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

[Lionel Landwerlin]

On 26/10/17 21:11, Chris Wilson wrote:
> Quoting Lionel Landwerlin (2017-10-26 18:13:13)
>> On 26/10/17 14:05, Chris Wilson wrote:
>>> Quoting Tvrtko Ursulin (2017-10-26 14:00:28)
>>>> 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.
>>> We don't even ship gputop or the perf generator in igt... Can we at
>>> least make noises towards owning that code...
>>> -Chris
>>>
>> It would be nice to have stuff in single repo but gputop has quite a few
>> dependencies that I'm not sure igt will want.
> Just add the dependencies as optional to the build scripts; we don't
> have to add them to the common libs or make them mandatory except if we
> want to build all the tools. (If they are that bad, just pull it in as a
> submodule.) But it would be nice to exploit the synergies between the
> existing tools and test/perf harnesses; even if long term ubergpuprofiler
> takes over as the defacto tool for all gpus.
>
> And we have to purge the current intel_gpu_top; killing machines since
> 2011.
> -Chris
>

It's just a massive amount of deps right now.
Including: node, emscripten, protobuf, a websocket lib, libuv and more.
I also had the good taste of pulling in some bits of mesa for the device 
descriptions (which contains bits that igt doesn't have).
It's hard to look forward to improve all of that when there are cool 
features to implement.

Completely agree on the synergy part. I'll try to work towards that.

-
Lionel