bridgman I was wondering what it would take to support the Vega M GPU (Polaris 22?) included in the new Kaby Lake G processors using the open-source AMDGPU drivers.
Are there any technical challenges that would make the task significantly harder than, for instance, supporting Raven Ridge APUs or a standalone Vega M ? Or is it just a question of time and limited developer resources ?
To me, it seems that the tricky part would be implementing the power sharing with the CPU, but I am mostly clueless about how this actually work so I might be entirely wrong...
Also, does anyone know whether the technical documentation has been released for this GPU, as it seems to be the case for other AMD GPUs ? Since it is essentially a semi-custom chip sold to Intel, I am not sure whether AMD's usual policy still applies here.
Just being curious
Are there any technical challenges that would make the task significantly harder than, for instance, supporting Raven Ridge APUs or a standalone Vega M ? Or is it just a question of time and limited developer resources ?
To me, it seems that the tricky part would be implementing the power sharing with the CPU, but I am mostly clueless about how this actually work so I might be entirely wrong...
Also, does anyone know whether the technical documentation has been released for this GPU, as it seems to be the case for other AMD GPUs ? Since it is essentially a semi-custom chip sold to Intel, I am not sure whether AMD's usual policy still applies here.
Just being curious
Comment