Like almost people with this card I got a lot of random black screens/reboot while playing/benchmarking this card.
I'm using a MSI RX 5700 XT Gaming X.
Like a lot of people, I tried almost everything, WITHOUT SUCCESS :
- clean windows install
- downclock RAM
- downgrade PCI-E to 3.0
- underclock GPU to standard frequencies
- boost fan curve
- Use 2 separates 8pins cables for power
- Change driver version
It's completely random.
My hardware is all up-to-date for drivers. I don't have any bios updates from MSI.
According the number of people with this issue, can we have an official response from AMD and a fix?
- B550-itx/ac (no OC)
- Ryzen 7 5800X
- Cooler Master 850W 80+ Gold
- Corsair XMS 3200 16GB
- MSI RX 5700 XT Gaming X
I can confirm:
january 2021, with adrenaline 21.1.1 and i'm on the pretty same situation as one year ago: I bought a ("expensive") gaming video card... and I can't really enjoying it.
I encourage anyone who still has issues with the various 5700XT cards to post their specs and make this post grow to finally get a solution from AMD.
As you say, many of us have this problem. And with graphics cards of different models, it happens to me with a Gigabyte Radeon RX Vega 56 Gaming OC 8GB. From everything I have read from other people and all the tests that I have been able to perform, we can only wait for a correction of the drivers by AMD.
Have a look at this:
system freezes with custom game profiles - AMD community
My own RX 5700 XT has been stable for 30+ days on "Standard" and "global".
I have set my boost clocks to advertised speeds from manufacturer and undervolted,
with custom fan curve that regulates junction temp below 90°C.
>are you using the Radeon Software on "Standard" profile? And are you using the "global" tuning profile or do you have game profiles installed?
"standard profile" and "global". I always did a DDU (tried amd clean up utility too) before installing a new driver.
All setting are untouched. I just tried for some days to underclocking at -5%... but that not resolved, so i reverted back to default.
I can run Furmark without problem reaching a junction temp of 110°C and a max GPU temp of 70°C, but yesterday I had a crash playing League of legends (really low GPU usage and temps). Games made in Unreal Engine are the most problematic. Sometimes I can play for hours... sometimes the games crashes with the Unreal Engine GPU "HUNG" message.
Yeah I have an RX 6800 and have the same problem but its with all my games but I agree the biggest issue is with games that run on Unreal Engine and Unity
Reinstalling the drivers seemed to help for a bite but not any more and it's gotten so bad for me that no matter what game I try to boot up they crash before the game is completely loaded up it crashes and I errors like this
GPU Crashed Or D3d Device Removed
So at the moment, I can't even play games and from what I can tell its driver related and I have tried so many things but I still can't find a fix I have not found one forum that actually has a fix that works.
For some of the game that I could get crash logs for I have sent to the developers of the game and they have said there is nothing they can do its all driver related
This was from the developers of rim world
"Ah ok a real rimworld/unity crash.
After investigating with the error
"d3d11 failed to create rendertexture error 0x887a0005""
@mrricciolo I am running my own RX 5700 XT with these settings:
P3 boost frequency: 1950 MHz
-> results in actual clock around 1895-1905 MHz under load
P3 boost max voltage 1050 mV
-> results in actual voltage 1041-1052 mV
My fan curve is also set to 100% at 85°C to prevent overheating.
Did you try anything similar?
As in: values close to these, rather than factory clocks.
I will try your P3 State.
however I don't think that know and go modify advanced settings has to be a required skill to avoiding system crashes.
A video card MUST run well on standard condition. in fact, advanced setting are under "user responsibility"
@mrricciolo Yes, I agree with you.
Buying a graphics card should not require any skill to avoid crashes.
After reading a lot of forum posts in several different communities,
my best guess is: AMD tried to reach nvidia's performance by increasing
the max clocks and therefore also overvolting the 5000 series cards by default.
Radeon Software enables all 5000 series cards to boost very high by default
with quite a bit of voltage, while most cooling solutions struggle to keep the 200-225W TDP cooled.
I have had really bad experience with my own RX 5700 XT right after building this system.
Instant crashes/freezes and super high temperatures.
My card was boosting way beyond 2050 MHz with default Radeon Settings.
This is above factory settings as well - which are 1905 MHz for my card.
I'm using the standard profile.
The only thing I change is the fan to get more RPMs.
I'm gonna try to reduce voltage, and limit the frequency (it goes to 2050mhz).
Can you share your voltages/frequencies settings?