Graphics Cards

[Sisco88]

I kust got this card a few weeks ago and it started showing some strange pink and sometimes blue artefacts. I have noticed that they sometimes disappear when setting an fps limit in certain games (ff14) or lowering the core voltage and clock speeds. Is the card defective?

Video:

https://drive.google.com/file/d/1lTp24IiMVLTRGgkPMxViaxGhOx9Et-NK/view?usp=drivesdk

Thanks!

Sisco

[ThreeDee]

What are the rest of your PC specs? .. the more detail the better .. include Make/Model of power supply too

Make sure you are running a separate power cable to each power input on your GPU

Run DDU and remove past drivers if you ran an Nvidia card prior to your current AMD GPU as well as all things AMD .. reboot .. and THEN install latest AMD GPU drivers from AMD.com

Make sure you have latest chipset drivers installed. (Intel.com for Intel, AMD.com for AMD)

Make sure your Windows 10/11 install is up to date

Make sure you have good air flow in & out of your case

[Sisco88]

Hello and thanks for the reply!

PC specs:

MSI B550M PRO-VDH wifi with latest BIOS
Ryzen 7 5800x
16GB DDR4 3600
Cooler master V850 sfx power supply

Fully updated windows 11 and all the latest drivers installed.

I currently have 2 power cables running from the power supply to the card of which one has 2 connectors. i will try adding a third separate cable and see if that does anything.

Sisco

[ThreeDee]

What case?

What motherboard? BIOS up to date?

How is the air flow in your case?

CPU/GPU temps?

How are you cooling your CPU?

Are you running any modded PSU cables? .. PCIe Riser cable?

[Sisco88]

MOBO is a msi b550m pro-vdh wifi  with the latest BIOS

Case is a JONSBO V9 with 2x120mm intake at the bottom and 2x120mm exhaust at the top. CPU is cooled with an alphacool eisbaer AIO. No modded or riser cables used.

CPU temp never reaches anything over 75C.
GPU temp highest is around 70C at 100% GPU usage with a maximum hotspot of 90C.

Sisco

[Geforsikus_2021]

Yes, it is possible, as we wrote above, that this is due to a lack of power, if the video card has 3 connectors of 8 pins, or 2 8 pins and one 6 pin, in any case, you need 3 cables separately from the power supply. If this is a reference consumption of up to 300 watts if from partners, perhaps up to 350 and higher depends on the model of course) one cable for 8 pins holds 150 watts, and for 6 pins 75 watts (separately)

[dude_98284]

You may want to start with the PSU. I find that a majority of GPU issues come from lack of power and trust me when I say it can cause some very odd and random issues.

[ThunderBeaver]

Running the Power Color Demon RX 6900 XT.

I was getting occasional black screen with my PC still running normal.

Turned out one of my 3 8 pin PCIE power cables was not fully seated into my Thermaltake Toughpower 1200 watt PSU.

These series of cards are not only power hungry but are also very finicky if they don't have a smooth constant power flow.

As for the odd screen artifacts same issue here. All early access games get horizontal tearing and triangular shaped flickering on ground and water surfaces. That I actually expected as these games are not finished and fully optimized yet.

I also found from some game developers that they place optimization for Invidia based GPUs ahead of AMD GPUs.  

A lot of my games work great though with no issue. How ever I have found through testing that you must keep tight control of your FPS with these cards. They will try to push FPS of up to 1048. I keep mine at a minimum of 60Hz and a max of 120Hz. Typically see FPS from 80 to 140Hz. For some reason all games get glitchy and unstable beyond 200Hz. 

I'm running a Viewsonic 50" OLED UHD 4K Smart Monitor with a max refresh of 240Hz and HDMI cable is rated up to 360Hz.

[Sisco88]

SO i currently have 3 separate 8 pin cables connected and the problem persists. The funny thing is when running FurMark the artifacts don't show up while the power draw is about the same as when running some games.

For example when running Halo Infinite in menus i am getting these artifacts with a power draw of  around 210W. I can also hear some coil whine.

While FurMark runs the power draw is around 250-260W without any artifacts.

[ThunderBeaver]

Unfortunately I don't have an answer to help solve the glitch.

I've seen several live play FPS games on YouTube running 6800 and 6900 series GPUs and a lot of them show some minor to moderate glitching.

Check those sources and see if they found solutions to the glitch/flickering issue.

So far the games that I experience this issue with still do it.

I find that YouTube has a lot of DIY solutions from game developers players programmers and PC builders. 

I think a lot of the glitching flickering screen tearing is due to needed optimization from the games Microsoft and AMD.

Considering this is the latest GPUs they're probably still working out the software bugs. 

[Geforsikus_2021]

Try to run the occt test on vram (you can set it to 99% with a slider), fullmark does not test the memory of the video card, but only the GPU crystal itself, if errors start pouring in, try to reset the frequency of RAM, or disable the xmp profile, or just manually lower the frequency of RAM to 3400 or 3200, and run the test again. the test is exactly vram. I hope that the problem is not in windows 11)

[Sisco88]

I am currently running the test and it seems to be running fine the past 30 minutes. I have uploaded another video showing the artefacts ingame and not in furmark.

https://drive.google.com/file/d/1lpFpbYup_5_CJFJhUG1UMjKq2egwkzPh/view?usp=sharing

EDIT: I have noticed that the artefacts appear behind the UI elements. and when the artifacts are on screen i can clearly hear a whining noise coming from the card.

regards,

Sisco

[Sisco88]

So after running the test for an hour, I did not see any of the issues. I did see random pixels on the screen changing color however.

What worries me is that there is a very noticeable coil whine every time the artifacts appear.

EDIT: I have found a video on youtube that shows the exact problem i am experiencing at around 1:33 so i guess the problem is memory related. i will try under-clocking the GPU memory tonight and update on the results.

https://www.youtube.com/watch?v=IDFmZXirzuo

Sisco

[Geforsikus_2021]

On the contrary, I would rather lower the memory frequency a little, it just all depends on the batch of chips, for example: there are memory chips that hold 4000 MHz at a voltage of 1.35v (I'm the proper memory), and there are those that I can't and at 1.4 -1.5 they can't. The same thing happens on the video card, if it can be a little bit of voltage and the problem will pass, or slightly lower the memory frequency, and the video card will be stable.)

[Sisco88]

Thanks for all your help, Gigabyte is not answering any of my questions so i really appreciate it.

So you are saying i should lower my system memory frequency instead of the card memory? I am using 2x8GB Crucial Ballistix 3600Mhz, Maybe trying to increase the voltage a little may help?

[Geforsikus_2021 · Accepted solution]

Try and raise the voltage usually if this is an xmp profile, it is 1.35V to put 1.38 on nothing else, or just lower the frequency from easy not 3600 to for example 3400 or 3200, let the manual xmp profile also stand. You can also try to slightly reduce the GPU frequency on the video you have 2400 +, for example, make 2350 MHz. Here's what I read: If you encounter artifacts or reboots, then do not rush to exchange the video card. The problem may be in the HDMI or DisplayPort cable. This problem occurs not only for owners of Radeon cards, but also for Nvidia. I have had periodic monitor flickers when using an Nvidia card, and after replacing it with a Radeon card, I observed exactly the same flickers.

For example, when using HDMI, there are flashing screens or artifacts in the form of incomprehensible dots or squares on the monitor screen. DisplayPort may have problems with the 20 pin, which can cause a reboot, there are two recommendations for this cable: use only proven and high-quality cables or glue (bite / crack) 20 contact on the cable. Alternatively, try to find another cable for the monitor, it is not so expensive, and the whistle of the drossels is due to the fact that the fps rises above 200 + can reach up to 1000 fps. I have an AORUS AD27QD monitor, I set 120 Hz, and in the amd adrenaline menu I set a limit of 115 fps + freesync. (connected via display port) And the drossels are not whistled and the card is not heated, even in heavy games))

[Sisco88]

I think you just saved me, enabling the FPS limiter in the Radeon software (max FPS 300) seems to have gotten rid of the artifacts (at least for now)! I'm still not a 100% convinced that the problem is solved but for now i'm happy! Will update when (if) the problem returns.

Thank you for al your time up until now!

Sisco

[Geforsikus_2021]

Glad I could help you, I hope the problem will not manifest itself anymore)