[Intel-gfx] [RFC 00/17] Per-context and per-client engine busyness

[Tvrtko Ursulin]

On 26/10/2017 10:50, Lionel Landwerlin wrote:
> On 26/10/17 08:34, Tvrtko Ursulin wrote:
>> On 25/10/2017 18:38, Chris Wilson wrote:
>>> Quoting Chris Wilson (2017-10-25 16:47:13)
>>>> Quoting Tvrtko Ursulin (2017-10-25 16:36:15)
>>>>> From: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
>>>> I've prototyped a quick demo of intel-client-top which produces 
>>>> output like:
>>>>
>>>>       neverball[  6011]:  rcs0:  41.01%  bcs0:   0.00%  vcs0:   
>>>> 0.00%  vecs0:   0.00%
>>>>            Xorg[  5664]:  rcs0:  31.16%  bcs0:   0.00%  vcs0:   
>>>> 0.00%  vecs0:   0.00%
>>>>           xfwm4[  5727]:  rcs0:   0.00%  bcs0:   0.00%  vcs0:   
>>>> 0.00%  vecs0:   0.00%
>>> +1
>>> +2 for a graph ;)
>> Where are those placement students when you need them! :)
> 
> I won't be your student, but I could like to wire this into gputop.

I was thinking gputop as well but did find the time to look at it yet.

If you want to play with it, I have uploaded my kernel and igt branches 
which implement this to below locations*.

But be aware the sysfs interface is at the moment a prototype and we 
haven't decided we are going with it 100%. So don't invest too much in 
any experiments you might decide to do. You'll see in the 
intel_client_top demo that I have also not invested a huge amount to 
make it nice and polished (and bug free!).

*)
https://cgit.freedesktop.org/~tursulin/drm-intel/log/?h=context-stats
https://cgit.freedesktop.org/~tursulin/intel-gpu-tools/log/?h=context-stats

Regards,

Tvrtko