dragonpilot

mirror of https://github.com/dragonpilot/dragonpilot.git synced 2026-03-03 03:43:58 +08:00

Files

George Hotz b6e355a933 modeld: PC Thneed prereqs (#25615 )

* pc thneed prereqs

* ugh, out of date

* that can stay private

* memcpy here is fine in SNPE variant

* release files

* thneed docs don't work anymore. they didn't look too useful

Co-authored-by: Comma Device <device@comma.ai>

2022-08-30 17:57:14 -07:00

include

thneed saves 45% of a core (#1512 )

2020-05-15 13:53:01 -07:00

kernels

minor cleanups, fix non binary compile (#23882 )

2022-03-01 16:05:44 -08:00

__init__.py

model: mse err from 0.02-> 0.000056 (#23891 )

2022-03-02 20:52:17 -08:00

compile.cc

Revert "Rocket league model (#24869 )" (#24936 )

2022-06-21 12:07:00 -07:00

lib.py

Mypy fixes for --check-untyped-defs (#24372 )

2022-05-12 14:08:11 +02:00

optimizer.cc

Falcon Punch Model: turn cutting improvements (#25413 )

2022-08-12 00:47:59 -07:00

README

thneed saves 45% of a core (#1512 )

2020-05-15 13:53:01 -07:00

serialize.cc

modeld: PC Thneed prereqs (#25615 )

2022-08-30 17:57:14 -07:00

thneed_common.cc

modeld: PC Thneed prereqs (#25615 )

2022-08-30 17:57:14 -07:00

thneed_qcom2.cc

modeld: PC Thneed prereqs (#25615 )

2022-08-30 17:57:14 -07:00

thneed.h

modeld: PC Thneed prereqs (#25615 )

2022-08-30 17:57:14 -07:00

weights_fixup.py

move models to selfdrive/modeld/ (#24539 )

2022-05-15 17:29:09 -07:00

README

thneed is an SNPE accelerator. I know SNPE is already an accelerator, but sometimes things need to go even faster..

It runs on the local device, and caches a single model run. Then it replays it, but fast.

thneed slices through abstraction layers like a fish.

You need a thneed.