Tweet
Can someone explain how Wayland would use two GPUs in the case of a four screen setup with each pair of GPUs connected to two screens. I'm using radeontop (GitHub - clbr/radeontop) to analyse the operation of each of the GPUs individually and (using glxgears as a test) I'm not seeing a correlation between which physical display glxgears is running in and the GPUs workload. For example, if I run glxgears in displays 1&2 which are connected to GPU1 I am seeing a significant increase in workload in the radeontop instance looking at GPU2 (displays 3&4). GPU1 is very cheap (Radeon X1300/X1550), GPU2 less so (HD4650). The work I'm doing is to analyse the operation of the HD4650 running legacy software. It's almost as if the GPU2 is making up for GPU1's lack of capability by running some of it's demand. Is this possible?