<< Solved >> RX 560 very high fan speed after wake from sleep on 10.14.4 (solved by 10.14.5 beta 2)

[justinhe]

If you check Gigabyte Support BIOS files for Aorus Z370 Gaming 5, you can see F5 addressed GPU compatibility. For your ASRock B360-HDV I understand this may not have been a problem, although see the details for BIOS v1.20.

Thanks, but i'm using BIOS version 4.0 , released Mar 28, 2019. Still the same problem.

PWM is a good way to control CPU fans, using a 4-pin connector, however RX-series GPUs have a specific temperature-curve fan-speed response built-in to their VBIOSs. The FrameBuffer critical error you experienced might suggest checking for the latest Lilu/Whatevergreen kexts, if you use them (?) or maybe even an earlier one. The RX560 has high fan-speed during boot, which then drops quickly once the OS is running.

I always use the latest Lilu and whatevergreen. still the same problem:

2019-03-31 15:43:45.201548+0800 0x742 Default 0x0 0 0 kernel: (AMDFramebuffer) CRITICAL ERROR : VUPDATE interrupt has not been generated in time!

 

[UtterDisbelief]

Thanks, but i'm using BIOS version 4.0 , released Mar 28, 2019. Still the same problem.



I always use the latest Lilu and whatevergreen. still the same problem:

2019-03-31 15:43:45.201548+0800 0x742 Default 0x0 0 0 kernel: (AMDFramebuffer) CRITICAL ERROR : VUPDATE interrupt has not been generated in time!

Okay, understood.

Does this problem does affect real Macs? There are many with new AMD Radeon GPUs. If this was, as you say, 'absolutely a macOS problem' we should hope Apple had fixed it during the beta stages going back a long way.

 

[justinhe]

Okay, understood.

Does this problem does affect real Macs? There are many with new AMD Radeon GPUs. If this was, as you say, 'absolutely a macOS problem' we should hope Apple had fixed it during the beta stages going back a long way.

Do you have amd card as main gpu to connect to display?

Put your computer to sleep, wake up 1m later, run:

log show —last 5m | grep -i amdframebuffer

 

[UtterDisbelief]

Do you have amd card as main gpu to connect to display?

Put your computer to sleep, wake up 1m later, run:

log show —last 5m | grep -i amdframebuffer

1) Yes.

2) Okay.

3) Done.

What are we looking for here? I have run the report and there are no errors or crashes.

I have also run both the scripts in your first post - again no errors at all.

Please understand I fully accept that you have a serious problem here and would like to help if I can. I feel though that blaming a GPU (or the OS) that is well-known as reliable and very compatible (Apple uses the Radeon Pro 560 but it seems very close) and works well with macOS may be premature. From other posts in this thread it also seems to affect RX470 and RX570.

It will be great if we can isolate this problem and let people know of a firm, reliable solution

 

[justinhe]

1) Yes.

2) Okay.

3) Done.

What are we looking for here? I have run the report and there are no errors or crashes.

I also ran the script from your first post - again no error.

Please understand I fully accept that you have a serious problem here and would like to help if I can. I feel though that blaming a GPU that is well-known as reliable and very compatible (Apple uses the Radeon Pro 560 but it seems very close) may be premature. From other posts in this thread it also seems to affect RX470 and RX570.

It will be great if we can isolate this problem and let people know of a firm, reliable solution

Not just my RX 560 /RX 470, but also my friends' have the fan issue on 10.14.4 after wake from sleep

Note that make sure it goes to sleep fully, power LED turns off or blinking is a sign of full sleep.

I'm looking for attached reports:

kernel: (AMDFramebuffer) --> IOP: FB:0 has been busy for 121 cycles with 0x51 transactions, busy? 1.

and sometimes:

kernel: (AMDFramebuffer) CRITICAL ERROR : VUPDATE interrupt has not been generated in time!

and sometimes:

kernel: (AMDFramebuffer) CRITICAL ERROR : VBLANK interrupt has not been generated in time!

it is just i don't know if these errors are the cause of the fan issue. There is no information on the internet, VBLANK, VUPDATE is something too deep inside the GPU bios, i don't understand what they mean, but i see them in Linux kernel source.

Anyway, with the cable trick, i no longer have to worry about the fan speed it is always 820RPM , now i made it 1120 RPM (change fan control from 15% to 20% in BIOS settings), GPU is now 28˚ with web browsing and terminal coding

 

[UtterDisbelief]

Not just my RX 560 /RX 470, but also my friends' have the fan issue on 10.14.4 after wake from sleep

Note that make sure it goes to sleep fully, power LED turns off or blinking is a sign of full sleep.

I'm looking for attached reports:

kernel: (AMDFramebuffer) --> IOP: FB:0 has been busy for 121 cycles with 0x51 transactions, busy? 1.

and sometimes:

kernel: (AMDFramebuffer) CRITICAL ERROR : VUPDATE interrupt has not been generated in time!

and sometimes:

kernel: (AMDFramebuffer) CRITICAL ERROR : VBLANK interrupt has not been generated in time!

it is just i don't know if these errors are the cause of the fan issue. There is no information on the internet, VBLANK, VUPDATE is something too deep inside the GPU bios, i don't understand what they mean, but i see them in Linux kernel source.

Anyway, with the cable trick, i no longer have to worry about the fan speed it is always 820RPM , now i made it 1120 RPM (change fan control from 15% to 20% in BIOS settings), GPU is now 28˚ with web browsing and terminal coding

Well good news that you can keep the temperature down - even if it is unusual

From the earlier run (log show --last 20m | grep -i -E "amd|radeon|gfx0"), if this helps:

2019-03-31 12:20:56.614806+0100 0xfca Default 0x0 418 0 SafeEjectGPUService: (SafeEjectGPU) Device PreExisted [00000001000003df] AMD Radeon RX 560

(Some spaces removed for clarity)

Now, I have just run the latest command, as you show in your screenshot above (log show --last boot | grep -i -E "amdframe") ... Still no errors or "busy for" messages. Actually no further output of note.

So I am unable replicate the error here.

 

[justinhe]

Well good news that you can keep the temperature down - even if it is unusual

From the earlier run (log show --last 20m | grep -i -E "amd|radeon|gfx0"), if this helps:

2019-03-31 12:20:56.614806+0100 0xfca Default 0x0 418 0 SafeEjectGPUService: (SafeEjectGPU) Device PreExisted [00000001000003df] AMD Radeon RX 560

(Some spaces removed for clarity)

Now, I have just run the latest command, as you show in your screenshot above (log show --last boot | grep -i -E "amdframe") ... Still no errors or "busy for" messages. Actually no further output of note.

So I am unable replicate the error here.

The “SafeEjectGPUService” message is normal, so i did not even mention it. Thanks though.

What’s the model of your GPU card? It is amazing that you did not experience high fan speed ( speed above 1500RPM for RX ?70/?80/?90 is considered very high, 2500RPM for RX ?60 is considered very high due to worse cooling design).

 

[UtterDisbelief]

The “SafeEjectGPUService” message is normal, so i did not even mention it. Thanks though.

What’s the model of your GPU card? It is amazing that you did not experience high fan speed ( speed above 1500RPM for RX ?70/?80/?90 is considered very high, 2500RPM for RX ?60 is considered very high due to worse cooling design).

It's a Gigabyte RX560 Gaming OC 4GB.

The fans stay off until the card is stressed (not very often) after the initial test pulse during boot. Waking from sleep briefly spins the fans again but they turn off once more - until stressed.

 

[justinhe]

It's a Gigabyte RX560 Gaming OC 4GB.

The fans stay off until the card is stressed (not very often) after the initial test pulse during boot. Waking from sleep briefly spins the fans again but they turn off once more - until stressed.

Darn! I see why, your RX 560 is unlike other RX 560s, it has fan stop feature, cooling tube, low max fan speed settings, low TDP design, i would buy that one if i knew it earlier....

 

[macuserim]

i just connected my RX 470's fan connector to Chassis Fan 2, with "4pin GPU Fan to SYS_FAN Cable"

where did you find such a cable?