problem but could be what davidone 86 said is good advise and only trying to help
https://forums.tomshardware.com/threads/psu-tier-list-psucultists.3624094/

 

[Merzak]

Thank you for the responses both. I still don't see how the PSU could be leading to my issue (though clearly the PSU is doing more stuff than just, "Providing electricity." - I'll be honest, the PSU is the part of a PC that I know least about).

I've done some looking into the list you linked, in addition to a few other articles about this, and have opted to go for a EVGA 550W SuperNova G3. Seems to have quality reviews for the price, and is pretty highly rated (in comparison to what does seem to be a highly non-recommended PSU in my current rig).

Considering my previous motherboard was seemingly one of the most cursed M-Boards to have ever existed, I'm not entirely surprised the PSU is of a similar fate...

As for my GPU, I don't sadly have the option (or not with ease) of trying it in another system. So that's a no-go in general. Considering the issue seems to be fairly on-and-off, even if I did try it, it might just work and make me think it's not an issue anyway. After doing some digging around, I've reverted my graphic drivers back a version, just in case the latest was causing an issue with CoD: Warzone specifically (as that game seems to be the only one that suffers). I rebooted my PC during that process (as well as unplugging and re-plugging every single connector on my PC), and CoD: Warzone worked there-after.. Coincidence? No idea. I've had a single screen flicker this afternoon, but otherwise no issues.

I did also monitor my GPU and CPU temperature during a game, and the GPU got up to 63 degrees C peak, and my CPU 58. A little higher than I'd like, but nothing problematic for sure. At this point, I guess I'll take closer note of what I'm doing when things happen, and see if there are any simple common denominators that might point to any one issue (and perhaps the PSU will even solve it; who knows!).

 

[RARRAF]

ok good luck with it see how it goes.

 

[Merzak]

Just for anyones curiosity, should anyone come looking at this for their own advice, upgrading my PSU did not fix anything, sadly. That said, it did effectively force me to learn a little bit more about PSUs in general, and I can highly recommend going with a Modular PSU. My old one was not, and so I had to remove ALL of the wires (near enough) like the good ol' days. My new one is modular, though, so in theory, should I ever upgrade, I now simply unplug the PSU only, and replug in the new. No need to rejig the wires over and over! (Modular meaning the wires clip into the PSU itself, as well as the GPU etc, so easily switched out.)

My PC also now looks much tidier, as my old PSU was multicoloured, loose wires - The new one came with full black, wrapped in a sleeve, so nothing is loose and it's much cleaner - Probably saves me a penny as well on leccy bills in the long run.

In any case, after doing a bit more reading and investigating, I do think my issues are a mix of two things: First, CoD: Warzone is trash. The lobby itself has MANY issues, let alone actually the game itself, and so it crashing due to my drivers being up to date was, as I suspected, entirely the cause of that. However, other issues, like the black screen, flickering, etc, all do suggest that my GPU is ageing, and slowly crawling towards its demise. I do intend to upgrade (preferably to an AMD RX 5700 XT), though this clearly will take me getting EXTREMELY lucky on the GPU market that is 2021 on-going: Reading a recent statement from both NVidia and AMD CFO, they expect the shortage to continue well into 2021, likely only really starting to go back to normal (prices and stock) around late Q1, early Q2, 2022. If that is true, then I really hope my GPU can just hang in there that little bit longer! Had it for 4 years, now, so I have confidence in the little guy!

 

[David0ne86]

Yeah man, altho you didn't truly fix anything, having a quality psu is so beneficial in many ways. As you said having a modular you can only use the cables you really need therefore improving airflow, temperatures and cleaning time. Also the components will have more stable voltages to work with and less ripple spikes that can potentially damage the whole rig. As i always say: NEVER cheap out on the psu. Glad you could see yourself why.

 

[spentshells]

if this were a me problem, I would clean my heatsink apply new paste after cleaning the old stuff and put some thermal pads on the memory.

 

[dotas1]

My new one is modular, though, so in theory, should I ever upgrade, I now simply unplug the PSU only, and replug in the new.

Unfortunately no. The wires are not universal for all brands. Most manufacturers use their own wiring that only works with their PSUs and when used with other brands they will either not work at all or shortcircuit something.

You did right changing the PSU even if it wasn't the issue. They degrade with age, that's why 3 years of warranty was nothing. The good ones tend to have 7+years.

I would try a small underclock of VRAM clock and see if the problems remain.