Getting rid of openpilot.common.numpy_fast #34368
Conversation
selfdrive/car/cruise.py
Outdated
| @@ -106,7 +106,7 @@ def _update_v_cruise_non_pcm(self, CS, enabled, is_metric): | |||
| if CS.gasPressed and button_type in (ButtonType.decelCruise, ButtonType.setCruise): | |||
| self.v_cruise_kph = max(self.v_cruise_kph, CS.vEgo * CV.MS_TO_KPH) | |||
|
|
|||
| self.v_cruise_kph = clip(round(self.v_cruise_kph, 1), V_CRUISE_MIN, V_CRUISE_MAX) | |||
| self.v_cruise_kph =clip(round(self.v_cruise_kph, 1), V_CRUISE_MIN, V_CRUISE_MAX) | |||
selfdrive/car/cruise.py
Outdated
| @@ -2,7 +2,7 @@ | |||
|
|
|||
| from cereal import car | |||
| from openpilot.common.conversions import Conversions as CV | |||
| from openpilot.common.numpy_fast import clip | |||
| from numpy import clip | |||
There was a problem hiding this comment.
let's stick to import numpy as np and use np.clip
There was a problem hiding this comment.
are you sure some source file like https://github.com/Sammohana/openpilot/tree/master/system/hardware/fan_controller.py are small and require only specific function. So I think it will only increase memory usage.
There was a problem hiding this comment.
yeah, let's keep it consistent
|
This broke model replay on master https://jenkins.comma.life/blue/organizations/jenkins/openpilot/detail/master/169/pipeline#step-338-log-31 |
|
Memory usage is up from 65 to 75, after fixing migration: https://jenkins.comma.life/blue/organizations/jenkins/openpilot/detail/migration-fix-long-plan-migration/1/pipeline |
|
Why does that matter for openpilot mem usage? |
|
oh nvm thats a different failure on my branch |
|
The memory increase is from the switch to tinygrad |
|
Fix for model_replay failures: #34383 |
Removing openpilot.common.numpy_fast as per #34331