Yes, GPU overclocking can lead to unstable behavior and glitches.
Yes, GPU overclocking can lead to unstable behavior and glitches.
So I configured my 6600K to 4.2Ghz stable, which is the best I can achieve. It might be possible to increase it further, but I believe it’s more than sufficient right now and I haven’t had to adjust the voltages.
I obtained my GPU (MSI GTX 1060 6GB) with an afterburner setting at 2252Mhz. In AfterBurner, the Core Voltage is at 0%, Power Limit is at maximum 108, Temperature is at 92, Core Clock is 150, and Memory Clock is 500.
I managed to reach 160 during a stress test, which caused the card to heat up quickly—around 60°C fast—but once the fans activated, it settled between 60-61°C, maybe a bit higher. At 160, Witcher 3 Ultra Hair works on AA8 crashes. At 150 it functions properly.
I think this might be due to needing more voltage, though. My main concern is that the graphics appear unstable, with visuals disappearing and reforming rapidly across the screen (extremely subtle, hard to capture a clear image). This issue also appears on my wallpaper.
Is this related to the overclocking? It seems like the card isn’t receiving enough voltage to maintain the higher clock speed. Not entirely sure—I’m new to overclocking, but I know MHz can increase, yet there’s a limit where more voltage isn’t desirable. Always monitor temperatures and performance.
Questions:
1. What voltage range would be best for long-term health?
2. Can overclocking lead to graphical glitches instead of crashes?
3. What is the typical average overclock setting on this card? (Did I win the Silicon lottery? 😅)
Slide everything fully to the right, you won't lose any lifespan. Also, if your power cap hits its limit, detach power and temperatures, set temperature to 83. 2.) Yes, the "bugs" are known as artifacts and can show up as black holes, misaligned 3D elements, random color shifts, odd boxiness, and strange behavior. Adding more voltage or reducing it will resolve these issues. I believe you've won the silicone lottery—most users are around 1700 core LCOK, but stability is still needed. This guide contains a lot of useful information; it helped me upgrade my GPU OC, but I suggest using FireStick or Time Spy instead of the original.
If you're looking for a percentage related to the card's longevity, I'd advise against overclocking and you're right about that. It's unclear what the typical overclock setting is for this card, but 2252 MHz seems quite high compared to its idle range of around 1700 MHz. If you didn't experience issues before the overclock, it likely points to the overclock itself being the problem.
From what I've understood, the average is OC that most of these cards can achieve is around 2100. I'm just hoping the silicon performed well in my case.
Slide everything all the way to the right, you won't lose any longevity. Also, if your power limit hits the maximum, disconnect power and temps, and set temp to 83. 2.) Yes, the "bugs" are known as artifacts and can show up as black voids, misaligned 3D objects and textures, random color splashes, odd boxiness, and strange behavior. Adding more voltage or reducing it will resolve this. I believe you've won the silicone lottery—most users are around 1700 core LCOK, but stability is still needed. This guide has a lot of useful information; it helped me get my GPU OCed. I suggest using FireStick or Time Spy instead of the default.
From what I've understood, the average is OC that most of these cards can achieve around 2100. I'm just hoping the silicon performed well.
You're correct about not encountering glitches, but I've run into several since installing this new hardware and updating to fresh Windows 10. I'm trying to see if this is typical for OC'd GPUs.
If you overclock too much, you might get digital artifacts along with other problems. Try reducing the clock speed gradually until it stabilizes. Anything pushed beyond factory limits can cause issues—keep experimenting until you find a balance. Also, check for any new driver updates since this card is relatively new.
I believe you should increase the core voltage (set the slider fully, as the GPU manufacturer recommends for safety) and if artifacts persist, reduce them gradually. If they disappear, raise it again before they return.
Adjust it all the way to the right, you won't lose any longevity. Also, if your power limit is at its maximum, unlink power and temps, set temp to 83.
Yes, the "bugs" are referred to as artifacts and can show up as black voids, misaligned 3D objects and textures, random splashes of color, odd boxiness, and what's happening to you now. Adding voltage or stepping back will resolve this.
I believe you've won the silicone lottery, most people are around 1700 core LOC, but that's only confirmed once stability is achieved.
This has a lot of useful information; it was how I achieved my GPU overclocking, though I suggest using FireStick or Time Spy instead of just guessing.
I wonder if going higher than 500 on the memory slider is wise? It's hard to say exactly how much it would impact performance.
I wan't talking about memory, I meant core voltage.
Here's how I OC a GPU:
1. Set my core voltage to max allowed by Afterburner (its safe and wont lower card lifespan)
2. Set my power and temp limit according to card (for a GTX 1060, un-link the two, set power limit to as high as it will go and temp limit to 83)
3. If you haven't already, install 3D mark Firestrike (it's a far better and more realistic test than valley, like recommended on the guide I posted)
3. Start to up the core clock.
Do this by raising it in increments and monitoring your temps each run-though. I usually start out by adding 100, then 50, then 25... If any of these jumps prove to much, I make a smaller one.
To check if your OC is stable, look carefully at the graphics each run-though, if you notice some artifacts, you've gone too high. Keep an eye on your temps also.
When you finally reach that magic clock speed where if you raise it just one or two more and you get artifacts, back off 5 MHz for stability.
4. Then do the memory clock this same way.
5. Stress test. If temps go to high or you get artifacts, back off in increments of 5.
6. Brag about how high it went
When encountering such images during stress testing, reduce the core clock. If you notice similar issues in your tests, monitor the memory clock in very small steps. These are extremely intense scenarios.