Commit Graph

163 Commits

Author SHA1 Message Date
ZwX1616 744d263288 compensate for tilt
old-commit-hash: 78cd373c4b
2021-01-14 21:23:52 -08:00
HaraldSchafer c6d9b9565a Mpc rework2 (#19660)
* start again

* need that too

* this actually works

* not needed

* do properly

* still works

* still works

* still good

* all G without ll

* still works

* all still good

* cleanup building

* cleanup sconscript

* new lane planner

* how on earth is this silent too....

* update

* add rotation radius

* update

* pathplanner first pass

* misc fixes

* fix

* need deep_interp

* local again

* fix

* fix test

* very old

* new replay

* interp properly

* correct length

* another horrible silent bug

* like master

* fix that

* do doubles

* different delay compensation

* make robust to empty msg

* make pass with hack for now

* add some extra

* update ref for increased leg

* test cpu usage on this pr

* tiny bit faster

* purge numpy

* update ref

* not needed

* ready for merge

* try again after recompile

Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
old-commit-hash: 158210cde8
2021-01-14 18:43:50 -08:00
Comma Device 25dbb18f73 fix rounding error
old-commit-hash: 451d4590cf
2021-01-14 21:59:26 +00:00
ZwX1616 fa97a4c7be fix tici dm
old-commit-hash: 9eb7688dfe
2021-01-12 21:17:54 -08:00
Dean Lee 99a831dd5c replace std::this_thread::sleep_for with util::sleep_fo (#19717)
old-commit-hash: 72bb55247c
2021-01-11 11:59:48 +01:00
Dean Lee 4c1919f41e merge utilpp.h into util.h (#19710)
old-commit-hash: 595830135b
2021-01-10 14:57:32 -08:00
Dean Lee 5561bb058d dmonitoringd: add helper function get_yuv_buf (#19590)
old-commit-hash: d39d737ae0
2021-01-08 15:30:06 -08:00
Willem Melching 3f950832b9 VisionIPC 2.0 (#19641)
* Squashed vipc

* Update release files

* Remove else

* add visionipc to release files

* use poller in vipc receive

* opencl framework instead of lib on macos

* Fix camera webcam

* Fix opencl on mac in ui

* more webcam fixes

* typo in ui sconsfile

* Use cur_yuv_buf

* visionbuf c++ class

* Camera qcom was still using visionbuf_allocate

* Turn loggerd back on

* fix snapshot

* No build needed

* update test camerad

* no more release callback

* make encoder c++

* Revert "no more release callback"

This reverts commit e5707b07002fee665d0483d90713154efc2d70d4.

* fix exit handlers

* No need to check errno

* move release callback call

* s/VIPCBufExtra/VisionIpcBufExtra/g

* use non blocking connect

* ui use non blocking connect

* Lower condition variable wait time

* Snapshot cleanup

* bump cereal

* bump cereal
old-commit-hash: fb496c692a
2021-01-08 14:54:25 +01:00
Dean Lee 5e557ed96d common exit handler (#19661)
* common signal handle

* fix typo

* rename to ExitHandler

* remove include signal.h

* remove include csignal

* rename variable exit to v

* rebase master
old-commit-hash: 86fe2d1697
2021-01-05 20:19:53 -08:00
Dean Lee b1b5baf5e9 shorten variable name (#19659)
old-commit-hash: 170ece1ddd
2021-01-05 12:57:31 +01:00
Dean Lee bbfb751d69 Replace usleep with util::sleep_for (#19643)
* replace usleep with sleep_for

* sleep microseconds

* sleep nanoseconds

* Trigger Build
old-commit-hash: c1cf0882bd
2021-01-04 18:03:11 +01:00
Adeeb Shihadeh 5bbe4f3d52 manager tests + make all processes exit cleanly (#19595)
* manager tests

* logcatd exits cleanly

* sigint

* boardd

* multiple dbus connections hangs for some reason

* clocksd proclogd

* network type from thermal

* fix tests

* fix android logcatd

* fix mac

* fix mac proclogd

* move on device athena tests

* build first

* build first

Co-authored-by: Comma Device <device@comma.ai>
old-commit-hash: ffa7e0cbdb
2020-12-29 22:32:03 -08:00
Adeeb Shihadeh 762eda95cd Set GPU priorities + improved modeld priorities (#2691)
* give gpu threads rt priority

* modeld gets highest prio on core 2

* fix frame stream

* lower prio of cal thread

Co-authored-by: Comma Device <device@comma.ai>
old-commit-hash: c57ee16e86
2020-12-15 19:22:03 -08:00
George Hotz 02a2f9ca15 Thneed refactors for future functions (#2673)
* delete debug

* thneed updates, but it seems slower

* thneed refactor

* refactor touchups

* add back asserts

* fix uaf

* track the size for local args

* final thneed refactor

* switch kgsl_command_object to avoid memory leak

* comments

* unused includes

Co-authored-by: Comma Device <device@comma.ai>
old-commit-hash: 5fdda8dbd8
2020-12-14 15:46:49 -08:00
Adeeb Shihadeh c09d84218d model timing benchmark script
old-commit-hash: 5ba7276118
2020-12-13 19:08:49 -08:00
Dean Lee 75e3d20dac convert loadyuv.c to loadyuv.cc (#2754)
old-commit-hash: d6eead25d0
2020-12-12 17:10:51 -08:00
Dean Lee 252b63ba15 transform.h: remove extern "C" (#2755)
old-commit-hash: 0fcfb07f1f
2020-12-12 00:00:07 -08:00
Dean Lee 9be81a6f99 clutil refactor (#2733)
* cleaup clutil.h

* clutil.c -> clutil.cc

* std::hash

* try simplify functions

* using namespace group help functions

* continue

* add function get_platform_info

* continue

* remove caching

* struct CLContext

* std::cout & CL_ERR_TO_STR

* remove CLU_NO_CACHE

* Revert "struct CLContext"

This reverts commit 882e413f227e4c1ba4bb83f956cc67739a189221.

* cl_get_error_string : remove undefied error codes

* replace all printf with std::cout

* util::read_file

* cleanup

* revert CL_ERR_TO_STR

* simplify

* add macro CL_ERR_TO_STR

* cl_get_error_string: fix undeclared errors

* log_size should not +1

* use std::stirng log

* apply review
old-commit-hash: 3c08d0f1f7
2020-12-11 13:05:32 +01:00
HaraldSchafer 89b03fa2ee Test model refactor (#2720)
* template funcion fill_meta

* add function get_plan_max_idx

* add function get_mdn_max_idx

* remove temp builder variables

* get_mdn_max_idx->get_lead_data

* get_pan_max_idx->get_plan_data

* fill_lane_line

* combine fill_lane_line&fill_path into one function

* fill_lead_vw

* using unique_ptr

* prefer using constexpr

* more constexpr,remove duplicate defined TRAJECTORY_SIZE

* remove suffix _arr from variable name

* misc

* remove extern c

* refactor model_publish

* remove unused paramaters

* traffic_convention to c style array

* c style array:prev_desire&pulse_desire

* fix error&make easy for review

* const mat3 &transform

* move cl_command_queue into ModelState

* use maco LEAD_MHP_SELECTION

* move constexpr from .h to .cc

* remove #define MODEL_NAME

* modeldata.h: contexpr

* remove param temporal from model_init

* helper function get_best_data

* fix probs

* int

Co-authored-by: deanlee <deanlee3@gmail.com>
old-commit-hash: 630c7309a5
2020-12-09 22:48:14 -08:00
Mitchell Goff 711d60239a Added model_replay and fixed some bugs in camera_replay (#2679)
* Added model_replay and fixed some bugs in camera_replay

* Unpack capnp logs on device

* add sync to device

* GPU now also works on PC

* update model ref

* update refs

* dont change this one

* Use pipeline calib instead of rlog calib

* remove that

* update refs

Co-authored-by: Willem Melching <willem.melching@gmail.com>
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
old-commit-hash: bbe9accd4d
2020-12-08 19:01:59 -08:00
Adeeb Shihadeh 8809fc68e7 log DSP execution time (#2723)
* log dsp execution time

* don't need that
old-commit-hash: b3fd759fe8
2020-12-08 17:28:47 -08:00
HoverB_SN_11 3ea014d35d too wide
old-commit-hash: 18c9e42520
2020-12-08 19:31:38 +00:00
Dean Lee fb9298b120 fix fille lead (#2703)
old-commit-hash: 531b85b983
2020-12-06 13:02:23 -08:00
ZwX1616 f4bdc2127c raw pred api (#2611)
* send both

* update cereal

* ptr type

* this is fine

* clean up

* v1 and v2

* header

* update cereal

* cereal
old-commit-hash: f30d15ea24
2020-12-02 18:46:12 -08:00
Dean Lee df9e0b2cb4 simplify building capnp messages with arrays (#2617)
old-commit-hash: bd4f6650fa
2020-11-30 12:08:52 +01:00
Dean Lee 57138a6e0f Error checking macros for opencl (#2615)
* macro CL_CHECK&CL_CHECK_ERR

* trigger checks
old-commit-hash: 42183d913f
2020-11-30 12:04:49 +01:00
George Hotz c22f0fc034 Make the DSP work everywhere (#2621)
* Use DSP again

* 20 Hz

Co-authored-by: Willem Melching <willem.melching@gmail.com>
old-commit-hash: bdb1b766d0
2020-11-30 11:21:06 +01:00
Willem Melching 19855c90f1 remove check on frame drop in modeld (#2638)
old-commit-hash: feaecbd615
2020-11-27 16:13:37 +01:00
Comma Device 2bc2c7c9b3 mask out the priority
old-commit-hash: 71930a4b96
2020-11-24 06:05:59 +00:00
Comma Device 8cb58c9d20 give the model GPU context highest priority
old-commit-hash: 5a8c90f32c
2020-11-24 05:57:34 +00:00
George Hotz 55df5b0ddf More governance work, fix thneed (#2610)
* more governance work

* fix thneed on qcom2

* waste doesn't OOM

Co-authored-by: Comma Device <device@comma.ai>
old-commit-hash: a14ce09018
2020-11-23 17:52:28 -08:00
George Hotz 7ce6b3e65e Prerotate dmonitoringmodeld input (#2606)
* not tested, but something like this

* more like this

* dl and ur may still be flipped

* they were flipped

* Revert "they were flipped"

This reverts commit fbd477bb4407c9e051fad7d6a3f0825ca3d23c11.

Co-authored-by: Comma Device <device@comma.ai>
old-commit-hash: 327aa123f4
2020-11-23 17:42:01 -08:00
Adeeb Shihadeh 4be6d905dc update frame drop filter (#2592)
old-commit-hash: d92f923d70
2020-11-21 14:50:53 -08:00
George Hotz 6b32f79ae6 New settings for governors, +30% memory bandwidth (#2590)
* new settings for governors, +30% memory bandwidth

* CPU waste is better

* warm up frame_drop

* optional CORE arg for rtshield

* back off priorities of processes not critical to driving

* if it doesn't drive, it's not realtime

* err, resource. this should have been included

* reverse order to not spike

Co-authored-by: Comma Device <device@comma.ai>
old-commit-hash: 917eff8c30
2020-11-20 19:25:12 -08:00
Adeeb Shihadeh 66aba30092 should be seconds
old-commit-hash: 85c9a71bc9
2020-11-20 14:32:22 -08:00
Adeeb Shihadeh 1a99d77746 log DM model execution time (#2589)
old-commit-hash: 580589b250
2020-11-20 13:38:31 -08:00
Comma Device 51711af7ff fix some lack of cache awareness in yuvframe2tensor
old-commit-hash: 826c21b045
2020-11-20 18:03:06 +00:00
Comma Device 8de3c8cf75 simpler repro for dmon lag issue
old-commit-hash: 06f6f245ff
2020-11-20 17:30:31 +00:00
George Hotz 599ee894cd other includes to build repro on pc
old-commit-hash: 6fc7c0c005
2020-11-20 09:09:43 -08:00
Comma Device 04e6e04bb4 simpler repro, no vector required
old-commit-hash: b9a8e7b209
2020-11-20 17:03:49 +00:00
Comma Device 6d764d4ade repros better with cache clears
old-commit-hash: 97b716512a
2020-11-20 16:50:25 +00:00
Comma Device 67d4a60288 reproduce dmon lag
old-commit-hash: 52673f0532
2020-11-20 16:39:12 +00:00
HaraldSchafer 7c8ea30532 fix frame drop percentage (#2578)
* correct, I think

* add model execution time

* remove unused

* token is a file

* this is gonna change

Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
old-commit-hash: 7378bc4284
2020-11-19 21:05:29 -08:00
HaraldSchafer faad598e29 Minor fixes 08 (#2565)
* correct indexes

* best metrics yet 30231f46-08af-477e-9d30-776593913c24/700

* sometimes x decreases

* update model refs

Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
old-commit-hash: 7b216e7cbb
2020-11-17 20:04:32 -08:00
HaraldSchafer 008f37c749 Torch model (#2452)
* refactor draw model

* rebase master

* correct valid_len

* rename function

* rename variables

* white space

* rebase to master

* e16c13ac-927d-455e-ae0a-81b482a2c787

* start rewriting

* save proress

* compiles!

* oops

* many fixes

* seems to work

* fix desires

* finally cleaned

* wrong std for ll

* dont pulse none

* compiles!

* ready to test

* WIP does not compile

* compiles

* various fixes

* does something!

* full 3d

* not needed

* draw up to 100m

* fix segfault

* wrong sign

* fix flicker

* add road edges

* finish v2 packet

* Added pytorch supercombo

* fix rebase

* no more keras

* Hacky solution to the NCHW/NHWC incompatibility between SNPE and our frame data

* dont break dmonitoringd, final model 229e3ce1-7259-412b-85e6-cc646d70f1d8/430

* fix hack

* Revert "fix hack"

This reverts commit 5550fc01a7881d065a5eddbbb42dac55ef7ec36c.

* Removed axis permutation hack

* Folded padding layers into conv layers

* Removed the last pad layer from the dlc

* Revert "Removed the last pad layer from the dlc"

This reverts commit b85f24b9e1d04abf64e85901a7ff49e00d82020a.

* Revert "Folded padding layers into conv layers"

This reverts commit b8d1773e4e76dea481acebbfad6a6235fbb58463.

* vision model: 5034ac8b-5703-4a49-948b-11c064d10880/780  temporal model: 229e3ce1-7259-412b-85e6-cc646d70f1d8/430  with permute + pool opt

* fix ui drawing with clips

* ./compile_torch.py 5034ac8b-5703-4a49-948b-11c064d10880/780 dfcd2375-81d8-49df-95bf-1d2d6ad86010/450 with variable history length

* std::clamp

* not sure how this compiled before

* 2895ace6-a296-47ac-86e6-17ea800a74e5/550

* db090195-8810-42de-ab38-bb835d775d87/601

* 5m is very little

* onnx runner

* add onnxruntime to pipfile

* run in real time without using the whole CPU

* bump cereal;

* add stds

* set road edge opacity based on stddev

* don't access the model packet in paint

* convert mat.h to a c++ header file (#2499)

* update tests

* safety first

Co-authored-by: deanlee <deanlee3@gmail.com>
Co-authored-by: mitchell <mitchell@comma.ai>
Co-authored-by: Comma Device <device@comma.ai>
Co-authored-by: George Hotz <george@comma.ai>
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
old-commit-hash: 08846b5c0e
2020-11-11 20:31:46 -08:00
Willem Melching fc0b6f099f Revert "convert mat.h to a c++ header file (#2499)"
This reverts commit 607866bb60ba1c69c6f777082c5ea787e895a251.

old-commit-hash: f2c8a6122d
2020-11-09 01:41:36 -08:00
Dean Lee 13a25b9e27 convert mat.h to a c++ header file (#2499)
old-commit-hash: 519c043a63
2020-11-09 01:19:09 -08:00
ZwX1616 4dc3fb994e generalize camera assumptions (#2423)
* clean up 1

* complain

* cleanup

* make models api generic

* clean up 2

* cleanup

* remove unused
old-commit-hash: c0946c8bdb
2020-11-05 13:22:28 -08:00
Willem Melching 651d4ce031 --std=c++17 (#2330)
* switch to c++17

* more 17

* 1z

* use std::clamp

* Fix types in clamp

* bump submodules

* bump laika

Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
old-commit-hash: 339d1967cb
2020-10-17 13:33:12 -07:00
ZwX1616 dae628002f Clean up and test camerad refactor (#2310)
* refactor

add function get_camerabuf_by_type

remove cl_free

use camera_id

move front stuff into frontview_thread

bigger refactor

more

typo

rename camerabuf *c to *b

cleanup header files

remove wideview_thread

camera_qcom.c to c++

fix SConscript

camera_process_buf

space

sendrgb same with before

do autoexposure in processing_thread

correct sendrgb

add camera_common.cc

move yuv_transform to CameraBuf

* use C to keep consitent

* rebase master

* rebase master

* continue refactor

* refactor all camera related data&function out of main

 done

* cleanup

* -use buf.rgb_width&height

* remoe ,

* rebase master

* remove unused std::vector

* add func common_camera_process_front

* fix max

* fix sendrgb

* less crap being logged

* fix indents

* add back thumbnails

* thumbnail fixed on tici

* junk junk junk

* unify api

* fix ptr ref

* lol!

* cleanup

* more simplified

* typos

* remove terminate

* use standard polling

* ops_thread refactor, too much code!

* fix build

* ok

* assert

* listen

* incl more

* wq!

* fix more build

* macos

* fix exp targets

* front is different

* const

* bye

* remath

* remove irrelevant

* do ops in a thread

* fix init

* return null

* QCOM2 ops

* qcom2 build

* missing comma

Co-authored-by: deanlee <deanlee3@gmail.com>
Co-authored-by: Comma Device <device@comma.ai>
old-commit-hash: 6b020241c9
2020-10-16 21:58:05 -07:00