Skip to content

Subaru: support BRZ 6MT 2024#579

Open
sshane wants to merge 338 commits intomasterfrom
brzpilot
Open

Subaru: support BRZ 6MT 2024#579
sshane wants to merge 338 commits intomasterfrom
brzpilot

Conversation

@sshane
Copy link
Owner

@sshane sshane commented Jan 23, 2026

No description provided.

adeebshihadeh and others added 30 commits February 7, 2026 21:00
…37115)

* Revert "revert tg calib and opencl cleanup (commaai#37113)"

This reverts commit 51312af.

* power draw is a lil higher

* just don't miss a cycle

* fix warp targets

* fix tinygrad dep
* Build vendored artifacts in CI

* parallel

* deterministic

* fix up

* fix gitignores

* lil more

* lil more consistency
* file chunking

* try this

* more cleanup

* cleaner
* from da bounce

* try this

* you can get it to clean up wow
Hi! The point of this pr is to make the model run easier to read. On the latest tinygrad numpy().flatten() empirically does the same thing as the internal contiguous().realize().uop.base.buffer.numpy(). numpy() is also documented (docstrings), which can assist new contributors in learning what each potential execution does. Torq_boi or yassine, I know you want proof in the code base, so here it is. As of tinygrad commit 2f55005:

 in tinygrad_repo/tinygrad/tensor.py
Lines 316-318 (def _buffer): ensure the tenso is contiguous() and realized() before accessing the raw buffer.
Line 378 (def numpy): Wraps the buffer access and adds a reshape to match the tensor shape.
self._buffer() is what executes contiguous().realize() and returns the buffer object.
Calling numpy() on that buffer object returns a 1D array (defined in tinygrad/device.py:193 via np.frombuffer).
The reshape(self.shape) at the end of Tensor.numpy() then adds dimensions to that 1D array. The added .flatten() removes those dimensions, flattening it back to a 1D array. Effectively the same as what is currently done, but less complex.
* rm upload

* use ci-artifacts

* sanitize

* rm ref_commit

* add ci

* handle exept

* bootstrap

* always

* fix

* replay

* keep ref_commit fork compatibility

* remove upload-only

* apply comments

* safe diffs in master

* Revert "safe diffs in master"

This reverts commit 369fcca.

* continue on master diff

* Update .github/workflows/tests.yaml

Co-authored-by: Shane Smiskol <shane@smiskol.com>

---------

Co-authored-by: Shane Smiskol <shane@smiskol.com>
v3 renamed inputs from kebab-case to snake_case (repo-token -> repo_token,
pr-message -> pr_message). The old names were silently ignored, causing
"Input required and not supplied: issue_message" errors.
The pinned SHA was v6.0.4, which is incompatible with actions/checkout@v6
and causes a "Duplicate header: Authorization" 400 error during git
remote operations. See peter-evans/create-pull-request#4272.
Revert "DM: Ford GT model (commaai#37013)"

This reverts commit 1459d35.
* rm side button

* fix

* fix
* correct from bottom alignment

* temp

* fix scale animation w/ btn_y

* home settings are always 64

* cleanup

* some clean up

* make 23 const

* rev

* more
* remove azure deps

* fix lint

* restore scripts
* Update Python packages

* fix

* bump panda

* revert tinygrad

---------

Co-authored-by: Vehicle Researcher <user@comma.ai>
Co-authored-by: Adeeb Shihadeh <adeebshihadeh@gmail.com>
sshane and others added 30 commits February 24, 2026 00:50
* rename

* make tuple

* blocking
* meh

* hmm

* can also do this

* keep behavior

* rm
* Give tf flags to onnx parse

* rm onnx again

* update lock
* one time test

* fix!

* cleanm up

* cleanm up
* should fail

* this works but i think i know a better way

* something like this

* hmm

* this works

* rm useless test

* good stash

* Revert "good stash"

This reverts commit c2dddf0.
* test wifi state machine

* clean up and another few tests

* no unittest :((

* clean up

* clean up

* try to repro on device

* try to repro on device

* nice, the flicker is covered by test_user_initiated_skips_dbus_lookup!

* add todo soon to be all fixed

* documentaiton

* test the thread races too

* _fire -> fire

* duplication

* new state

* fix some tests

* format

* combine similar tests

* use process_callbacks

* clean up

* collapse two tests

* rm nl

* previous messy test

* delete old

* asked another to ask questions
…ommaai#37397)

* fix a few

* document

* now remove unused prev_ssid

* more
* update docs

* Update paths for brand-specific safety files
* hmm

* revert to master

* context too big

* fresh context

* early return

* early return

* tests

* restore cmts

* lester nester

* note

* add

* final review

* cmt
fix connect/activate failure resetting connected/connecting ssid
* interesting epoch approach

* repro

* determ fix

* cmts

* new issue

* test

* clean up

* cmt

* add back

* reorg cmt

* cmt

* clean up

* cmt
* failing test

* fix

* rename

* better
* real traces for some tests

combine and new test for low strength/turn off hotspot while connecting

revert wifiui

* stupid llm

* clean up
* try this

* deactivate

* faiilures!

* starting

* try

* ...

* starting

* fix strength

* revert

* debug

* more

* override for display network

* try

* nvm it fixes a few things

* cmt

* clean up
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.