* 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>
* 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 fbd477bb44.
Co-authored-by: Comma Device <device@comma.ai>
* 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>
* 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>
* 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>
* 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>
* 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>
* always c++
* Create C++ params class
* get works
* tests hang now
* passes tests
* cleanup string formatting
* handle interrupt in blocking read
* fix memory leak
* remove unused constructor
* Use delete_db_value directly
* Rename put -> write_db_value
* filename cleanup
* no semicolons in cython
* Update common/SConscript
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
* add std::string version of delete_db_value
* This is handled
* cleanup encoding
* Add clear method to clear all
* add persistent params
* fix android build
* Should be called clear_all
* only import params when needed
* set params path on manager import
* recusrively create directories
* Fix function order
* cleanup mkdirp
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
Co-authored-by: Comma Device <device@comma.ai>
* protect highlights
* fix modeld scale factor
* rewrite logic for accurate and smooth AE
* fix condition
* tune speed
* move to qcom2
* not exactly 2x
* avoid hunting due to ettr policy
* tune to new curve
* gain kickdowns
* log params
* 4x legroom
* no limit needed here
* simple lowpass
* tweak
* pin to a remote core
* speedup AE and lower night exp
* this curve makes more sense
* tuning from fleet data
* fix ui
* only qcom2
Co-authored-by: Comma Device <device@comma.ai>
* fix setting core affinity
* dmonitoringd doesn't need rt priority
* android only gets two cores
* model and planner get core 2
* log missed cycle count
* neos update
* prod image
* revert NEOS changes
* still need this
* equalize performance on three threads
* need to request first
* clean up indexing
* obsolete
* handling skips and drops
* clean up repeated chunks
* use main frame id
* oneshot
* real skip check
* move to test
* should be cropped H here
Co-authored-by: Comma Device <device@comma.ai>
* use custom MessageBuilder from cereal
* use MesageBuilder.toBytes
* bump cereal
* unused
* another one
* fix ui msg
Co-authored-by: deanlee <deanlee3@gmail.com>
* use cpu for now on tici
* pin to core 5
* fix driver view
* probably wrong
* correct this time
* RHD too
Co-authored-by: Comma Device <device@comma.ai>
* dockerize carla + openpilot
* separate dockerfile
* bring back CI dockerfile
* cleanup bridge
* add op docker build and start script
* build container in CI
* fix camerad hack
* remove most magic numbers from bridge.py
* openpilot-sim docker build and run scripts
* fix dmonitoring hacks
* revert controlsd hacks
* clean up build scripts
* singular
* fix path
* fix image name
* modify sim readme
* sim readme and start script changes
* dockerfile with working opengl
* working opengl + passing panda build_st in docker
* fix bug in sim docker file
* bugfix sim docker file
* bugfix all op-sim docker issues
* modify readme + run script
* IT DRIVES
* clean this up
* more cleanup
* cleanup docker stuff
* more cleanup
* start with openpilot-base
* install carla python package
* Script is not in lib
* chmod
* everything should be running in docker now, the code may not be nice though
* works locally...
* rhdChecked is deprecated
* Checkout using git lfs when building sim container
* try to pass the tests
* pull latest docker
* gps should not throw an error on openpilot launch in bridge.py
* fixed a coding style error
* Only start ubloxd in car
* fixed more style problems
* revert typo
* Use enviromental variable to prevent errors in a simulator
* Remove unused import
* Attempt to fix missing enviromental variable
* fix typo
* less work for users, auto tmux engagement
* less work for users, auto tmux engagement
* fix check for nvidia
* clean up nvidia check
* remove typo, shorted dockerfile
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
Co-authored-by: Willem Melching <willem.melching@gmail.com>
Co-authored-by: Bruce Wayne <batman@workstation-eu-gregor.eu.local>
Co-authored-by: Gregor Kikelj <gregor1234567890@gmail.com>
* read rhd from param
* don't need that anymore
* fix build
* bump cereal
* read param from camerad too
* add read_db_bool helper
* bump model replay ref commit, IsRHD wasn't set
* remove read_model
* remove structs for c-capnp
* remove duplicate #define from modeld
* add function fill_path_points
* fix Indentation
* use MODEL_PATH_DISTANCE instead of 192
* fix type
use validLen
* rename left_path_points&right_path_points to xxx_lane_points
* Revert "Fix modeld dropping frames if processing takes over 50 ms"
This reverts commit a8e7a187ca.
* track frame drop
* fix unused
* alert on frame drop perc
* reduce thresholds
* posenet is invalid on non-consecutive frames
* Refactor filter time constant
* Use vipc_dropped_frames for posenet valid check
* Actually set frame drop percentage
* Add explicit cast just to be sure
* Prevent frames dropped to go up to quickly on startup
* bump cereal
* reduce soft disable threshold to 1%
Co-authored-by: Willem Melching <willem.melching@gmail.com>
* enable Wunused, first pass
* unused stuff in snpe model
* these are used on phone
* handle sigint and sigterm in modeld
* fix phone build
* camera qcom
* QCOM build works
* delete unused camerad vars
Co-authored-by: Comma Device <device@comma.ai>
* fix mac nui
* fix cereal
* builds on mac without warnings
* ui swap shouldn't be instant on pc
* camerad builds on mac
* modeld, and now everything, builds on mac
* ui draws pictures on mac now
* fix camerad shutdown
* support ui looping on pc
* python3 on that
* SCNu64 should be cross platform
* bump cereal and new apk
* clean up
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
* Running pre-commit in CI
* fix dockerfile syntax
* dont run on submodule repo folders
* Fix some import errors in ci
* more stuff
* That should be the last one
* remove the clCreateProgramWithSource interceptor
* that's old code, thneed is better
* label them thneed_, we shouldn't need to touch CL for anything not SNPE related
* add PubMaster & SubMaster
remove 'delete msg'
remove headers
* use constructor to initial all submster
* modify drain sockets
* fix typo in ssconscript.remove lines
no checkValid in loggerd
last modify
handle_message:event->&event
fix type
remove heads
event to auto
* new interface
* api changed
* Revert "use constructor to initial all submster"
This reverts commit 73be7ea46250a325ce41d3a0445e34395a2ae692.
* change to new api
* revert loggerd
* dd
* use new PubSub api
* update to new interface
don't modify loggerd
reset panda
reset opendbc
remove empty lines
* switch to new pubMaster
* update to the new inteface
change
remove error code
. to ->
merge paramsd.cc
update panda
fix typo
simplify
fix typo
* Fix build
* always conflate
Co-authored-by: deanlee <deanlee3@gmail.com>
* bigmodel
* more debug print
* debugging bigmodel
* remove the tanh, debugging
* print images/buffers
* disassemble the command queues
* decompiler
* dump the shaders
* full disasm
* support patching kernel and fixing convolution_horizontal_reduced_reads_1x1
* microbenchmark
* 42 GFLOPS, 1 GB/s
* gemm benchmark
* 75 GFLOPS vs 42 GFLOPS
* 115 GFLOPS
* oops, never mind
* gemm image is slow
* this is pretty hopeless
* gemm image gets 62 GFLOPS
* this is addictive and still a waste of time
* cleanup cleanup
* that hook was dumb
* tabbing
* more tabbing
Co-authored-by: Comma Device <device@comma.ai>
* thneed runs the model
* thneed is doing the hooking
* set kernel args
* thneeding the bufferS
* print the images well
* thneeds with better buffers
* includes
* disasm adreno
* parse packets
* disasm works
* disasm better
* more thneeding
* much thneeding
* much more thneeding
* thneed works i think
* thneed is patient
* thneed works
* 7.7%
* gpuobj sync
* yay, it mallocs now
* cleaning it up, Thneed
* sync objs and set power
* thneed needs inputs and outputs
* thneed in modeld
* special modeld runs
* can't thneed the DSP
* test is weird
* thneed modeld uses 6.4% CPU
* add thneed to release
* move to debug
* delete some junk from the pr
* always track the timestamp
* timestamp hacks in thneed
* create a new command queue
* fix timestamp
* pretty much back to what we had, you can't use SNPE with thneed
* improve thneed test
* disable save log
Co-authored-by: Comma Device <device@comma.ai>
* convert c-capnp to c++
change paramater type
* remove c-capnp header from commonmodel.c
* unsigned short -> uint16_t
* MSM_CAMERA_I2C_ADDR_TYPE_MAX->4
* lgtm in readme, and mac nui fixes
* call super inits in radars
* unused imports, dup lines
* more radars, more unused imports
* pass CP into RadarInterfaceBase
* more fixups
* unused imports
* delete unused lines
* ugh, new unused import
Co-authored-by: George Hotz <geohot@gmail.com>
* add traffic convention
* hope this work
* no comment
* latest and gratest
* big gru model
* 1af55c7d-ee15-414a-9e98-a0cb08c3441f/75
* much later in training
* wrong temporal size
* converged
* fix lane changes
* capnparm
* building in progress
* scons build works
* that script fixes opencl
* start new camera code
* includes and more camera scripts
* control c works now
* no device control yet
* phy too
* just one camera for now
* fix capnparm
* hmm, the inits are needed
* more cameras
* link stop start
* doesn't work yet
* fix ion on qcom2
* start poll ish
* 4 pictures and done
* no jpeg
* it works to picture
* destroy sync obj
* both work for now
* defined QCOM2
* fix fd leak
* run modeld
* 10 bit mode
* real frame stride
* needs digital gain
* dnew
* no color correcting on new
* that snpe doesn't work
* qcom2 gate
* cleanups
* oops, fix aarch64 detector
* update cereal
* modeld works with SNPE
* fix driver monitoring model
Co-authored-by: Tici <robbe@comma.ai>
* snpe version 1.35.0.698
* fix files common for new snpe
* efficient net with normal names
* compiling with 1.32 works
* actual snpe files
* fix monitoring, delete unneeded
* point the symlinks to the right places
* err, modeld needs the CPU
* benchmark for 50 its
* put back old model for now
Co-authored-by: Comma Device <device@comma.ai>
Co-authored-by: Harald Schafer <harald.the.engineer@gmail.com>
Co-authored-by: George Hotz <geohot@gmail.com>
* cleanup simulator files
* minor updates
* update readme
* keras runner builds
* hmm, still doesn't work
* keras runner works
* should work with python3 keras mod
* touchups