[UMR] Add GPU_POWER (v3) + refactored cleanup
kai at dev.carbon-project.org
Sat Feb 11 19:18:36 UTC 2017
Tom St Denis wrote on 11.02.2017 20:05:
> On 02/11/2017 01:57 PM, Kai Wasserbäch wrote:
>> Tom St Denis wrote on 11.02.2017 18:52:
>>> I split out the deletion of the unused bits in --top.
>> thank you so much for bearing with me (and Edward, though I can't speak for him,
>> obviously)! You can still have my R-b. ;-)
> No worries. It's no secret I'm no fan of NFC re-submits but hard to get upset
> in the face of genuinely good feedback. And it is a community project in the
> long run so it's not like community pushback/feedback is not welcomed (my
> rumblings notwithstanding).
I'm not sure what AMD's internal policy might be on this, but NFC re-submits are
usually covered by the "with this/these fixed, you can have my R-b". Which is
the reason I had that kind of language in there. I didn't expect a resend to the
list, to be honest. (And I would guess the same applied to Edward's R-b.)
> Thanks to you and Edward for the feedback.
> I've pushed them out. You'll need a staging kernel to use GPU_POWER (with my
> and Eric's patches) to see any of this on a VI dGPU.
> It's kinda interesting to see the power usage graphs (using umr's --top 'logger'
> mode) comparing Polaris10 and fiji while running say unigine-heaven. I think
> the sensors will help debug/optimize power usage going forward.
Oh, should there have been a DRM version check in there then? Like Mesa guards
In any case I will have to wait for the Vega-upgrade I'm planning for this year
to actually test these features out.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 659 bytes
Desc: OpenPGP digital signature
More information about the amd-gfx