980ti showing signs of early death?

[Mazer]

I had this for 3 years now in another build been working good for me, runs all my games at ultra settings with some over 100+ fps made me buy a 1080p 144hz refresh monitor. Like a dream come true.
However, lately I've been playing games and noticing weird things in games for example, out of the blue I was playing Rage 2 and after some in game time it would freeze for a second and it did this randomly. Also seems to happen when playing battlefield 1 and V odd. I took a look with msi afterburner and saw my core clock was fluctuating from 1379 to 1366

I installed gpuz on this and saw the core problem along with dips in the gpu load which to me doesn't look right or maybe it's a boost thing? I'm not sure. I then tried to overclock it a little nothing big 30mhz+ on the core and to my surprise it wasn't stable when using heaven and firestrike also saw artifacts with no memory OC on firestrike what gives? I went back to stock and then it was fine ok so no OC. Can anyone help me in determining if my 980ti is dying or just really unlucky with it?

System specs:
Asrock b360 pro-v
Core i5 8400
Corsair LPX 16gb 2x8 ram 2666mhz
Gigabyte 980ti gaming G1
3x Samsung 860 evo 500gb
Corsair rm750x 80+ gold
NZXT s340s
Three exhaust fans
Two intake fans

 

[Vya Domus]

Seems alright to me.

Running any game at high framerates (144hz+) makes it more prone to stutter as your are starting to hit numerous other bottlenecks including the game engine itself. Makes sense you never noticed this if you have just recently switched to this 144hz monitor. The 980ti isn't exactly the fastest around the block either.

Signs of death would be artifacting, black screens, PC shuts itself, etc all at stock. I don't think there is a system in existence that wont stutter under certain conditions.

 

[xkm1948]

ANY overclocking will slowly degrade your silicon over time. It is normal.

 

[P4-630]

ANY overclocking will slowly degrade your silicon over time. It is normal.

How about overclocking vram.

 

[neatfeatguy]

I'd hate to think your 980Ti would be going out on you after 3 years, but it is always possible. I've just hit 4 years with my 980Ti and she's still going strong.

I've got a Zotac AMP! Omega that comes with a decent factory OC on the core and very minimal on the memory. I've had both the core and memory pushed up an extra 100/150MHz, respectively, since nearly day one. I can push them further, but I feel the extra 1-2% performance it may provide isn't worth maxing out the OC range on my card.

When it comes to OCing, it can very a lot. I've had multiple GPUs of the same brand/model to run in SLI. One card could overclock well, the other wouldn't take an OC to save its life. It happens. GTX 280s, I had one I could push the clocks up nearly 100 across the board and no issues, the other card if I moved the clocks even 10MHz and even pushed more voltage, she still crashed. I had a similar experience with 8800GTS 512MB I ran - both the 280s and 8800s were the same make/model (280s were BFG and 8800s were EVGA) and using the same BIOS versions.

I found that when it comes to OCing my 980Ti, I found I got better results for pushing the memory over the core. If you feel you want to try and OC a little, try pushing the memory up just a bit. I found I could OC the crap out of the memory on my 980Ti, I can push it up 250MHz and stay stable in games. Core, not so much, 150 above factory settings causes stability issues - so that's why I stuck with 100 on the core and 150 on the memory.

I'm not at home, but I use MSI Afterburner. I slide the power limit all the way up (I think it stops at 111% for my card), I setup a fan curve to run 80% speed once temps exceed 80C (which I never hit) and then I have a handful of steps all the way down to no fans running when under 35C. I started by moving up the memory by 25MHz and then running a few benchmarks to see how things go. If no instability, I move the memory up another 25MHz and repeat the benchmarks. Going off memory here, I think once the Maxwell cards start getting over 80C, they begin their throttle down process and 92C is the max the card will go before it really starts to pull back to keep the card cool. The cooler you can keep the card, generally the better OC you'll be able to sustain.

As for hiccups you're seeing, it's nothing out of the ordinary. I've got a few games this happens on that I've been playing, but I figure the hiccups to be tied more to the game engine itself than the hardware trying to run it.

 

[Dinnercore]

ANY overclocking will slowly degrade your silicon over time. It is normal.

Let me correct that: 'ANY *USAGE* will slowly degrade your silicon over time. It is normal.' And all the other components that are in use with it too. Like VRMs, caps etc.

 

[freeagent]

My 970FTW (6+6) does that, under certain loads it runs at 1404 and dips to 1396ish? I bought the card used.. the guy said he didn't oc it, and I think I may know why.. As soon as you try to change the clocks or even fan speed through afterburner or precision it throttles itself down into the 900s on the core, and will not run its stock clocks. If you just leave it alone its fine, runs good, especially for 150. My 980 (bought for 200) doesn't suffer from the same nonsense, it doesn't budge downwards. If I oc that one, it boosts on its own to 1 bin higher than what I set. My last card was a GTX 580 so still getting used to this boost stuff.

 

[Mazer]

What I'm most concern is the GPU load should it be consistent?

 

[Dinnercore]

What I'm most concern is the GPU load should it be consistent?

No it does not need to be consistent, in fact it will vary a lot during gameplay. The big dips hint towards a bottleneck somewhere else in the chain, be it game engine or other hardware not keeping up.

EDIT: It is very hard to see from the graphs, but the dips in GPU load seem to hit a small CPU temp spike every time indicating that the CPU may be the one holding it back. Either that or it has to wait for RAM to be cleared. Hard to say what exactly would help you avoid those spikes without seeing how much RAM usage you get in-game as well as CPU load during gameplay.

 

[EarthDog]

I don't see a problem here. GPU load pretty consistent and clock fluctuations are normal as that is how boost works.

 

[Mazer]

No it does not need to be consistent, in fact it will vary a lot during gameplay. The big dips hint towards a bottleneck somewhere else in the chain, be it game engine or other hardware not keeping up.

EDIT: It is very hard to see from the graphs, but the dips in GPU load seem to hit a small CPU temp spike every time indicating that the CPU may be the one holding it back. Either that or it has to wait for RAM to be cleared. Hard to say what exactly would help you avoid those spikes without seeing how much RAM usage you get in-game as well as CPU load during gameplay.

I redid a cinebench r15 now I'm getting worried because of the low score

 

[neatfeatguy]

I redid a cinebench r15 now I'm getting worried because of the low score

View attachment 126493

Don't sweat the small stuff.

A base clocked 980Ti is roughly 15% slower than a nice, highly OCed 980Ti. If you can't get your card to OC much, or at all, expect results from benchmarks to be on the lower end of the spectrum for that GPU. If you see folks posting results that show 10-20% better over what you can do, that's probably because of a higher OC on the GPU.

Just keep temps as low as possible by having good air flow and a clean case. Your card should be just fine for hopefully a few more years. If you start getting artifacts, driver crashes, lockups and such things, then you should start to worry about possible GPU failure.

 

[Dinnercore]

Amen to everything that @neatfeatguy guy said.

And dont worry about cinebench scores, they differ every time you run it and just windows doing background stuff while you click start can be enough to throw you down 5% or more.
On top of that, the GPU test is not really suited for testing recent gaming cards in my eyes.

To sum everything up, dont worry about the small hickups. Your system seems to be fine, if you want to do something for your peace of mind just clean it all out (dont forget the cpu cooler) and maybe try 3 intake 2 exhaust instead of 3 exhaust/2 intake fans to see if it makes any difference on the temps.