• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

[Fixed in 16.9.1] RX480 GPUs don't go idle on waking up from S3 sleep

N

NomanA

Member
[Edited: The bug is now fixed in 16.9.1 driver. Card resumes from sleep correctly every time]

I picked up a custom RX480, and while it has performed pretty well, I have hit this annoying bug which is affecting pretty much all RX480 users - all who use S3 sleep on their PCs, regardless of their motherboard chipset.

Every time the PC wakes up, there's about 95% chance that the GPU will get stuck in state 7, with 100% activity and maximum GPU boost clocks. The card heats up a bit, and the fans have to run non-stop. No amount of coming in and out of 3D applications, can then bring the GPU out of state 7. You have to restart the PC from there, or try multiple sleep/wake cycles until the GPU goes idle. This affects reference and custom RX 480s alike, whether they are overclocked or not, or whether you have used Wattman or not to set up custom clocks and voltages.

It's a very annoying bug, especially for HTPC usages, where PCs often come in and out of sleep mode a lot.

The workarounds are all impractical.
1) Don't let PC go into sleep mode.
2) Let the PC hibernate instead of sleep. GPU clocks are fine then. However waking up from hibernate takes a lot longer.
3) Use Wattman to down-volt and down-clock state7, because that's where the GPU is stuck. However this won't work in games.

The bug has existed since RX 480 release, and has not yet been acknowledged, as far as I know. Hopefully with notebook GPU support, AMD will have to fix the bug, since S3 sleep is a common occurrence for such PCs.
 
Last edited:
AMD needs to obviously research the problem. I'd imagine they're aware of it if it's widespread. But who knows. Microsoft just killed millions of webcams somehow not realizing that it would happen upon release of the Anniversary Update even though thousands of people are in the insider program and were having problems with webcams. Nobody put 2 and 2 together.

I suspect a driver update will fix it eventually.
 
nerp said:
I suspect a driver update will fix it eventually.
Click to expand...

Yes, and I hope that the fix comes soon. At least the AMD issue-reporting form has a whole category about resuming PCs from sleep state.

There's a thread at Overclockers.uk mostly about similar issue, with some comments from an AMD representative.
https://forums.overclockers.co.uk/showthread.php?p=29944860

And a lot more active thread at AMD's community forums
https://community.amd.com/thread/204521

Few more things about this issue. While the GPU clock and activity almost always are at max, after resuming from sleep, the memory clock stays low more than two thirds of the time. Also the fan comes on full-blast for few seconds (5-10 sec) on some retries too.
 
My Sapphire Nitro RX470 8 GB doesn't have this issue and I've tried putting the computer to sleep multiple times. I even tried resetting to default settings. I'm running the 16.8.1 drivers, an i7-3770 processor, and B75 chipset on the motherboard.
 
Thanks for trying out the sleep mode on your RX 470 based setup. It's great to know, and gives me hope that the problem will eventually be fixed for RX 480s as well.
 
The issue is supposedly fixed in 16.9.1 according to the release notes. I'll try out the driver and update the original post.

Good job AMD!
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top