Updating all power profiles using the same scripts. Previously,
every power profile was generated using different algorithims.
b/62389228
Test: Run a normal android workload and verify the app blame
is reasonable
Change-Id: Idff05eb0fa776a3dae7bde2952ec80b031834d54
New hardware units have a mapping that's actually linear, so we assume
that the curve is linear here and have the kernel fix it up for us if it
needs to.
Bug: 63606090
Test: manual
Change-Id: I05c98115b20e0328654ca42ac58244ea81e19940
This reduces the minimum brightness to 0.009 nits from 0.09 nits
(on EVT1 and 2).
Bug: 63834045
Test: manual
Change-Id: I7d76eae6e76de5ed3a125f8f4b42d1bba2f57ac7
The control points from another device were directly copied over to
Taimen, but Taimen's display's display has a very different mapping
between backlight value and actual produced brightness. These new values
bring it inline without our ML learned curve compensating for this fact.
The one difference here is that Taimen can go brighter than the devices
the model was trained on so an additional point was introduced. Based on
visual inspection the curve looks almost exactly linear in the last few
points before it hits the ceiling, so the final point is just a linear
extrapolation from Marlin's final two control points to Taimen's maximum
brightness.
Note also that Taimen's display can't be set above 223 in normal use.
Every value about this is an "overdrive" setting and greatly increases
the risk of burn-in not to mention using significantly more power.
Bug: 62027049
Test: manual
Change-Id: Iad725edd7adbe4505fd13237a6cb3aa2efd9bbcf
Make EFFECT_CLICK 10ms and LONG_PRESS 12ms.
Bug: 62176703
Change-Id: Id676a6581e8beff9ad2d951b2d9f405d38d374b9
Signed-off-by: David Lin <dtwlin@google.com>
