Overclocking Causing Red Dots and Artifacting

[lasdant]

Hello,

Recently I decided to overclock my ASUS R9 290 DC2OC-4GD5, and it seems to be stable on most games I play. For whatever reason, it causes flashing dots to appear after I've been playing Skyrim for a bit. This happens almost every time I play Skyrim specifically, I'll play for a bit, it starts going crazy so I stop, and then I can start playing again. It seems to me like it's caused by overheating, but AMD cards run way hotter than normal, and AMD have said that the card is good even in the 90C range. Here are my specs;

CPU: AMD Ryzen 5 1600 @3.2GHz
MOBO: MSI B350 Tomahawk ATX AM4
RAM: Corsair Vengeance LPX 8Gb x1
GPU: ASUS R9 290 DC2OC-4GD5

OVERCLOCK
Core Voltage: +13mV
Power Limit: +50%
Temp Limit: 94C
Core Clock: 1000MHz -> 1020MHz
Memory Clock: 1260MHz -> 1380MHz

I used MSI Afterburner for overclock. Here's a picture of the monitor while I was playing, the rising temperature at the end is when I started playing, and when it drops off is when I had to close the game because it started freaking out.

https://imgur.com/a/XubeCnh

 

[velocityg4]

Sounds like the VRAM is overheating. The temps you see are normally just for the GPU. Not the VRAM. You can try setting a more aggressive fan curve in AMD Wattman. So, that the GPU fan is at a higher RPM at a lower temp.

 

[lasdant]

Sounds like the VRAM is overheating. The temps you see are normally just for the GPU. Not the VRAM. You can try setting a more aggressive fan curve in AMD Wattman. So, that the GPU fan is at a higher RPM at a lower temp.

Is there anyway to monitor VRAM temperatures? My fan curve is already quite aggressive, it normally doesn't go below 70% fan speed.

 

[Rogue Leader]

Back that memory clock down that should bring your temps down. Thats exactly whats happening, your memory is overheating.

 

[scout_03]

your overclcock is to high and see if you lower the game settings that could help .

 

[lasdant]

Back that memory clock down that should bring your temps down. Thats exactly whats happening, your memory is overheating.

I turned the memory clock back down to 1260MHz and it happened again, I play for a bit and then dots start appearing on the screen.

 

[Rogue Leader]

Back that memory clock down that should bring your temps down. Thats exactly whats happening, your memory is overheating.

I turned the memory clock back down to 1260MHz and it happened again, I play for a bit and then dots start appearing on the screen.

Interesting, still probably a heat issue.

IMO start over with everything at stock and see if it does it, and check your temps. Lets see what kinda range we have to work with here.

 

[lasdant]

Back that memory clock down that should bring your temps down. Thats exactly whats happening, your memory is overheating.

I turned the memory clock back down to 1260MHz and it happened again, I play for a bit and then dots start appearing on the screen.

Interesting, still probably a heat issue.

IMO start over with everything at stock and see if it does it, and check your temps. Lets see what kinda range we have to work with here.

Sorry didn't notice in time, I just tested by resetting everything back to stock except the core clock, which I left at 1020MHz. The red dots appeared again after a while, but now when I left and checked the temperature the card was sitting at 76C when it start freaking out. I'll try it again completely at stock settings.

 

[Rogue Leader]

Back that memory clock down that should bring your temps down. Thats exactly whats happening, your memory is overheating.

I turned the memory clock back down to 1260MHz and it happened again, I play for a bit and then dots start appearing on the screen.

Interesting, still probably a heat issue.

IMO start over with everything at stock and see if it does it, and check your temps. Lets see what kinda range we have to work with here.

Sorry didn't notice in time, I just tested by resetting everything back to stock except the core clock, which I left at 1020MHz. The red dots appeared again after a while, but now when I left and checked the temperature the card was sitting at 76C when it start freaking out. I'll try it again completely at stock settings.

20 mhz should not cause artifacts. Let me know what happens.

 

[10tacle]

Back both memory and core to stock speed and see if things return to normal. It could be that your GPU just doesn't like to be overclocked. I'm not sure about AMD cards, but ASUS cards for Nvidia are notorious for being among the bottom of the AIB board partners for overclockability.

 

[Neur0nauT]

I agree that it sounds like overheating VRAM/VRMs on your GPU. In my past experience, if you ignore these artefacts, you run the risk of killing your graphics card....so be wary if you are deciding to overclock.

AIDA 64 has an on screen display functionality that you can set multiple temperature diodes from your system including the VRMs on your GPU which would correlate to overheating VRAM. I have it setup to display on my second monitor whilst in game. That way I can monitor all system temps.

 

[R0GG]

Although your card accused a mild overclock it's overheating disproportionally, most card do but AMD R9 290 series are renowned for that, and despite showing an overall within specs temperature, some of the card internal components might be suffering from excessive heat.
- Increase graphic card fans RPM by adjusting the fans curve making it more abrupt to get fans to spin quicker for same given temperature, and " learning" card behavior with temperature, airflow and overclocking.
-Key element is improving the case airflow and CPU cooling: These cards need a real good case airflow with strong 120mm fans with configurable speed at bottom and top of card by drive cage blowing horizontally sustained air currents above and below the card
- incremental small steps overclock/card testing (cards are different and so is there environment, so results may vary)
- And finally getting a card to run at higher clocks does not guaranty a better performance or higher usage, at times it even has negative effect on the card usage and performance not to mention overheating and that is even for a stable overclock without any signs of instability or artifact.

 

[lasdant]

Back that memory clock down that should bring your temps down. Thats exactly whats happening, your memory is overheating.

I turned the memory clock back down to 1260MHz and it happened again, I play for a bit and then dots start appearing on the screen.

Interesting, still probably a heat issue.

IMO start over with everything at stock and see if it does it, and check your temps. Lets see what kinda range we have to work with here.

Sorry didn't notice in time, I just tested by resetting everything back to stock except the core clock, which I left at 1020MHz. The red dots appeared again after a while, but now when I left and checked the temperature the card was sitting at 76C when it start freaking out. I'll try it again completely at stock settings.

20 mhz should not cause artifacts. Let me know what happens.

Done, no artifacting or craziness happening on screen, card was running between 70-75C. Interestingly enough, I think the card is downclocking itself to combat heat, when the card reached 75 the core clock dropped to about 500MHz, and then once the temps dropped to 65, the core clock went back up to the base 1000MHz.

https://imgur.com/a/Ws2meF3

 

[lasdant]

Although your card accused a mild overclock it's overheating disproportionally, most card do but AMD R9 290 series are renowned for that, and despite showing an overall within specs temperature, some of the card internal components might suffering from excessive heat.
- Increase graphic card fans RPM by adjusting the fans curve making it more abrupt to get fans to spin quicker for same given temperature, and " learning" card behavior with temperature, airflow and overclocking.
-Key element is improving the case airflow and CPU cooling: These cards need a real good case airflow with strong 120mm fans at bottom and top of card by drive cage blowing horizontally sustained air currents above and below the card
- incremental small steps overclock/card testing (cards are different and so is there environment, so results may vary)
- And finally getting a card to run at higher clocks does not guaranty a better performance or higher usage, at times it even has negative effect on the card usage and performance not to mention overheating and that is even for a stable overclock without any signs of instability or artifact.

Currently I only have 3 case fans, two intake in the front and one exhaust in the back, that might be the culprit for poor airflow. I'll get some more fans that I can put in the top of the case as well.

Is it better to have more intake or exhaust fans? I can't quite remember.

 

[Rogue Leader]

Back that memory clock down that should bring your temps down. Thats exactly whats happening, your memory is overheating.

I turned the memory clock back down to 1260MHz and it happened again, I play for a bit and then dots start appearing on the screen.

Interesting, still probably a heat issue.

IMO start over with everything at stock and see if it does it, and check your temps. Lets see what kinda range we have to work with here.

Sorry didn't notice in time, I just tested by resetting everything back to stock except the core clock, which I left at 1020MHz. The red dots appeared again after a while, but now when I left and checked the temperature the card was sitting at 76C when it start freaking out. I'll try it again completely at stock settings.

20 mhz should not cause artifacts. Let me know what happens.

Done, no artifacting or craziness happening on screen, card was running between 70-75C. Interestingly enough, I think the card is downclocking itself to combat heat, when the card reached 75 the core clock dropped to about 500MHz, and then once the temps dropped to 65, the core clock went back up to the base 1000MHz.

https://imgur.com/a/Ws2meF3

Temps in the 70's is fine. It sure sounds to me like you have a GPU that came form the lower end of the pile and doesn't like being overclocked. I take back the memory overheat problem. if you are adding only 20mhz to the core clock, and you get artifacts, then your GPU is not up to snuff at all.

 

[Neur0nauT]

Done, no artifacting or craziness happening on screen, card was running between 70-75C. Interestingly enough, I think the card is downclocking itself to combat heat, when the card reached 75 the core clock dropped to about 500MHz, and then once the temps dropped to 65, the core clock went back up to the base 1000MHz.

My guess is that your GPU's VRMs are overheating. I had a 280X and it's core regularly ran at about 90c full load! My 390X has been known to hit 94C under load...These cards are designed to run hot, so 75 is nominal to be honest. You should not be artefacting at 75C.

Now when I did notice artefacts on my 280x before it died, the VRMs were always maxing out at nearly 100c. Go and download a trial for AIDA 64 and activate the temperatures on your GPU VRMs with the on-screen-display. You'll probably find they are maxing out at about 100C when the artefacts are showing.

Another tip...if your card is overheating....try using the FPS cap in the Adrenalin control panel. I have noticed my 390X VRMs never go beyond 82c since I have started capping the frame rate in games. If I switch off capping.....it jumps up to 100C when being pushed.

 

[lasdant]

Back that memory clock down that should bring your temps down. Thats exactly whats happening, your memory is overheating.

I turned the memory clock back down to 1260MHz and it happened again, I play for a bit and then dots start appearing on the screen.

Interesting, still probably a heat issue.

IMO start over with everything at stock and see if it does it, and check your temps. Lets see what kinda range we have to work with here.

Sorry didn't notice in time, I just tested by resetting everything back to stock except the core clock, which I left at 1020MHz. The red dots appeared again after a while, but now when I left and checked the temperature the card was sitting at 76C when it start freaking out. I'll try it again completely at stock settings.

20 mhz should not cause artifacts. Let me know what happens.

Done, no artifacting or craziness happening on screen, card was running between 70-75C. Interestingly enough, I think the card is downclocking itself to combat heat, when the card reached 75 the core clock dropped to about 500MHz, and then once the temps dropped to 65, the core clock went back up to the base 1000MHz.

https://imgur.com/a/Ws2meF3

Temps in the 70's is fine. It sure sounds to me like you have a GPU that came form the lower end of the pile and doesn't like being overclocked. I take back the memory overheat problem. if you are adding only 20mhz to the core clock, and you get artifacts, then your GPU is not up to snuff at all.

I mean I have had the card for almost 4 years now, might have something to do with it. But I suppose I just have to live with running it at stock speeds. Doesn't seem to like being overclocked.

 

[Rogue Leader]

I mean I have had the card for almost 4 years now, might have something to do with it. But I suppose I just have to live with running it at stock speeds. Doesn't seem to like being overclocked.

If its been overclocked that whole time the VRMs may have worn out. I mean your overclock isn't crazy but these GPUs have never been known for being super durable in overclocking.

 

[lasdant]

If its been overclocked that whole time the VRMs may have worn out. I mean your overclock isn't crazy but these GPUs have never been known for being super durable in overclocking.

I've only started overclocking recently, for the most part I've been using it at stock speeds. Now that it's reaching the end of its life and I don't really care what happens to it anymore, I've been having some fun pushing the limits.

 

[Rogue Leader]

If its been overclocked that whole time the VRMs may have worn out. I mean your overclock isn't crazy but these GPUs have never been known for being super durable in overclocking.

I've only started overclocking recently, for the most part I've been using it at stock speeds. Now that it's reaching the end of its life and I don't really care what happens to it anymore, I've been having some fun pushing the limits.

If thats the case then I would up the power limit as high as you can go and try adding some voltage. You may well still be able to get some non artifact extra power. GO SLOW, you don't want to just cook the thing, but setting the power limit to its highest level just tells the GPU it can draw more, it doesn't necessarily feed it more, only what it needs.

Worth a try, and you already know what happens when it doesn't work.

 

[lasdant]

If thats the case then I would up the power limit as high as you can go and try adding some voltage. You may well still be able to get some non artifact extra power. GO SLOW, you don't want to just cook the thing, but setting the power limit to its highest level just tells the GPU it can draw more, it doesn't necessarily feed it more, only what it needs.

Worth a try, and you already know what happens when it doesn't work.

Just gave it a shot, turned up power limit to the max and voltage to +13mV, and it caused artifacting, but a completely different kind. Normally when it was happening before it was causing different colored dots to flash on screen, but now it's causing pixelated squares to flash at me all over the screen in addition to the dots. Here's what the monitor looked like while under load.

https://imgur.com/a/ZHBu71r

 

[Rogue Leader]

If thats the case then I would up the power limit as high as you can go and try adding some voltage. You may well still be able to get some non artifact extra power. GO SLOW, you don't want to just cook the thing, but setting the power limit to its highest level just tells the GPU it can draw more, it doesn't necessarily feed it more, only what it needs.

Worth a try, and you already know what happens when it doesn't work.

Just gave it a shot, turned up power limit to the max and voltage to +13mV, and it caused artifacting, but a completely different kind. Normally when it was happening before it was causing different colored dots to flash on screen, but now it's causing pixelated squares to flash at me all over the screen in addition to the dots. Here's what the monitor looked like while under load.

https://imgur.com/a/ZHBu71r

Squares is worse. Yeah I think stock is it, don't mess with it.

 

[lasdant]

Squares is worse. Yeah I think stock is it, don't mess with it.

Seems like it, voltage was definitely the offender there, leaving everything at stock and turning power limit up to the max seems to be running just as smoothly as stock.

 

[Rogue Leader]

Squares is worse. Yeah I think stock is it, don't mess with it.

Seems like it, voltage was definitely the offender there, leaving everything at stock and turning power limit up to the max seems to be running just as smoothly as stock.

Yeah power limit doesn't do anything unless the GPU needs the extra power, such as if you're overclocking.

 

[Neur0nauT]

Recently I have seen a bunch of people with 290X's and different "Southern Island" cards end up re-pasting their heat sinks with positive results. Have you considered doing this? It may improve the temps on the card overall and give you that required headroom for OCing. Since you've nothing to lose.