The RX 6000 series Owners' Club

[GamerGuy]

85c hotspot is beyond fine. You are not cooking anything.

Yep, I'm even fine seeing 90C max on the Hotspot temp, it's only if it's above that that my OCD kicks in.

 

[Frick]

Instead of starting a thread I'm just going to say that something's up with my 6950xt. When the secondary monitor is on (old 1080p on HDMI) is on games on the main monitor (Samsung U32J590U on DP) occasionally has white lines/artifacts flashing. It's like some rows gathered in the top third of the monitor flash white very briefly. It does not happen when the secondady monitor is turned off. This is new behaviour, not related to drivers. No overclocks or undervolting, just standard settings. So something's up. I'm assuming GPU, but maybe it could be the motherboard but I don't see why. Or PSU.

 

[AusWolf]

Instead of starting a thread I'm just going to say that something's up with my 6950xt. When the secondary monitor is on (old 1080p on HDMI) is on games on the main monitor (Samsung U32J590U on DP) occasionally has white lines/artifacts flashing. It's like some rows gathered in the top third of the monitor flash white very briefly. It does not happen when the secondady monitor is turned off. This is new behaviour, not related to drivers. No overclocks or undervolting, just standard settings. So something's up. I'm assuming GPU, but maybe it could be the motherboard but I don't see why. Or PSU.

It sounds like VRAM failure to me, but I don't understand why you only see it with two monitors, not with one. Can you try the card in a different system with the same monitor setup?

 

[TheBense]

Pardon my ignorance, the last time I've messed with this stuff was March 2015 with an R9 290.

I have a Gigabyte Aorus RX 6900 XT Xtreme Waterforce WB GPU that I installed in my X399 Threadripper machine, (clickable link)

My machine is unstable. I get random lockups whenever I'm doing tasks that do not put stress on the GPU such as switching between browser tabs.

I have switched between drivers, different installs of windows, I've tested RAM, disabled XMP on my RAM, switched out RAM entirely.
Switching between browser tabs, disabled MPO. I've removed the waterblock from the GPU and I've reapplied thermal paste and pads. The only thing that seems to have helped is underclocking the GPU core by about 30MHz. However, it really is hit or miss as to when this thing will lock up.

What I am considering doing is, flashing the vbios to a more conservative vbios with more conservative clocks such as this PowerColor Liquid Devil RX 6900 XT Ultimate. However, this is not a dual vbios card, so I am a bit reluctant.

I have used GPU-Z to dump the vbios, then searched google with the md5sum of it, and found this.


If any of you have any insight, suggestions, please advise.

 

[Veseleil]

Pardon my ignorance, the last time I've messed with this stuff was March 2015 with an R9 290.

I have a Gigabyte Aorus RX 6900 XT Xtreme Waterforce WB GPU that I installed in my X399 Threadripper machine, (clickable link)

My machine is unstable. I get random lockups whenever I'm doing tasks that do not put stress on the GPU such as switching between browser tabs.

I have switched between drivers, different installs of windows, I've tested RAM, disabled XMP on my RAM, switched out RAM entirely.
Switching between browser tabs, disabled MPO. I've removed the waterblock from the GPU and I've reapplied thermal paste and pads. The only thing that seems to have helped is underclocking the GPU core by about 30MHz. However, it really is hit or miss as to when this thing will lock up.

What I am considering doing is, flashing the vbios to a more conservative vbios with more conservative clocks such as this PowerColor Liquid Devil RX 6900 XT Ultimate. However, this is not a dual vbios card, so I am a bit reluctant.

I have used GPU-Z to dump the vbios, then searched google with the md5sum of it, and found this.


If any of you have any insight, suggestions, please advise.

You bought the card used I presume? Have you stress tested it after purchase? Please try this tool - https://github.com/GpuZelenograd/memtest_vulkan

It will spit out errors visible in the CMD window. Another thing to look for is wildly oscillating performance in GB/s, just make sure you're not using the PC during the test as it can throw off the performance metric.

How long you've been using the card, and for which tasks exactly, games or work? Did you notice lockups while using the GPU at full load (demanding games, unlocked FPS)?
Did you try to underclock it further, let's say slider in adrenaline to 2300MHz? It's my setting for 99% of usage scenarios, as I like to game quietly and efficiently under 200W. Please share more info about the GPU itself. Have you tried using it in another mobo slot, also using a different GPU and try to replicate the issue and eliminate the rest of the system as a possible cause of the problem?

 

[AusWolf]

I've just installed the Space Marine 2 beta driver (24.10.something), and now the driver control panel annoyingly opens every time my monitor wakes up from sleep. Anyone noticed anything similar?

 

[physicsjunky]

yeah, adrenaline has been doing that recently I'm not sure if there is a fix. I don't think the AFMF 2-V2 beta driver is doing that.

 

[MrBanana]

I've just installed the Space Marine 2 beta driver (24.10.something), and now the driver control panel annoyingly opens every time my monitor wakes up from sleep. Anyone noticed anything similar?

It's doing the same for me since 24.8.1.

 

[Deleted member 223428]

I've just installed the Space Marine 2 beta driver (24.10.something), and now the driver control panel annoyingly opens every time my monitor wakes up from sleep. Anyone noticed anything similar?

Yes, with the latest (WHQL 24.8.1) on my laptop that's a combo of RDNA2 (680M) and RDNA3 (7700S). Going back to 24.7.x 'fixed' it.

 

[Lew Zealand]

I think I noticed the same on my 780m / 6600 XT combo after updating to 24.8.1, I guess it's a weird but harmless bug? But I've swapped the 6600 XT out for something else for testing recently and it hasn't happened since then.

 

[Veseleil]

I've stopped using drivers from the year 2024, and went back to the latest driver from 2023. Enjoying life ever since. Although one might think that the game support drivers are essential for that specific game, they might be, but it's not a rule. With Horizon: Forbidden West I've experienced exactly the opposite. I had constant game crashes, and that's when I went back to check. Got at least 30FPS more, and the game never crashed again. Also noticed large FPS boost in almost all other games I played at the time. My conclusion is that after 23.12.1, the drivers aren't optimized for RDNA2 anymore.

 

[Terronium-12]

I'm having a bit of a trouble with More Power Tool and my 6900 XT.

So, for point of reference I have a MERC 319 Black (as you can peep here) and the base power limit is 289W, with a TDC limit of 320A as you can see in the picture below:

I set the power limit to 310W and the TDC to 330A (as you can see above), wrote it to the registry and have noticed some anomalies since. The Radeon performance overlay would bug out and flash on and off (a restart seemed to have fixed it) and it's hard reset twice now. I'm wondering what the issue could be, because I highly doubt that 310W (355 or so if I crank up the limit to the max 15%) is too much for the card's rather robust power delivery layout. Could it be that I need to DDU the drivers and install them from a blank slate?

Could I just be unlucky with the GPU silicon?

 

[Veseleil]

I'm having a bit of a trouble with More Power Tool and my 6900 XT.

So, for point of reference I have a MERC 319 Black (as you can peep here) and the base power limit is 289W, with a TDC limit of 320A as you can see in the picture below:

View attachment 371094

View attachment 371098

I set the power limit to 310W and the TDC to 330A (as you can see above), wrote it to the registry and have noticed some anomalies since. The Radeon performance overlay would bug out and flash on and off (a restart seemed to have fixed it) and it's hard reset twice now. I'm wondering what the issue could be, because I highly doubt that 310W (355 or so if I crank up the limit to the max 15%) is too much for the card's rather robust power delivery layout. Could it be that I need to DDU the drivers and install them from a blank slate?

Could I just be unlucky with the GPU silicon?

IDK about MPT, as I never used it, but I can recommend complete DDU cleanup, and what I mentioned in the last sentence of my previous post (above yours, so to not repeat myself over the matter).

 

[chris189]

@Terronium-12 May you consider flashing your BIOS? There is a ton & I mean a ton more options that you really need to consider changing bro. If you desire to keep your card alive forever & ever, that is.

You may need to search or even upload your BIOS (Make sure you have dual BIOS switch) & ask BITBY.RU AMD VBIOS FIRMWARE DECRYPTION or if you can find out who can decrypt the RDNA 2 VBIOS for flashing. It's important to flash to give your card the freedom & performance & life it deserves. Rather than the torture AMD bestows unto us with it's 118C temperature limits.

Later

 

[Jism]

I've read somewhere that the latest drivers of AMD seem to block any changes done with MPT?