RX 580 and 5K displays, bandwidth validation failed whith multiple monitors

Gaël HERMET linuzel at gmail.com
Tue Jun 4 13:10:09 UTC 2019


Le 30 mai 2019 à 18:01, Harry Wentland <hwentlan at amd.com> a écrit :
> 
> On 2019-05-27 10:58 a.m., Gaël HERMET wrote:
>> Hi,
>> 
>> I have been facing an issue with my 5K display (iiyama ProLite
>> XB2779QQS-S1).
>> 
>> It works fine as long as it is the only active monitor, as soon as I
>> activate another monitor the main one (5k) can't display more than 4k.
>> 
>> Debug using "echo 0x4 > /sys/module/drm/parameters/debug" show this :
>> mai 23 09:01:22 bureau-gael /usr/lib/gdm3/gdm-x-session[3465]: (EE)
>> AMDGPU(0): failed to set mode: Invalid argument
>> mai 23 09:01:22 bureau-gael kernel: [drm:dce112_validate_bandwidth
>> [amdgpu]] dce112_validate_bandwidth: Bandwidth validation failed!
>> 
>> I disabled the check by forcing is_display_configuration_supported to
>> return true in dce_calcs.c and it works fine.
>> 
>> Anything I can do to correct this bandwidth calculation ?
>> 
> 
> The bandwidth formulas come from our HW teams and usually leave a good
> margin of error. Changing the formulas to allow for your case isn't a
> correction as it might cause issues in certain scenarious, i.e. there's
> no more guarantee that things work as expected.
> 
> Examples are 4k video playback (especially multiple videos), gaming.
> Anything that does a lot of GPU memory access.
> 
> Issues that might appear are underflow, i.e. white lines on the screen.
> In extreme cases underflow might even hang the entire display pipe.
> 
> Harry

Ok I understand.

Is there any reason why it works without any trouble or setting to mess with in both Windows and MacOS ?

This setup is not that rare, 5k iMacs with Radeon graphics have existed for years now and they all support external displays fine.

Thanks again,
Gaël.

> 
>> 
>> _______________________________________________
>> amd-gfx mailing list
>> amd-gfx at lists.freedesktop.org
>> https://lists.freedesktop.org/mailman/listinfo/amd-gfx



More information about the amd-gfx mailing list