Sign in with
Sign up | Sign in
Your question
Solved

What else causes artifacts? OC GPU temps are 65C (cold for card) under full load

Last response: in Overclocking
Share
December 14, 2013 12:46:24 AM

What else causes artifacts if your OCed GPU is staying cool under full load? As an example, if a card is at 65C with a theoretical cap of 90C, what else could be causing the artifacts?

The artifacts in question are the black and white flashing triangles in multiple sizes and the white/black, white/green, white/red checkerboard squares in various sizes.
December 14, 2013 12:50:39 AM

Do you see this on the Windows desktop or specific games?
Have you installed the latest drivers?
m
0
l
a b K Overclocking
December 14, 2013 1:04:39 AM

Is it overclocked? If not, then it's defective (97% chance) or there's some strange driver issue (3% chance).
m
0
l
December 14, 2013 8:45:30 AM

The card in question is a r9 280x. It is over clocked and using the latest beta driver from AMD's website. It wasn't really geared towards my own card, but more of a broad question. To narrow down the reason as to why the card may be causing artifacts because temperatures are so low under full load.
m
0
l
a c 224 K Overclocking
December 14, 2013 11:30:39 AM

Low overclocked GPU temperature can be a safe illusion especially if you're overclocking the memory as the memory chips do not report their temperatures, you are only getting the GPU temperature.

You could be pushing your memory chips to the door of failure.
m
0
l
December 14, 2013 3:57:48 PM

4Ryan6 said:
Low overclocked GPU temperature can be a safe illusion especially if you're overclocking the memory as the memory chips do not report their temperatures, you are only getting the GPU temperature.

You could be pushing your memory chips to the door of failure.


Ryan,

Are you talking about the RAM memory or my GPU memory? And if it's my RAM memory, how do I go about checking its performance? I watch my GPU CPU PSU and MOBO but never the RAM.
m
0
l
a c 224 K Overclocking
December 15, 2013 7:12:43 AM

ztrob21 said:
4Ryan6 said:
Low overclocked GPU temperature can be a safe illusion especially if you're overclocking the memory as the memory chips do not report their temperatures, you are only getting the GPU temperature.

You could be pushing your memory chips to the door of failure.


Ryan,

Are you talking about the RAM memory or my GPU memory? And if it's my RAM memory, how do I go about checking its performance? I watch my GPU CPU PSU and MOBO but never the RAM.


I'm talking about the memory chips on the graphics card itself, they are not temperature monitored and if they overheat you won't know it until they fail.

Some graphics cards just don't have overclockable memory chips, some have overclockable memory chips that are not fully covered under the heat sink and are not being properly cooled.

Once you reached an artifacting stage in a graphics card you're pushing it too far with your overclocking, unless of course it was artifacting at stock speeds before overclocking then you have a factory bad card.

m
0
l
December 15, 2013 10:20:22 AM

You've overclocked the memory too much. Artifacts like that are memory-related, and could just be the luck of the card, or the memory is heating up too much, and you're only reading the core temperature. Reset the memory to defaults, and try overclocking slowly again. You tend to get white dots, when the core gets too hot. The core tends to be most affected by heat, IMO.

If you still get artifacts at stock on the memory, you have to RMA that sucka!
m
0
l
December 15, 2013 7:28:56 PM

Odd that GPU's memory temps aren't monitored because it's just as detrimental to the performance of the card as the core clock. Even though for instance, in my case, the r9 280x doesn't benefit much to OCing the memory. Doing a BF3 benchmark I only got a 1.2 fps increase at 123 frames from stock at 6400 (1600Mhz) to 7200 (1800MHz). So it seems rather pointless for my card. I had the memory OCed to 6800, so I have since brought it back down to stock.

The core clock on the other hand I had at 1220Mhz, from stock 1070MHz, which is quite the OC. I believe what was causing the black tears on my screen.

Before I was using OCCT to test the stability of my system, but I soon realized, even after 4 hours of 0 errors, the card was indeed not stable. Last night I downloaded Furmark and I noticed the tearing was happening at 1220MHz at around 69 degrees C. Tried a benchmark run in Unigine Heaven and actually noticed for the first time a couple green squares pop up at 1220 along with a few screen tears.

After turning it down to 1200MHz and reaching 76C in furmark with no tearing and several runs in Unigine Heaven 4.0 with what looked to be a perfect run, everything seems stable for now.

I was going to OC the memory but for a .5 fps increase, I figured I better not worry about it for now. I will definitely be RMAing the card if I see the memory pooping out of me at stock speeds.
m
0
l

Best solution

December 16, 2013 5:06:08 AM

An increase of 130 on the core is toasty. I only get 85--and that's after overvolting! ;) 

Glad you got it sorted out!
Share
December 16, 2013 10:43:19 AM

I was surprised to see that my core clock temps didn't go up very high after the OC. I only went up 5C or so bringing up the voltage to 1.290. Haven't gone past 65C in any of my most GPU intensive games yet. Tomb Raider and BF4.

I appreciate the input Ryan and Alex!
m
0
l
May 1, 2014 5:10:13 PM

ztrob21 said:
4Ryan6 said:
Low overclocked GPU temperature can be a safe illusion especially if you're overclocking the memory as the memory chips do not report their temperatures, you are only getting the GPU temperature.

You could be pushing your memory chips to the door of failure.


Ryan,

Are you talking about the RAM memory or my GPU memory? And if it's my RAM memory, how do I go about checking its performance? I watch my GPU CPU PSU and MOBO but never the RAM.


RAM = Random access memory, its redundant to say RAM memory. Your motherboard and your GPU both have RAM - calling it "ram memory" doesnt differentiate between the two. your gpu's RAM is obviously what he is referring to, since you are overclocking your GPU (core and ram)
m
0
l
!