On 17/11/17 10:53, Chris Wilson wrote:
Quoting Lionel Landwerlin (2017-11-16 16:00:03)
With the introduction of asymetric slices in CNL, we cannot rely on
the previous SUBSLICE_MASK getparam. Here we introduce a more detailed
way of querying the Gen's GPU topology that doesn't aggregate numbers.
This is essential for monitoring parts of the GPU with the OA unit,
because signals need to be accounted properly based on whether part of
the GPU has been fused off. The current aggregated numbers like
EU_TOTAL do not gives us sufficient information.
Here is the sysfs layout on a Skylake GT4 :
/sys/devices/pci0000:00/0000:00:02.0/drm/card0/topology/
Ok, bikeshedding time!
We already use topology in conjunction with DP-MST, so at a toplevel
this would be confusing.
I would start with a gt/ dir for all of this info.
Is this subslicing only for the render unit; are all platforms going to
have the same fusing across all units? At the least, I thought we would
be able to configure the powergating of the different slices on the
different units. It seems a logical extension that fusing would be
similar.
-Chris
I just realized that 'enabled_mask' might not be future proof enough.
We might want to expose other fusing information in the slices/subslices...
I might rename that to enabled_slices/enabled_subslices/enabled_eus in a v3.
-
Lionel
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
