troubleshooting RX480 Stuttering & Shutdown under load

[AhmedAm]

I FIXED IT: JUST REPLACE THE THERMAL PASTE

I have no idea why this fixed it, but I stumbled upon another thread with the same issue and he fixed it with replacing the thermal paste, I did the same and my GPU is better than ever  ( keep in mind that it worked for 1.5 years flawlessly, but I guess GPUs are finicky) 

 

 

UPDATE: Turns out that the GPU is at fault, tried it on a friend's PC (runs a 1080 on a vs650) and the issue persisted. But it went away when I underclocked it to 1150ish mhz so the power draw is less than 110w, but I couldn't replicate that on my PC unfortunately. I'm planning to either RMA the GPU (might be hard in Saudi Arabia) or opening it up and padding the VRMs

 

Hey there, it's my first post here , Hope you can help me, as I've been pulling my hair out for the last 3 months trying to solve this. (going to list the specs at the bottom to not clutter the top)
So I'm having 2 issues that may or may not be connected, the first one is the stuttering.

 

For some reason, whenever there is a certain amount of particles on screen (e.g CSGO Smoke, or DiRT Rally dirt particles) I would get a frametime spike of 100+ ms every one second, Increasing the texture quality on some games would cause the issue to appear sometimes too (PUBG High Textures would trigger it sometimes)

 

I've tried the following:
DDU

Older drivers

Downclock

Overclock

increasing and decreasing the power limit

running the game on my SSD

Cleaning the GPU 

Reinstalling windows (Completely wiped my SSD and switched from MBR to GPT)

Verifying vram using VMT (the test runs perfectly, but if I increase the surface size from 1024x1024 to 2048x2048 the test would fail every single check)

 

My second issue is the GPU shutting down under load, Furmark is the most reliable at crashing the PC (the GPU would crash within 1 minute), there's also a possible issue with the temp. sensors (Temp. would wildly change from 58c to 65c, which is obviously impossible)

 

I really hope my GPU isn't dying, as I can't replace it for the next ~5 months, please tell me if you need any more info.

 

Specs:

CPU: i5-4460 (4c4t, 3.2ghz boost)

GPU: RX480 GAMING X 4G (downclocked to 1305mhz core and 1750mhz mem, no change currently to power limit/voltage)

Mobo: Gigabyte H81M-S2PH (Latest BIOS)

RAM: 4+8 Hyperx Fury DDR3-1600

Storage: 240gb Sandisk SSD + 1tb 5400rpm WD HDD (Both are running perfectly)

PSU: Aerocool KCAS 500w 80+ Bronze

 

 

[SkilledRebuilds]

Even under furmark load it shouldn't shut down... it should operate normally (but warmer)

Temp Sensor statement makes me believe the Power Supply feed was altered for a second... maybe throttled and dethrottled... dropped down to 58's and back to near 70's..

Others who come here and read... - PSU IMO?

[AhmedAm]

I forgot to mention that the build had no issues for a long time (the build was originally made in 2014/2015, and I upgraded the GPU in 2016).

Plus the temp. wouldn't just dip once, it would keep bouncing between ~55 to around ~70 (looks like a sine graph :P)

[YoloSwag]

do you have a spare system you could test this on?

 

or maybe a friend's PC?

 

I'm hoping too it's a problem with the PSU and not your GPU.

[SkilledRebuilds]

Ive seen the same swings (sinewave graph) on Coreclocks with a GTX980 and a cheap PSU. (Mine was cheap, is what Im saying, not yours)

I actually killed that GPU, because the PSU was crap.
While diagnosing,... Core clocks were 600mhz from the 1100Mhz it should be.. swinging up and down 600Mhz-1100Mhz ranges... WTF!!
So I come over...
Against my advice....My friend (HIS PC) just ramped the Power Limit to Max from Default....while we were testing (actually under 100% load at the time) and POP_BANG. PC Turns OFF with PSU Sparks.

So the danger is real in certain situations.
Test GPU in another PC if possible to rule out PSU (If it happens in another PC, if not,.. GPU is good)
Test Different PSU in your machine (I know,.. without spare parts or willing friends,.. super hard to do)

I'm 100% nailing it on the PSU,.. from experience.
But... don't take MY WORD for it.

 

Store testing or Friend/Owners Testing MUST BE DONE.
I'd not be gaming on that PC until sorted.... because any flickers of bullshit from the PSU can take the GPU to 11 out of 10,...and then kill either the GPU/PSU or Both.
 

[YoloSwag]

55 minutes ago, SkilledRebuilds said:

-snip-

I saw a kid before he killed both his CPU and GPU with a crappy PSU. I was there to get parts for a new build and this kid comes in and tells the store dude his problem but it's like the store dude and I know what went down. Poor kid lol.

[AhmedAm]

I'm going to try testing it on a friend's PC today, I should just move the SSD and the GPU, right ?

[AhmedAm]

Another question, can unstable electricity from the grid wear the PSU faster ? Because that would explain why I had to replace my HDD on a yearly basis ; _ ;