Tools: Update setup command for macOS native setup

.gitattributes

@@ -9,7 +9,6 @@
 *.ttf filter=lfs diff=lfs merge=lfs -text
 *.wav filter=lfs diff=lfs merge=lfs -text
 
-selfdrive/test/process_replay/fakedata/*.zst filter=lfs diff=lfs merge=lfs -text
 selfdrive/car/tests/test_models_segs.txt filter=lfs diff=lfs merge=lfs -text
 system/hardware/tici/updater filter=lfs diff=lfs merge=lfs -text
 selfdrive/ui/qt/spinner_larch64 filter=lfs diff=lfs merge=lfs -text

.github/workflows/auto_pr_review.yaml

@@ -24,7 +24,7 @@ jobs:
     # Check PR target branch
     - name: check branch
       uses: Vankka/pr-target-branch-action@def32ec9d93514138d6ac0132ee62e120a72aed5
-      if: github.repository == 'sunnypilot/sunnypilot'
+      if: github.repository == 'commaai/openpilot'
       env:
         GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
       with:
@@ -37,17 +37,17 @@ jobs:
     # Welcome comment
     - name: "First timers PR"
       uses: actions/first-interaction@v1
-      if: github.event.pull_request.head.repo.full_name != 'sunnypilot/sunnypilot'
+      if: github.event.pull_request.head.repo.full_name != 'commaai/openpilot'
       with:
         repo-token: ${{ secrets.GITHUB_TOKEN }}
         pr-message: |
             <!-- _(run_id **${{ github.run_id }}**)_ -->
             Thanks for contributing to openpilot! In order for us to review your PR as quickly as possible, check the following:
             * Convert your PR to a draft unless it's ready to review
-            * Read the [contributing docs](https://github.com/sunnypilot/sunnypilot/blob/master/docs/CONTRIBUTING.md)
+            * Read the [contributing docs](https://github.com/commaai/openpilot/blob/master/docs/CONTRIBUTING.md)
             * Before marking as "ready for review", ensure:
               * the goal is clearly stated in the description
               * all the tests are passing
-              * the change is [something we merge](https://github.com/sunnypilot/sunnypilot/blob/master/docs/CONTRIBUTING.md#what-gets-merged)
+              * the change is [something we merge](https://github.com/commaai/openpilot/blob/master/docs/CONTRIBUTING.md#what-gets-merged)
               * include a route or your device' dongle ID if relevant
-
+      

.github/workflows/badges.yaml

@@ -13,7 +13,7 @@ jobs:
   badges:
     name: create badges
     runs-on: ubuntu-latest
-    if: github.repository == 'sunnypilot/sunnypilot'
+    if: github.repository == 'commaai/openpilot'
     permissions:
       contents: write
     steps:
@@ -29,7 +29,7 @@ jobs:
 
         git checkout --orphan badges
         git rm -rf --cached .
-        git config user.email "badge-researcher@sunnypilot.ai"
+        git config user.email "badge-researcher@comma.ai"
         git config user.name "Badge Researcher"
 
         git add translation_badge.svg

.github/workflows/ci_weekly_report.yaml

@@ -15,7 +15,7 @@ env:
 
 jobs:
   setup:
-    if: github.repository == 'sunnypilot/sunnypilot'
+    if: github.repository == 'commaai/openpilot'
     runs-on: ubuntu-latest
     outputs:
       ci_runs: ${{ steps.ci_runs_setup.outputs.matrix }}
@@ -31,7 +31,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix: ${{fromJSON(needs.setup.outputs.ci_runs)}}
-    uses: sunnypilot/sunnypilot/.github/workflows/ci_weekly_run.yaml@master
+    uses: commaai/openpilot/.github/workflows/ci_weekly_run.yaml@master
     with:
       run_number: ${{ matrix.run_number }}
 

.github/workflows/ci_weekly_run.yaml

@@ -12,6 +12,6 @@ concurrency:
 
 jobs:
   selfdrive_tests:
-    uses: sunnypilot/sunnypilot/.github/workflows/selfdrive_tests.yaml@master
+    uses: commaai/openpilot/.github/workflows/selfdrive_tests.yaml@master
     with:
       run_number: ${{ inputs.run_number }}

.github/workflows/compile-openpilot/action.yaml

@@ -15,7 +15,7 @@ runs:
                           scons -j$(nproc) --cache-populate"
     - name: Save scons cache
       uses: actions/cache/save@v4
-      if: (github.ref == 'refs/heads/master' || github.ref == 'refs/heads/master-new')
+      if: github.ref == 'refs/heads/master'
       with:
         path: .ci_cache/scons_cache
         key: scons-${{ runner.arch }}-${{ env.CACHE_COMMIT_DATE }}-${{ github.sha }}

.github/workflows/docs.yaml

@@ -18,7 +18,7 @@ concurrency:
 jobs:
   docs:
     name: build docs
-    runs-on: ubuntu-24.04
+    runs-on: ubuntu-latest
     steps:
     - uses: commaai/timeout@v1
 
@@ -35,13 +35,13 @@ jobs:
 
     # Push to docs.comma.ai
     - uses: actions/checkout@v4
-      if: github.ref == 'refs/heads/master' && github.repository == 'sunnypilot/sunnypilot'
+      if: github.ref == 'refs/heads/master' && github.repository == 'commaai/openpilot'
       with:
         path: openpilot-docs
         ssh-key: ${{ secrets.OPENPILOT_DOCS_KEY }}
-        repository: sunnypilot/sunnypilot-docs
+        repository: commaai/openpilot-docs
     - name: Push
-      if: github.ref == 'refs/heads/master' && github.repository == 'sunnypilot/sunnypilot'
+      if: github.ref == 'refs/heads/master' && github.repository == 'commaai/openpilot'
       run: |
         set -x
 

.github/workflows/lfs-maintenance.yaml

@@ -1,72 +0,0 @@
-name: Sync comma's LFS
-
-env:
-  LFS_URL: 'https://gitlab.com/sunnypilot/public/sunnypilot-new-lfs.git/info/lfs'
-  LFS_PUSH_URL: 'ssh://git@gitlab.com/sunnypilot/public/sunnypilot-new-lfs.git'
-
-on:
-  schedule:
-    - cron: '0 0 * * *'  # Runs at 00:00 UTC every day
-  push:
-    branches:
-      - 'master-new'
-  pull_request:
-    branches:
-      - 'master-new'
-  workflow_dispatch:  # enables manual triggering
-    inputs:
-      upstream_branch:
-        default: 'master'
-        type: string
-
-jobs:
-  sync:
-    runs-on: ubuntu-latest
-    # Skip if PR is in draft mode
-    if: github.event_name != 'pull_request' || (github.event_name == 'pull_request' && github.event.pull_request.draft == false)
-    steps:
-      - name: Checkout Repository
-        uses: actions/checkout@v4
-        with:
-          repository: 'commaai/openpilot'
-          ref: ${{ inputs.upstream_branch }}
-      
-      - name: LFS Fetch
-        run: |
-          git lfs fetch
-
-      - name: Set up Git
-        run: |
-          git config --global user.name 'GitHub Action'
-          git config --global user.email 'action@github.com'
-
-      - name: Set up SSH
-        uses: webfactory/ssh-agent@v0.9.0
-        with:
-          ssh-private-key: ${{ secrets.SSH_PRIVATE_KEY }}
-
-      - name: Add GitLab public keys
-        run: |
-          ssh-keyscan -H gitlab.com >> ~/.ssh/known_hosts
-      
-      - name: Ensure branch
-        run: |
-          if git symbolic-ref -q HEAD >/dev/null; then
-              echo "Already on a branch, proceeding with push"
-            else
-              echo "Detached HEAD state detected, creating temporary branch"
-            git checkout -b temp_branch
-          fi
-
-      - name: Update LFS Config
-        run: |
-          echo '[lfs]' > .lfsconfig
-          echo '    url = ${{ env.LFS_URL }}' >> .lfsconfig
-          echo '    pushurl = ${{ env.LFS_PUSH_URL }}' >> .lfsconfig
-          echo '    locksverify = false' >> .lfsconfig
-
-      - name: Push LFS
-        id: sync-and-commit
-        run: |
-          git lfs ls-files -l
-          git lfs push --all origin

.github/workflows/prebuilt.yaml

@@ -12,7 +12,7 @@ jobs:
   build_prebuilt:
     name: build prebuilt
     runs-on: ubuntu-latest
-    if: github.repository == 'sunnypilot/sunnypilot'
+    if: github.repository == 'commaai/openpilot'
     env:
       PUSH_IMAGE: true
     permissions:

.github/workflows/release.yaml

@@ -13,7 +13,7 @@ jobs:
     container:
       image: ghcr.io/commaai/openpilot-base:latest
     runs-on: ubuntu-latest
-    if: github.repository == 'sunnypilot/sunnypilot'
+    if: github.repository == 'commaai/openpilot'
     permissions:
       checks: read
       contents: write

.github/workflows/repo-maintenance.yaml

@@ -11,7 +11,7 @@ jobs:
     runs-on: ubuntu-latest
     container:
       image: ghcr.io/commaai/openpilot-base:latest
-    if: github.repository == 'sunnypilot/sunnypilot'
+    if: github.repository == 'commaai/openpilot'
     steps:
     - uses: actions/checkout@v4
       with:

.github/workflows/selfdrive_tests.yaml

@@ -4,7 +4,6 @@ on:
   push:
     branches:
       - master
-      - master-new
   pull_request:
   workflow_dispatch:
   workflow_call:
@@ -15,11 +14,10 @@ on:
         type: string
 
 concurrency:
-  group: selfdrive-tests-ci-run-${{ inputs.run_number }}-${{ github.event_name == 'push' && (github.ref == 'refs/heads/master' || github.ref == 'refs/heads/master-new') && github.run_id || github.head_ref || github.ref }}-${{ github.workflow }}-${{ github.event_name }}
+  group: selfdrive-tests-ci-run-${{ inputs.run_number }}-${{ github.event_name == 'push' && github.ref == 'refs/heads/master' && github.run_id || github.head_ref || github.ref }}-${{ github.workflow }}-${{ github.event_name }}
   cancel-in-progress: true
 
 env:
-  REPORT_NAME: report-${{ inputs.run_number || '1' }}-${{ github.event_name == 'push' && (github.ref == 'refs/heads/master' || github.ref == 'refs/heads/master-new') && 'master' || github.event.number }}
   PYTHONWARNINGS: error
   BASE_IMAGE: openpilot-base
   AZURE_TOKEN: ${{ secrets.AZURE_COMMADATACI_OPENPILOTCI_TOKEN }}
@@ -33,7 +31,6 @@ env:
 
 jobs:
   build_release:
-    if: github.repository == 'commaai/openpilot' # build_release blocked for the time being to only comma as we may have a different process.
     name: build release
     runs-on:
       - ${{ ((github.repository == 'commaai/openpilot') && ((github.event_name != 'pull_request') || (github.event.pull_request.head.repo.full_name == 'commaai/openpilot'))) && 'namespace-profile-amd64-8x16' || 'ubuntu-24.04' }}
@@ -55,7 +52,7 @@ jobs:
       run: TARGET_DIR=$STRIPPED_DIR release/build_devel.sh
     - uses: ./.github/workflows/setup-with-retry
     - name: Check submodules
-      if: github.repository == 'sunnypilot/sunnypilot'
+      if: github.repository == 'commaai/openpilot'
       timeout-minutes: 3
       run: release/check-submodules.sh
     - name: Build openpilot and run checks
@@ -96,9 +93,7 @@ jobs:
 
   build_mac:
     name: build macOS
-    runs-on: ${{ ((github.repository == 'commaai/openpilot') &&
-                   ((github.event_name != 'pull_request') ||
-                    (github.event.pull_request.head.repo.full_name == 'commaai/openpilot'))) && 'namespace-profile-macos-8x14' || 'macos-latest' }}
+    runs-on: ${{ github.repository == 'commaai/openpilot' && 'namespace-profile-macos-8x14' || 'macos-latest' }}
     steps:
     - uses: actions/checkout@v4
       with:
@@ -107,7 +102,6 @@ jobs:
       uses: ./.github/workflows/auto-cache
       with:
         path: ~/Library/Caches/Homebrew
-        key: build_macos_${{ hashFiles('.github/workflows/selfdrive_tests.yaml') }}
     - name: Install dependencies
       run: ./tools/mac_setup.sh
       env:
@@ -119,7 +113,6 @@ jobs:
       uses: ./.github/workflows/auto-cache
       with:
         path: /tmp/scons_cache
-        key: build_macos_${{ hashFiles('.github/workflows/selfdrive_tests.yaml') }}
     - name: Building openpilot
       run: . .venv/bin/activate && scons -j$(nproc)
 
@@ -238,7 +231,7 @@ jobs:
       uses: actions/cache@v4
       with:
         path: .ci_cache/comma_download_cache
-        key: car_models-${{ hashFiles('selfdrive/car/tests/test_models.py', 'opendbc/car/tests/routes.py') }}-${{ matrix.job }}
+        key: car_models-${{ hashFiles('selfdrive/car/tests/test_models.py', 'selfdrive/car/tests/routes.py') }}-${{ matrix.job }}
     - name: Build openpilot
       run: ${{ env.RUN }} "scons -j$(nproc)"
     - name: Test car models
@@ -316,7 +309,6 @@ jobs:
     runs-on:
       - ${{ ((github.repository == 'commaai/openpilot') && ((github.event_name != 'pull_request') || (github.event.pull_request.head.repo.full_name == 'commaai/openpilot'))) && 'namespace-profile-amd64-8x16' || 'ubuntu-24.04' }}
       - ${{ ((github.repository == 'commaai/openpilot') && ((github.event_name != 'pull_request') || (github.event.pull_request.head.repo.full_name == 'commaai/openpilot'))) && 'namespace-experiments:docker.builds.local-cache=separate' || 'ubuntu-24.04' }}
-    if: (github.repository == 'commaai/openpilot') && ((github.event_name != 'pull_request') || (github.event.pull_request.head.repo.full_name == 'commaai/openpilot'))
     steps:
     - uses: actions/checkout@v4
       with:
@@ -361,5 +353,5 @@ jobs:
       - name: Upload Test Report
         uses: actions/upload-artifact@v4
         with:
-          name: ${{ env.REPORT_NAME }}
+          name: report-${{ inputs.run_number || '1' }}-${{ github.event_name == 'push' && github.ref == 'refs/heads/master' && 'master' || github.event.number }}
           path: selfdrive/ui/tests/test_ui/report_1/screenshots

.github/workflows/stale.yaml

@@ -20,7 +20,7 @@ jobs:
           stale-pr-message: 'This PR has had no activity for ${{ env.DAYS_BEFORE_PR_STALE }} days. It will be automatically closed in ${{ env.DAYS_BEFORE_PR_CLOSE }} days if there is no activity.'
           close-pr-message: 'This PR has been automatically closed due to inactivity. Feel free to re-open once activity resumes.'
           stale-pr-label: stale
-          delete-branch: ${{ github.event.pull_request.head.repo.full_name == 'sunnypilot/sunnypilot' }} # only delete branches on the main repo
+          delete-branch: ${{ github.event.pull_request.head.repo.full_name == 'commaai/openpilot' }} # only delete branches on the main repo
           exempt-pr-labels: "ignore stale,needs testing" # if wip or it needs testing from the community, don't mark as stale
           days-before-pr-stale: ${{ env.DAYS_BEFORE_PR_STALE }}
           days-before-pr-close: ${{ env.DAYS_BEFORE_PR_CLOSE }}

.github/workflows/sunnypilot-build-model.yaml

@@ -1,86 +0,0 @@
-name: Build Model from Upstream
-
-env:
-  BUILD_DIR: "/data/openpilot"
-  OUTPUT_DIR: ${{ github.workspace }}/output
-  SCONS_CACHE_DIR: ${{ github.workspace }}/release/ci/scons_cache
-  UPSTREAM_REPO: "commaai/openpilot"
-
-on:
-  workflow_dispatch:
-    inputs:
-      upstream_branch:
-        description: 'Upstream branch to build from'
-        required: true
-        default: 'master'
-        type: string
-      custom_name:
-        description: 'Custom name for the model'
-        required: false
-        type: string
-
-jobs:
-  build_model:
-    runs-on: self-hosted
-
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          repository: ${{ env.UPSTREAM_REPO }}
-          ref: ${{ github.event.inputs.upstream_branch }}
-          submodules: recursive
-
-      - run: git lfs pull
-
-      - name: Cache SCons
-        uses: actions/cache@v4
-        with:
-          path: ${{env.SCONS_CACHE_DIR}}
-          key: scons-${{ runner.os }}-${{ runner.arch }}-${{ github.head_ref || github.ref_name }}-model-${{ github.sha }}
-          restore-keys: |
-            scons-${{ runner.os }}-${{ runner.arch }}-${{ github.head_ref || github.ref_name }}-model-${{ github.sha }}
-            scons-${{ runner.os }}-${{ runner.arch }}-${{ github.head_ref || github.ref_name }}-model
-            scons-${{ runner.os }}-${{ runner.arch }}-${{ github.head_ref || github.ref_name }}
-            scons-${{ runner.os }}-${{ runner.arch }}-master-new
-            scons-${{ runner.os }}-${{ runner.arch }}-master
-            scons-${{ runner.os }}-${{ runner.arch }}
-
-      - name: Setup build environment
-        run: |
-          mkdir -p "${BUILD_DIR}/"
-          sudo find $BUILD_DIR/ -mindepth 1 -delete
-          echo "Starting build stage..."
-          echo "Building from: ${{ env.UPSTREAM_REPO }} branch: ${{ github.event.inputs.upstream_branch }}"
-      
-      - name: Patch SConstruct to pass arbitrary cache
-        run: |
-          sed -i.bak 's#cache_dir =#default_cache_dir =#' ${{ github.workspace }}/SConstruct
-          printf '/default_cache_dir/a\\\ncache_dir = ARGUMENTS.get("cache_dir", default_cache_dir)\n' | sed -i.bak -f - ${{ github.workspace }}/SConstruct
-          cat ${{ github.workspace }}/SConstruct
-
-      - name: Build Model
-        run: |
-          source /etc/profile
-          export UV_PROJECT_ENVIRONMENT=${HOME}/venv
-          export VIRTUAL_ENV=$UV_PROJECT_ENVIRONMENT
-          scons -j$(nproc) cache_dir=${{ env.SCONS_CACHE_DIR }} ${{ github.workspace }}/selfdrive/modeld
-
-      - name: Prepare Output
-        run: |
-          sudo rm -rf ${OUTPUT_DIR}
-          mkdir -p ${OUTPUT_DIR}
-          rsync -avm \
-            --include='*.dlc' \
-            --include='*.thneed' \
-            --include='*.pkl' \
-            --include='*.onnx' \
-            --exclude='*' \
-            --delete-excluded \
-            --chown=comma:comma \
-            ./selfdrive/modeld/models/ ${OUTPUT_DIR}/
-
-      - name: Upload Build Artifacts
-        uses: actions/upload-artifact@v4
-        with:
-          name: model-${{ github.event.inputs.custom_name || github.event.inputs.upstream_branch }}-${{ github.run_number }}
-          path: ${{ env.OUTPUT_DIR }}

.github/workflows/sunnypilot-build-prebuilt.yaml

@@ -1,268 +0,0 @@
-name: sunnypilot prebuilt action
-
-env:
-  BUILD_DIR: "/data/openpilot"
-  OUTPUT_DIR: ${{ github.workspace }}/output
-  CI_DIR: ${{ github.workspace }}/release/ci
-  SCONS_CACHE_DIR: ${{ github.workspace }}/release/ci/scons_cache
-  PUBLIC_REPO_URL: "https://github.com/sunnypilot/sunnypilot"
-  
-  # Branch configurations
-  MASTER_BRANCH: "master"
-  MASTER_NEW_BRANCH: "master-new"
-  DEV_C3_SOURCE_BRANCH: "master-dev-c3-new"
-  
-  # Target branch configurations
-  STAGING_TARGET_BRANCH: "staging-c3-new"
-  DEV_TARGET_BRANCH: "dev-c3-new"
-  RELEASE_TARGET_BRANCH: "release-c3-new"
-
-on:
-  push:
-    branches: [ master, master-new, master-dev-c3-new ]
-    tags: [ '*' ]
-  pull_request:
-    branches: [ master, master-new ]
-  workflow_dispatch:
-    inputs:
-      extra_version:
-        description: 'Extra version identifier'
-        required: false
-        default: ''
-
-jobs:
-  build:
-    concurrency:
-      group: build-${{ github.head_ref || github.ref_name }}
-      cancel-in-progress: false
-    runs-on: self-hosted
-    outputs:
-      new_branch: ${{ steps.set-env.outputs.new_branch }}
-      version: ${{ steps.set-env.outputs.version }}
-      extra_version_identifier: ${{ steps.set-env.outputs.extra_version_identifier }}
-    steps:
-      - uses: actions/checkout@v4
-        with:
-          submodules: recursive
-      - run: git lfs pull
-
-      - name: Cache SCons
-        uses: actions/cache@v4
-        with:
-          path: ${{env.SCONS_CACHE_DIR}}
-          key: scons-${{ runner.os }}-${{ runner.arch }}-${{ github.head_ref || github.ref_name }}-${{ github.sha }}
-          restore-keys: |
-            scons-${{ runner.os }}-${{ runner.arch }}-${{ github.head_ref }}
-            scons-${{ runner.os }}-${{ runner.arch }}-${{ github.ref_name }}
-            scons-${{ runner.os }}-${{ runner.arch }}-${{ env.MASTER_NEW_BRANCH }}
-            scons-${{ runner.os }}-${{ runner.arch }}-${{ env.MASTER_BRANCH }}
-            scons-${{ runner.os }}-${{ runner.arch }}
-
-      - name: Set Configuration
-        run: |
-          if [[ "${{ github.ref_name }}" == "${{ env.DEV_C3_SOURCE_BRANCH }}" ]]; then
-            # Dev configuration
-            echo "BRANCH_TYPE=dev" >> $GITHUB_ENV
-            echo "NEW_BRANCH=${{ env.DEV_TARGET_BRANCH }}" >> $GITHUB_ENV
-            echo "VERSION=$(date '+%Y.%m.%d')-${{ github.run_number }}" >> $GITHUB_ENV
-            echo "EXTRA_VERSION_IDENTIFIER=${{ github.run_number }}" >> $GITHUB_ENV
-          
-          elif [[ "${{ github.ref_name }}" == "${{ env.MASTER_BRANCH }}" || "${{ github.ref_name }}" == "${{ env.MASTER_NEW_BRANCH }}" ]]; then
-            # Master configuration
-            echo "BRANCH_TYPE=master" >> $GITHUB_ENV
-            echo "NEW_BRANCH=${{ env.STAGING_TARGET_BRANCH }}" >> $GITHUB_ENV
-            echo "EXTRA_VERSION_IDENTIFIER=staging" >> $GITHUB_ENV
-            echo "VERSION=$(cat common/version.h | grep COMMA_VERSION | sed -e 's/[^0-9|.]//g')-staging" >> $GITHUB_ENV
-          
-          elif [[ "${{ github.ref }}" == refs/tags/* ]]; then
-            # Tag configuration
-            echo "BRANCH_TYPE=tag" >> $GITHUB_ENV
-            echo "NEW_BRANCH=${{ env.RELEASE_TARGET_BRANCH }}" >> $GITHUB_ENV
-            echo "EXTRA_VERSION_IDENTIFIER=release" >> $GITHUB_ENV
-            echo "VERSION=$(cat common/version.h | grep COMMA_VERSION | sed -e 's/[^0-9|.]//g')-release" >> $GITHUB_ENV
-          
-          else
-            # Feature branch configuration
-            echo "BRANCH_TYPE=dispatch" >> $GITHUB_ENV
-            echo "NEW_BRANCH=${{ github.head_ref || github.ref_name }}-prebuilt" >> $GITHUB_ENV
-            echo "VERSION=$(date '+%Y.%m.%d')-${{ github.run_number }}" >> $GITHUB_ENV
-          fi
-
-      - name: Set environment variables
-        id: set-env
-        run: |
-          # Write to GITHUB_OUTPUT from environment variables
-          echo "new_branch=$NEW_BRANCH" >> $GITHUB_OUTPUT
-          [[ ! -z "$EXTRA_VERSION_IDENTIFIER" ]] && echo "extra_version_identifier=$EXTRA_VERSION_IDENTIFIER" >> $GITHUB_OUTPUT
-          [[ ! -z "$VERSION" ]] && echo "version=$VERSION" >> $GITHUB_OUTPUT
-          
-          # Set up common environment
-          source /etc/profile;
-          export UV_PROJECT_ENVIRONMENT=${HOME}/venv
-          export VIRTUAL_ENV=$UV_PROJECT_ENVIRONMENT
-          printenv >> $GITHUB_ENV
-          if [[ "${{ runner.debug }}" == "1" ]]; then
-            cat $GITHUB_OUTPUT
-          fi
-
-      - name: Setup build environment
-        run: |
-          mkdir -p "${BUILD_DIR}/"
-          sudo find $BUILD_DIR/ -mindepth 1 -delete
-          echo "Starting build stage..."
-          echo "BUILD_DIR: ${BUILD_DIR}"
-          echo "CI_DIR: ${CI_DIR}"
-          echo "VERSION: ${{ steps.set-env.outputs.version }}"
-          echo "UV_PROJECT_ENVIRONMENT: ${UV_PROJECT_ENVIRONMENT}"
-          echo "VIRTUAL_ENV: ${VIRTUAL_ENV}"
-          echo "-------"
-          if [[ "${{ runner.debug }}" == "1" ]]; then
-            printenv
-          fi
-
-      - name: Build Panda
-        run: |
-          scons -j$(nproc) cache_dir=${{env.SCONS_CACHE_DIR}} ${{ github.workspace }}/panda
-
-      - name: Build Main Project
-        run: |
-          export PYTHONPATH="$BUILD_DIR"
-          ./release/release_files.py | sort | uniq | rsync -rRl${RUNNER_DEBUG:+v} --files-from=- . $BUILD_DIR/
-          cd $BUILD_DIR
-          sed -i '/from .board.jungle import PandaJungle, PandaJungleDFU/s/^/#/' panda/__init__.py
-          scons -j$(nproc) cache_dir=${{env.SCONS_CACHE_DIR}} --minimal
-          touch ${BUILD_DIR}/prebuilt
-          if [[ "${{ runner.debug }}" == "1" ]]; then
-            ls -la ${BUILD_DIR}
-          fi
-
-      - name: Prepare Output
-        run: |
-          sudo rm -rf ${OUTPUT_DIR}
-          mkdir -p ${OUTPUT_DIR}
-          rsync -am${RUNNER_DEBUG:+v} \
-            --include='**/panda/board/' \
-            --include='**/panda/board/obj' \
-            --include='**/panda/board/obj/panda.bin.signed' \
-            --include='**/panda/board/obj/panda_h7.bin.signed' \
-            --include='**/panda/board/obj/bootstub.panda.bin' \
-            --include='**/panda/board/obj/bootstub.panda_h7.bin' \
-            --exclude='.sconsign.dblite' \
-            --exclude='*.a' \
-            --exclude='*.o' \
-            --exclude='*.os' \
-            --exclude='*.pyc' \
-            --exclude='moc_*' \
-            --exclude='*.cc' \
-            --exclude='Jenkinsfile' \
-            --exclude='supercombo.onnx' \
-            --exclude='**/panda/board/*' \
-            --exclude='**/panda/board/obj/**' \
-            --exclude='**/panda/certs/' \
-            --exclude='**/panda/crypto/' \
-            --exclude='**/release/' \
-            --exclude='**/.github/' \
-            --exclude='**/selfdrive/ui/replay/' \
-            --exclude='**/__pycache__/' \
-            --exclude='**/selfdrive/ui/*.h' \
-            --exclude='**/selfdrive/ui/**/*.h' \
-            --exclude='**/selfdrive/ui/qt/offroad/sunnypilot/' \
-            --exclude='${{env.SCONS_CACHE_DIR}}' \
-            --exclude='**/.git/' \
-            --exclude='**/SConstruct' \
-            --exclude='**/SConscript' \
-            --exclude='**/.venv/' \
-            --delete-excluded \
-            --chown=comma:comma \
-            ${BUILD_DIR}/ ${OUTPUT_DIR}/
-
-      - name: 'Tar.gz files'
-        run: |
-          tar czf prebuilt.tar.gz -C ${{ env.OUTPUT_DIR }} .
-          ls -la prebuilt.tar.gz
-
-      - name: 'Upload Artifact'
-        uses: actions/upload-artifact@v4
-        with:
-          name: prebuilt
-          path: prebuilt.tar.gz
-
-  publish:
-    concurrency:
-      group: publish-${{ github.head_ref || github.ref_name }}
-      cancel-in-progress: true
-    if: ${{ github.event_name != 'pull_request' || github.event_name == 'pull_request' && github.event.pull_request.draft }} 
-    needs: build
-    runs-on: ubuntu-24.04
-    environment: ${{ contains(fromJSON(vars.AUTO_DEPLOY_PREBUILT_BRANCHES), github.head_ref || github.ref_name) && 'auto-deploy' || 'feature-branch' }}
-    steps:
-      - uses: actions/checkout@v4
-
-      - name: Download build artifacts
-        uses: actions/download-artifact@v4
-        with:
-          name: prebuilt
-
-      - name: Untar prebuilt
-        run: |
-          mkdir -p ${{ env.OUTPUT_DIR }}
-          tar xzf prebuilt.tar.gz -C ${{ env.OUTPUT_DIR }}
-
-      - name: Configure Git
-        run: |
-          git config --global user.email "github-actions[bot]@users.noreply.github.com"
-          git config --global user.name "github-actions[bot]"
-
-      - name: Publish to Public Repository
-        env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        run: |
-          echo '${{ toJSON(needs.build.outputs) }}'
-          ls -la ${{ env.OUTPUT_DIR }}
-
-          ${{ env.CI_DIR }}/publish.sh \
-            "${{ github.workspace }}" \
-            "${{ env.OUTPUT_DIR }}" \
-            "${{ needs.build.outputs.new_branch }}" \
-            "${{ needs.build.outputs.version }}" \
-            "https://x-access-token:${{github.token}}@github.com/sunnypilot/sunnypilot.git" \
-            "-${{ needs.build.outputs.extra_version_identifier }}"
-          
-          echo ""
-          echo "----   ℹ️ To update the list of branches that auto deploy prebuilts  -----"
-          echo ""
-          echo "1. Go to: ${{ github.server_url }}/${{ github.repository }}/settings/variables/actions/AUTO_DEPLOY_PREBUILT_BRANCHES"
-          echo "2. Current value: ${{ vars.AUTO_DEPLOY_PREBUILT_BRANCHES }}"
-          echo "3. Update as needed (JSON array with no spaces)"
-
-  notify:
-    needs: [ build, publish ]
-    runs-on: ubuntu-24.04
-    if: success()
-    steps:
-      - uses: actions/checkout@v4
-      - name: Setup Alpine Linux environment
-        uses: jirutka/setup-alpine@v1.2.0
-        with:
-          packages: 'jq gettext curl'
-
-      - name: Send Discord Notification
-        env:
-          DISCORD_WEBHOOK: ${{ contains(fromJSON(vars.DEV_FEEDBACK_NOTIFICATION_BRANCHES), github.head_ref || github.ref_name) && secrets.DISCORD_DEV_FEEDBACK_CHANNEL_WEBHOOK || secrets.DISCORD_DEV_PRIVATE_CHANNEL_WEBHOOK }}
-        run: |
-          TEMPLATE='${{ vars.DISCORD_GENERAL_UPDATE_NOTICE }}'
-          export EXTRA_VERSION_IDENTIFIER="${{ needs.build.outputs.extra_version_identifier }}"
-          export VERSION="${{ needs.build.outputs.version }}"
-          export branch_name=${{ github.head_ref || github.ref_name }}
-          export new_branch=${{ needs.build.outputs.new_branch }}
-          export extra_version_identifier=${{ needs.build.outputs.extra_version_identifier || github.run_number}}
-          echo ${TEMPLATE} | envsubst | jq -c '.' | tee payload.json
-          curl -X POST -H "Content-Type: application/json" -d @payload.json $DISCORD_WEBHOOK
-          
-          echo ""
-          echo "----   ℹ️ To update the list of branches that notify to dev-feedback  -----"
-          echo ""
-          echo "1. Go to: ${{ github.server_url }}/${{ github.repository }}/settings/variables/actions/DEV_FEEDBACK_NOTIFICATION_BRANCHES"
-          echo "2. Current value: ${{ vars.DEV_FEEDBACK_NOTIFICATION_BRANCHES }}"
-          echo "3. Update as needed (JSON array with no spaces)"
-        shell: alpine.sh {0}

.github/workflows/ui_preview.yaml

@@ -3,25 +3,23 @@ on:
   push:
     branches:
       - master
-      - master-new
   pull_request_target:
     types: [assigned, opened, synchronize, reopened, edited]
     branches:
       - 'master'
-      - 'master-new'
     paths:
       - 'selfdrive/ui/**'
   workflow_dispatch:
 
 env:
   UI_JOB_NAME: "Create UI Report"
-  REPORT_NAME: ${{ github.event_name == 'push' && (github.ref == 'refs/heads/master' || github.ref == 'refs/heads/master-new') && 'master' || github.event.number }}
-  SHA: ${{ github.event_name == 'push' && (github.ref == 'refs/heads/master' || github.ref == 'refs/heads/master-new') && github.sha || github.event.pull_request.head.sha }}
+  REPORT_NAME: ${{ github.event_name == 'push' && github.ref == 'refs/heads/master' && 'master' || github.event.number }}
+  SHA: ${{ github.event_name == 'push' && github.ref == 'refs/heads/master' && github.sha || github.event.pull_request.head.sha }}
   BRANCH_NAME: "openpilot/pr-${{ github.event.number }}"
 
 jobs:
   preview:
-    if: github.repository == 'sunnypilot/sunnypilot'
+    if: github.repository == 'commaai/openpilot'
     name: preview
     runs-on: ubuntu-latest
     timeout-minutes: 20
@@ -60,13 +58,13 @@ jobs:
       - name: Getting master ui
         uses: actions/checkout@v4
         with:
-          repository: sunnypilot/ci-artifacts
+          repository: commaai/ci-artifacts
           ssh-key: ${{ secrets.CI_ARTIFACTS_DEPLOY_KEY }}
           path: ${{ github.workspace }}/master_ui
           ref: openpilot_master_ui
 
       - name: Saving new master ui
-        if: (github.ref == 'refs/heads/master' || github.ref == 'refs/heads/master-new') && github.event_name == 'push'
+        if: github.ref == 'refs/heads/master' && github.event_name == 'push'
         working-directory: ${{ github.workspace }}/master_ui
         run: |
           git checkout --orphan=new_master_ui
@@ -108,13 +106,13 @@ jobs:
               DIFF="${DIFF}<table>"
 
               DIFF="${DIFF}<tr>"
-              DIFF="${DIFF}  <td> master <img src=\"https://raw.githubusercontent.com/sunnypilot/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}_master_ref.png\"> </td>"
-              DIFF="${DIFF}  <td> proposed <img src=\"https://raw.githubusercontent.com/sunnypilot/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}.png\"> </td>"
+              DIFF="${DIFF}  <td> master <img src=\"https://raw.githubusercontent.com/commaai/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}_master_ref.png\"> </td>"
+              DIFF="${DIFF}  <td> proposed <img src=\"https://raw.githubusercontent.com/commaai/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}.png\"> </td>"
               DIFF="${DIFF}</tr>"
 
               DIFF="${DIFF}<tr>"
-              DIFF="${DIFF}  <td> diff <img src=\"https://raw.githubusercontent.com/sunnypilot/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}_diff.png\"> </td>"
-              DIFF="${DIFF}  <td> composite diff <img src=\"https://raw.githubusercontent.com/sunnypilot/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}_diff.gif\"> </td>"
+              DIFF="${DIFF}  <td> diff <img src=\"https://raw.githubusercontent.com/commaai/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}_diff.png\"> </td>"
+              DIFF="${DIFF}  <td> composite diff <img src=\"https://raw.githubusercontent.com/commaai/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}_diff.gif\"> </td>"
               DIFF="${DIFF}</tr>"
 
               DIFF="${DIFF}</table>"
@@ -127,7 +125,7 @@ jobs:
             if [[ $INDEX -eq 0 ]]; then
               TABLE="${TABLE}<tr>"
             fi
-            TABLE="${TABLE}  <td> <img src=\"https://raw.githubusercontent.com/sunnypilot/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}.png\"> </td>"
+            TABLE="${TABLE}  <td> <img src=\"https://raw.githubusercontent.com/commaai/ci-artifacts/${{ env.BRANCH_NAME }}/${A[$i]}.png\"> </td>"
             if [[ $INDEX -eq 1 || $(($i + 1)) -eq ${#A[*]} ]]; then
               TABLE="${TABLE}</tr>"
             fi

.gitignore

@@ -103,8 +103,3 @@ Pipfile
 # Ignore all local history of files
 .history
 .ionide
-
-### JetBrains ###
-!.idea/customTargets.xml
-!.idea/tools/*
-!.run/*

.gitmodules

@@ -1,18 +1,18 @@
 [submodule "panda"]
   path = panda
-  url = https://github.com/sunnyhaibin/panda.git
+  url = ../../commaai/panda.git
 [submodule "opendbc"]
   path = opendbc_repo
-  url = https://github.com/sunnypilot/opendbc.git
+  url = ../../commaai/opendbc.git
 [submodule "msgq"]
   path = msgq_repo
-  url = https://github.com/sunnypilot/msgq.git
+  url = ../../commaai/msgq.git
 [submodule "rednose_repo"]
   path = rednose_repo
-  url = https://github.com/commaai/rednose.git
+  url = ../../commaai/rednose.git
 [submodule "teleoprtc_repo"]
   path = teleoprtc_repo
-  url = https://github.com/commaai/teleoprtc
+  url = ../../commaai/teleoprtc
 [submodule "tinygrad"]
   path = tinygrad_repo
-  url = https://github.com/tinygrad/tinygrad.git
+  url = https://github.com/commaai/tinygrad.git

.idea/customTargets.xml

@@ -1,25 +0,0 @@
-<?xml version="1.0" encoding="UTF-8"?>
-<project version="4">
-  <component name="CLionExternalBuildManager">
-    <target id="a62f99e8-5ec4-434c-8122-49efed5af108" name="uv Scons Build Debug" defaultType="TOOL">
-      <configuration id="b93ec964-16e5-4962-a12e-3ed360ce8f02" name="uv Scons Build Debug">
-        <build type="TOOL">
-          <tool actionId="Tool_External Tools_uv Scons Build Debug" />
-        </build>
-        <clean type="TOOL">
-          <tool actionId="Tool_External Tools_uv Scons Clean" />
-        </clean>
-      </configuration>
-    </target>
-    <target id="edd8ad9d-183b-467c-a355-0d9a0ecab026" name="uv Scons Build Release" defaultType="TOOL">
-      <configuration id="09523339-5ce3-4223-ab9e-904f38ad7752" name="uv Scons Build Release">
-        <build type="TOOL">
-          <tool actionId="Tool_External Tools_uv Scons Build Release" />
-        </build>
-        <clean type="TOOL">
-          <tool actionId="Tool_External Tools_uv Scons Clean" />
-        </clean>
-      </configuration>
-    </target>
-  </component>
-</project>

.idea/tools/External Tools.xml

@@ -1,23 +0,0 @@
-<toolSet name="External Tools">
-  <tool name="uv Scons Build Debug" showInMainMenu="false" showInEditor="false" showInProject="false" showInSearchPopup="false" disabled="false" useConsole="true" showConsoleOnStdOut="false" showConsoleOnStdErr="false" synchronizeAfterRun="true">
-    <exec>
-      <option name="COMMAND" value="bash" />
-      <option name="PARAMETERS" value="-c &quot;source .venv/bin/activate &amp;&amp; scons -u -j$(nproc) --compile_db --ccflags=\&quot;-fno-inline\&quot;&quot;" />
-      <option name="WORKING_DIRECTORY" value="$ProjectFileDir$" />
-    </exec>
-  </tool>
-  <tool name="uv Scons Clean" showInMainMenu="false" showInEditor="false" showInProject="false" showInSearchPopup="false" disabled="false" useConsole="true" showConsoleOnStdOut="false" showConsoleOnStdErr="false" synchronizeAfterRun="true">
-    <exec>
-      <option name="COMMAND" value="bash" />
-      <option name="PARAMETERS" value="-c &quot;source .venv/bin/activate &amp;&amp; scons -c&quot; " />
-      <option name="WORKING_DIRECTORY" value="$ProjectFileDir$" />
-    </exec>
-  </tool>
-  <tool name="uv Scons Build Release" showInMainMenu="false" showInEditor="false" showInProject="false" showInSearchPopup="false" disabled="false" useConsole="true" showConsoleOnStdOut="false" showConsoleOnStdErr="false" synchronizeAfterRun="true">
-    <exec>
-      <option name="COMMAND" value="bash" />
-      <option name="PARAMETERS" value="-c &quot;source .venv/bin/activate &amp;&amp; scons -u -j$(nproc) --compile_db&quot; " />
-      <option name="WORKING_DIRECTORY" value="$ProjectFileDir$" />
-    </exec>
-  </tool>
-</toolSet>

.lfsconfig

@@ -1,4 +1,4 @@
 [lfs]
-	url = https://gitlab.com/sunnypilot/public/sunnypilot-new-lfs.git/info/lfs
-	pushurl = ssh://git@gitlab.com/sunnypilot/public/sunnypilot-new-lfs.git
+	url = https://gitlab.com/commaai/openpilot-lfs.git/info/lfs
+	pushurl = ssh://git@gitlab.com/commaai/openpilot-lfs.git
 	locksverify = false

.lfsconfig-comma

@@ -1,4 +0,0 @@
-[lfs]
-	url = https://gitlab.com/commaai/openpilot-lfs.git/info/lfs
-	pushurl = ssh://git@gitlab.com/commaai/openpilot-lfs.git
-	locksverify = false

.run/Build Debug.run.xml

@@ -1,10 +0,0 @@
-<component name="ProjectRunConfigurationManager">
-  <configuration default="false" name="Build Debug" type="CLionExternalRunConfiguration" factoryName="Application" REDIRECT_INPUT="false" ELEVATE="false" USE_EXTERNAL_CONSOLE="false" EMULATE_TERMINAL="false" WORKING_DIR="file://$ProjectFileDir$/selfdrive/ui" PASS_PARENT_ENVS_2="true" PROJECT_NAME="sunnypilot" TARGET_NAME="uv Scons Build Debug" CONFIG_NAME="uv Scons Build Debug" RUN_PATH="ui">
-    <envs>
-      <env name="QT_DBL_CLICK_DIST" value="150" />
-    </envs>
-    <method v="2">
-      <option name="CLION.EXTERNAL.BUILD" enabled="true" />
-    </method>
-  </configuration>
-</component>

.run/Build Release.run.xml

@@ -1,10 +0,0 @@
-<component name="ProjectRunConfigurationManager">
-  <configuration default="false" name="Build Release" type="CLionExternalRunConfiguration" factoryName="Application" REDIRECT_INPUT="false" ELEVATE="false" USE_EXTERNAL_CONSOLE="false" EMULATE_TERMINAL="false" WORKING_DIR="file://$ProjectFileDir$/selfdrive/ui" PASS_PARENT_ENVS_2="true" PROJECT_NAME="sunnypilot" TARGET_NAME="uv Scons Build Release" CONFIG_NAME="uv Scons Build Release" RUN_PATH="ui">
-    <envs>
-      <env name="QT_DBL_CLICK_DIST" value="150" />
-    </envs>
-    <method v="2">
-      <option name="CLION.EXTERNAL.BUILD" enabled="true" />
-    </method>
-  </configuration>
-</component>

Dockerfile.openpilot

@@ -1,9 +1,9 @@
 FROM ghcr.io/commaai/openpilot-base:latest
 
-ENV PYTHONUNBUFFERED=1
+ENV PYTHONUNBUFFERED 1
 
-ENV OPENPILOT_PATH=/home/batman/openpilot
-ENV PYTHONPATH=${OPENPILOT_PATH}:${PYTHONPATH}
+ENV OPENPILOT_PATH /home/batman/openpilot
+ENV PYTHONPATH ${OPENPILOT_PATH}:${PYTHONPATH}
 
 RUN mkdir -p ${OPENPILOT_PATH}
 WORKDIR ${OPENPILOT_PATH}

Dockerfile.openpilot_base

@@ -1,16 +1,16 @@
 FROM ubuntu:24.04
 
-ENV PYTHONUNBUFFERED=1
+ENV PYTHONUNBUFFERED 1
 
 ENV DEBIAN_FRONTEND=noninteractive
 RUN apt-get update && \
-    apt-get install -y --no-install-recommends sudo tzdata locales ssh pulseaudio xvfb x11-xserver-utils gnome-screenshot python3-tk python3-dev && \
+    apt-get install -y --no-install-recommends sudo tzdata locales ssh pulseaudio xvfb x11-xserver-utils gnome-screenshot && \
     rm -rf /var/lib/apt/lists/*
 
 RUN sed -i -e 's/# en_US.UTF-8 UTF-8/en_US.UTF-8 UTF-8/' /etc/locale.gen && locale-gen
-ENV LANG=en_US.UTF-8
-ENV LANGUAGE=en_US:en
-ENV LC_ALL=en_US.UTF-8
+ENV LANG en_US.UTF-8
+ENV LANGUAGE en_US:en
+ENV LC_ALL en_US.UTF-8
 
 COPY tools/install_ubuntu_dependencies.sh /tmp/tools/
 RUN /tmp/tools/install_ubuntu_dependencies.sh && \
@@ -55,9 +55,9 @@ RUN mkdir -p /tmp/opencl-driver-intel && \
     cd / && \
     rm -rf /tmp/opencl-driver-intel
 
-ENV NVIDIA_VISIBLE_DEVICES=all
-ENV NVIDIA_DRIVER_CAPABILITIES=graphics,utility,compute
-ENV QTWEBENGINE_DISABLE_SANDBOX=1
+ENV NVIDIA_VISIBLE_DEVICES all
+ENV NVIDIA_DRIVER_CAPABILITIES graphics,utility,compute
+ENV QTWEBENGINE_DISABLE_SANDBOX 1
 
 RUN dbus-uuidgen > /etc/machine-id
 

LICENSE.md

@@ -1,21 +0,0 @@
-# Custom MIT License
-
-Copyright (c) 2024, Haibin Wen, SUNNYPILOT LLC
-
-Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to view and modify the Software, subject to the following conditions:
-
-1. **Permission Required**: Permission Required for Commercial, For-Profit, or Closed Source Use: Use of the Software, in whole or in part, for any commercial purposes, for-profit projects, or in closed source projects requires explicit written permission from the original author(s).
-
-2. **Redistribution**: Any redistribution of the Software, modified or unmodified, must retain this license notice and the following acknowledgment:
-   "This software is licensed under a custom license requiring permission for use."
-
-3. **Visibility**: Any project that uses the Software must visibly mention the following acknowledgment:
-   "This project uses software from Haibin Wen and SUNNYPILOT LLC and is licensed under a custom license requiring permission for use."
-
-4. **No Warranty**: THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-Contact sunnypilot Support <support@sunnypilot.ai> for permission requests.
-
----
-
-Haibin Wen, SUNNYPILOT LLC

README.md

@@ -38,8 +38,7 @@ Quick start: `bash <(curl -fsSL openpilot.comma.ai)`
   </tr>
 </table>
 
-
-Using openpilot in a car
+To start using openpilot in a car
 ------
 
 To use openpilot in a car, you need four things:
@@ -50,14 +49,6 @@ To use openpilot in a car, you need four things:
 
 We have detailed instructions for [how to install the harness and device in a car](https://comma.ai/setup). Note that it's possible to run openpilot on [other hardware](https://blog.comma.ai/self-driving-car-for-free/), although it's not plug-and-play.
 
-### Branches
-| branch           | URL                                    | description                                                                         |
-|------------------|----------------------------------------|-------------------------------------------------------------------------------------|
-| `release3`         | openpilot.comma.ai                      | This is openpilot's release branch.                                                 |
-| `release3-staging` | openpilot-test.comma.ai                | This is the staging branch for releases. Use it to get new releases slightly early. |
-| `nightly`          | openpilot-nightly.comma.ai             | This is the bleeding edge development branch. Do not expect this to be stable.      |
-| `nightly-dev`      | installer.comma.ai/commaai/nightly-dev | Same as nightly, but includes experimental development features for some cars.      |
-
 To start developing openpilot
 ------
 

SConstruct

@@ -237,8 +237,7 @@ if GetOption('compile_db'):
   env.CompilationDatabase('compile_commands.json')
 
 # Setup cache dir
-default_cache_dir = '/data/scons_cache' if AGNOS else '/tmp/scons_cache'
-cache_dir = ARGUMENTS.get('cache_dir', default_cache_dir)
+cache_dir = '/data/scons_cache' if AGNOS else '/tmp/scons_cache'
 CacheDir(cache_dir)
 Clean(["."], cache_dir)
 

cereal/log.capnp

@@ -2440,14 +2440,6 @@ struct Microphone {
   filteredSoundPressureWeightedDb @2 :Float32;
 }
 
-struct Touch {
-  sec @0 :Int64;
-  usec @1 :Int64;
-  type @2 :UInt8;
-  code @3 :Int32;
-  value @4 :Int32;
-}
-
 struct Event {
   logMonoTime @0 :UInt64;  # nanoseconds
   valid @67 :Bool = true;
@@ -2528,9 +2520,6 @@ struct Event {
     logMessage @18 :Text;
     errorLogMessage @85 :Text;
 
-    # touch frame
-    touch @135 :List(Touch);
-
     # navigation
     navInstruction @82 :NavInstruction;
     navRoute @83 :NavRoute;

cereal/services.py

@@ -22,7 +22,6 @@ _services: dict[str, tuple] = {
   "temperatureSensor2": (True, 2., 200),
   "gpsNMEA": (True, 9.),
   "deviceState": (True, 2., 1),
-  "touch": (True, 20., 1),
   "can": (True, 100., 2053),  # decimation gives ~3 msgs in a full segment
   "controlsState": (True, 100., 10),
   "selfdriveState": (True, 100., 10),

common/params.cc

@@ -200,7 +200,6 @@ std::unordered_map<std::string, uint32_t> keys = {
     {"UpdaterTargetBranch", CLEAR_ON_MANAGER_START},
     {"UpdaterLastFetchTime", PERSISTENT},
     {"Version", PERSISTENT},
-    {"EnableGithubRunner", PERSISTENT},
 };
 
 } // namespace

common/transformations/model.py

@@ -1,7 +1,7 @@
 import numpy as np
 
 from openpilot.common.transformations.orientation import rot_from_euler
-from openpilot.common.transformations.camera import get_view_frame_from_calib_frame, view_frame_from_device_frame, _ar_ox_fisheye
+from openpilot.common.transformations.camera import get_view_frame_from_calib_frame, view_frame_from_device_frame
 
 # segnet
 SEGNET_SIZE = (512, 384)
@@ -39,13 +39,6 @@ sbigmodel_intrinsics = np.array([
   [0.0,  sbigmodel_fl,      0.5 * (256 + MEDMODEL_CY)],
   [0.0,  0.0,                                     1.0]])
 
-DM_INPUT_SIZE = (1440, 960)
-dmonitoringmodel_fl = _ar_ox_fisheye.focal_length
-dmonitoringmodel_intrinsics = np.array([
-  [dmonitoringmodel_fl,  0.0, DM_INPUT_SIZE[0]/2],
-  [0.0, dmonitoringmodel_fl, DM_INPUT_SIZE[1]/2 - (_ar_ox_fisheye.height - DM_INPUT_SIZE[1])/2],
-  [0.0,  0.0, 1.0]])
-
 bigmodel_frame_from_calib_frame = np.dot(bigmodel_intrinsics,
   get_view_frame_from_calib_frame(0, 0, 0, 0))
 

docs/CARS.md

@@ -103,7 +103,7 @@ A supported vehicle is one that just works when you install a comma device. All
 |Hyundai|Ioniq Plug-in Hybrid 2020-22|All|openpilot available[<sup>1</sup>](#footnotes)|0 mph|0 mph|[![star](assets/icon-star-full.svg)](##)|[![star](assets/icon-star-full.svg)](##)|<details><summary>Parts</summary><sub>- 1 Hyundai H connector<br>- 1 RJ45 cable (7 ft)<br>- 1 comma 3X<br>- 1 comma power v2<br>- 1 harness box<br>- 1 mount<br>- 1 right angle OBD-C cable (1.5 ft)<br><a href="https://comma.ai/shop/comma-3x.html?make=Hyundai&model=Ioniq Plug-in Hybrid 2020-22">Buy Here</a></sub></details>||
 |Hyundai|Kona 2020|Smart Cruise Control (SCC)|openpilot available[<sup>1</sup>](#footnotes)|6 mph|0 mph|[![star](assets/icon-star-full.svg)](##)|[![star](assets/icon-star-empty.svg)](##)|<details><summary>Parts</summary><sub>- 1 Hyundai B connector<br>- 1 RJ45 cable (7 ft)<br>- 1 comma 3X<br>- 1 comma power v2<br>- 1 harness box<br>- 1 mount<br>- 1 right angle OBD-C cable (1.5 ft)<br><a href="https://comma.ai/shop/comma-3x.html?make=Hyundai&model=Kona 2020">Buy Here</a></sub></details>||
 |Hyundai|Kona Electric 2018-21|Smart Cruise Control (SCC)|openpilot available[<sup>1</sup>](#footnotes)|0 mph|0 mph|[![star](assets/icon-star-full.svg)](##)|[![star](assets/icon-star-full.svg)](##)|<details><summary>Parts</summary><sub>- 1 Hyundai G connector<br>- 1 RJ45 cable (7 ft)<br>- 1 comma 3X<br>- 1 comma power v2<br>- 1 harness box<br>- 1 mount<br>- 1 right angle OBD-C cable (1.5 ft)<br><a href="https://comma.ai/shop/comma-3x.html?make=Hyundai&model=Kona Electric 2018-21">Buy Here</a></sub></details>||
-|Hyundai|Kona Electric 2022-23|Smart Cruise Control (SCC)|openpilot available[<sup>1</sup>](#footnotes)|0 mph|0 mph|[![star](assets/icon-star-full.svg)](##)|[![star](assets/icon-star-full.svg)](##)|<details><summary>Parts</summary><sub>- 1 Hyundai O connector<br>- 1 RJ45 cable (7 ft)<br>- 1 comma 3X<br>- 1 comma power v2<br>- 1 harness box<br>- 1 mount<br>- 1 right angle OBD-C cable (1.5 ft)<br><a href="https://comma.ai/shop/comma-3x.html?make=Hyundai&model=Kona Electric 2022-23">Buy Here</a></sub></details>||
+|Hyundai|Kona Electric 2022-23|Smart Cruise Control (SCC)|Stock|0 mph|0 mph|[![star](assets/icon-star-full.svg)](##)|[![star](assets/icon-star-full.svg)](##)|<details><summary>Parts</summary><sub>- 1 Hyundai O connector<br>- 1 RJ45 cable (7 ft)<br>- 1 comma 3X<br>- 1 comma power v2<br>- 1 harness box<br>- 1 mount<br>- 1 right angle OBD-C cable (1.5 ft)<br><a href="https://comma.ai/shop/comma-3x.html?make=Hyundai&model=Kona Electric 2022-23">Buy Here</a></sub></details>||
 |Hyundai|Kona Electric (with HDA II, Korea only) 2023[<sup>5</sup>](#footnotes)|Smart Cruise Control (SCC)|Stock|0 mph|0 mph|[![star](assets/icon-star-full.svg)](##)|[![star](assets/icon-star-full.svg)](##)|<details><summary>Parts</summary><sub>- 1 Hyundai R connector<br>- 1 RJ45 cable (7 ft)<br>- 1 comma 3X<br>- 1 comma power v2<br>- 1 harness box<br>- 1 mount<br>- 1 right angle OBD-C cable (1.5 ft)<br><a href="https://comma.ai/shop/comma-3x.html?make=Hyundai&model=Kona Electric (with HDA II, Korea only) 2023">Buy Here</a></sub></details>|<a href="https://www.youtube.com/watch?v=U2fOCmcQ8hw" target="_blank"><img height="18px" src="assets/icon-youtube.svg"></img></a>|
 |Hyundai|Kona Hybrid 2020|Smart Cruise Control (SCC)|openpilot available[<sup>1</sup>](#footnotes)|0 mph|0 mph|[![star](assets/icon-star-full.svg)](##)|[![star](assets/icon-star-full.svg)](##)|<details><summary>Parts</summary><sub>- 1 Hyundai I connector<br>- 1 RJ45 cable (7 ft)<br>- 1 comma 3X<br>- 1 comma power v2<br>- 1 harness box<br>- 1 mount<br>- 1 right angle OBD-C cable (1.5 ft)<br><a href="https://comma.ai/shop/comma-3x.html?make=Hyundai&model=Kona Hybrid 2020">Buy Here</a></sub></details>||
 |Hyundai|Palisade 2020-22|All|openpilot available[<sup>1</sup>](#footnotes)|0 mph|0 mph|[![star](assets/icon-star-full.svg)](##)|[![star](assets/icon-star-full.svg)](##)|<details><summary>Parts</summary><sub>- 1 Hyundai H connector<br>- 1 RJ45 cable (7 ft)<br>- 1 comma 3X<br>- 1 comma power v2<br>- 1 harness box<br>- 1 mount<br>- 1 right angle OBD-C cable (1.5 ft)<br><a href="https://comma.ai/shop/comma-3x.html?make=Hyundai&model=Palisade 2020-22">Buy Here</a></sub></details>|<a href="https://youtu.be/TAnDqjF4fDY?t=456" target="_blank"><img height="18px" src="assets/icon-youtube.svg"></img></a>|

docs/css/tooltip.css

@@ -1,44 +0,0 @@
-[data-tooltip] {
-  position: relative;
-  display: inline-block;
-  border-bottom: 1px dotted black;
-}
-
-[data-tooltip] .tooltip-content {
-  width: max-content;
-  max-width: 25em;
-  position: absolute;
-  top: 100%;
-  left: 50%;
-  transform: translateX(-50%);
-  background-color: white;
-  color: #404040;
-  box-shadow: 0 4px 14px 0 rgba(0,0,0,.2), 0 0 0 1px rgba(0,0,0,.05);
-  padding: 10px;
-  font: 14px/1.5 Lato, proxima-nova, Helvetica Neue, Arial, sans-serif;
-  text-decoration: none;
-  opacity: 0;
-  visibility: hidden;
-  transition: opacity 0.1s, visibility 0s;
-  z-index: 1000;
-  pointer-events: none; /* Prevent accidental interaction */
-}
-
-[data-tooltip]:hover .tooltip-content {
-  opacity: 1;
-  visibility: visible;
-  pointer-events: auto; /* Allow interaction when visible */
-}
-
-.tooltip-content .tooltip-glossary-link {
-  display: inline-block;
-  margin-top: 8px;
-  font-size: 12px;
-  color: #007bff;
-  text-decoration: none;
-}
-
-.tooltip-content .tooltip-glossary-link:hover {
-  color: #0056b3;
-  text-decoration: underline;
-}

docs/hooks/glossary.py

@@ -1,68 +0,0 @@
-import re
-import tomllib
-
-def load_glossary(file_path="docs/glossary.toml"):
-  with open(file_path, "rb") as f:
-    glossary_data = tomllib.load(f)
-  return glossary_data.get("glossary", {})
-
-def generate_anchor_id(name):
-  return name.replace(" ", "-").replace("_", "-").lower()
-
-def format_markdown_term(name, definition):
-  anchor_id = generate_anchor_id(name)
-  markdown = f"* [**{name.replace('_', ' ').title()}**](#{anchor_id})"
-  if definition.get("abbreviation"):
-    markdown += f" *({definition['abbreviation']})*"
-  if definition.get("description"):
-    markdown += f": {definition['description']}\n"
-  return markdown
-
-def glossary_markdown(vocabulary):
-  markdown = ""
-  for category, terms in vocabulary.items():
-    markdown += f"## {category.replace('_', ' ').title()}\n\n"
-    for name, definition in terms.items():
-      markdown += format_markdown_term(name, definition)
-  return markdown
-
-def format_tooltip_html(term_key, definition, html):
-  display_term = term_key.replace("_", " ").title()
-  clean_description = re.sub(r"\[(.+)]\(.+\)", r"\1", definition["description"])
-  glossary_link = (
-    f"<a href='/concepts/glossary#{term_key}' class='tooltip-glossary-link' title='View in glossary'>Glossary🔗</a>"
-  )
-  return re.sub(
-    re.escape(display_term),
-    lambda
-    match: f"<span data-tooltip>{match.group(0)}<span class='tooltip-content'>{clean_description} {glossary_link}</span></span>",
-    html,
-    flags=re.IGNORECASE,
-  )
-
-def apply_tooltip(_term_key, _definition, pattern, html):
-  return re.sub(
-    pattern,
-    lambda match: format_tooltip_html(_term_key, _definition, match.group(0)),
-    html,
-    flags=re.IGNORECASE,
-  )
-
-def tooltip_html(vocabulary, html):
-  for _category, terms in vocabulary.items():
-    for term_key, definition in terms.items():
-      if definition.get("description"):
-        pattern = rf"(?<!\w){re.escape(term_key.replace('_', ' ').title())}(?![^<]*<\/a>)(?!\([^)]*\))"
-        html = apply_tooltip(term_key, definition, pattern, html)
-  return html
-
-# Page Hooks
-def on_page_markdown(markdown, **kwargs):
-  glossary = load_glossary()
-  return markdown.replace("{{GLOSSARY_DEFINITIONS}}", glossary_markdown(glossary))
-
-def on_page_content(html, **kwargs):
-  if kwargs.get("page").title == "Glossary":
-    return html
-  glossary = load_glossary()
-  return tooltip_html(glossary, html)

launch_env.sh

@@ -7,7 +7,7 @@ export OPENBLAS_NUM_THREADS=1
 export VECLIB_MAXIMUM_THREADS=1
 
 if [ -z "$AGNOS_VERSION" ]; then
-  export AGNOS_VERSION="11.4"
+  export AGNOS_VERSION="11.3"
 fi
 
 export STAGING_ROOT="/data/safe_staging"

mkdocs.yml

@@ -8,10 +8,6 @@ strict: true
 docs_dir: docs
 site_dir: docs_site/
 
-hooks:
-  - docs/hooks/glossary.py
-extra_css:
-  - css/tooltip.css
 theme:
   name: readthedocs
   navigation_depth: 3

openpilot/sunnypilot

@@ -1 +0,0 @@
-../sunnypilot

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "openpilot"
-requires-python = ">= 3.11, < 3.13"
+requires-python = ">= 3.11, <= 3.12"
 license = {text = "MIT License"}
 version = "0.1.0"
 description = "an open source driver assistance system"
@@ -42,7 +42,8 @@ dependencies = [
 
   # modeld
   "onnx >= 1.14.0",
-  "onnxruntime >=1.16.3",
+  "onnxruntime >=1.16.3; platform_system == 'Linux' and platform_machine == 'aarch64'",
+  "onnxruntime-gpu >=1.16.3; platform_system == 'Linux' and platform_machine == 'x86_64'",
 
   # logging
   "pyzmq",

release/ci/install_github_runner.sh

@@ -1,147 +0,0 @@
-#!/usr/bin/env bash
-set -e
-
-# Default values
-DEFAULT_REPO_URL="https://github.com/sunnypilot"
-START_AT_BOOT=false
-
-# Parse command line arguments
-while [[ $# -gt 0 ]]; do
-    case $1 in
-        --start-at-boot)
-            START_AT_BOOT=true
-            shift
-            ;;
-        --token)
-            GITHUB_TOKEN="$2"
-            shift 2
-            ;;
-        --repo)
-            REPO_URL="$2"
-            shift 2
-            ;;
-        *)
-            if [ -z "$GITHUB_TOKEN" ]; then
-                GITHUB_TOKEN="$1"
-            elif [ -z "$REPO_URL" ]; then
-                REPO_URL="$1"
-            fi
-            shift
-            ;;
-    esac
-done
-
-# Check required arguments
-if [ -z "$GITHUB_TOKEN" ]; then
-    echo "Usage: $0 [--start-at-boot] [--token <github_token>] [--repo <repository_url>]"
-    echo "Required argument: github_token"
-    echo "Optional arguments:"
-    echo "  --start-at-boot    Enable auto-start at boot (default: false)"
-    echo "  --repo            Repository URL (default: ${DEFAULT_REPO_URL})"
-    exit 1
-fi
-
-# Set repository URL if not provided
-REPO_URL="${REPO_URL:-$DEFAULT_REPO_URL}"
-
-# Determine BASE_DIR based on mount point
-if mountpoint -q /data/media; then
-    BASE_DIR="/data/media/0/github"
-else
-    BASE_DIR="/data/github"
-fi
-
-# Constants
-RUNNER_USER="github-runner"
-USER_GROUPS="comma,gpu,gpio,sudo"
-RUNNER_DIR="${BASE_DIR}/runner"
-BUILDS_DIR="${BASE_DIR}/builds"
-LOGS_DIR="${BASE_DIR}/logs"
-CACHE_DIR="${BASE_DIR}/cache"
-OPENPILOT_DIR="${BASE_DIR}/openpilot"
-
-create_directories() {
-    sudo mkdir -p "$RUNNER_DIR" "$BUILDS_DIR" "$LOGS_DIR" "$CACHE_DIR" "$OPENPILOT_DIR"
-    mkdir -p "/data/openpilot"
-    sudo chown -R comma:comma "/data/openpilot"
-}
-
-download_and_setup_runner() {
-    cd "$RUNNER_DIR"
-    curl -o actions-runner-linux-arm64-2.321.0.tar.gz -L https://github.com/actions/runner/releases/download/v2.321.0/actions-runner-linux-arm64-2.321.0.tar.gz
-    tar xzf ./actions-runner-linux-arm64-2.321.0.tar.gz
-    rm ./actions-runner-linux-arm64-2.321.0.tar.gz
-    chmod +x ./config.sh
-}
-
-setup_runner_user() {
-    sudo useradd --comment 'GitHub Runner' --create-home --home-dir ${BASE_DIR} ${RUNNER_USER} --shell /bin/bash -G ${USER_GROUPS} || sudo usermod -aG ${USER_GROUPS} ${RUNNER_USER}
-    export BASE_DIR
-    sudo -u ${RUNNER_USER} bash -c "truncate -s 0 '${BASE_DIR}/.bash_logout'"
-}
-
-create_sudoers_entry() {
-    sudo grep -qxF "${RUNNER_USER} ALL=(ALL) NOPASSWD: ALL" /etc/sudoers || echo "${RUNNER_USER} ALL=(ALL) NOPASSWD: ALL" | sudo tee -a /etc/sudoers
-}
-
-configure_runner() {
-    cd "$RUNNER_DIR"
-    sudo -u ${RUNNER_USER} ./config.sh --url "$REPO_URL" --token "$GITHUB_TOKEN" --name $(hostname) --runnergroup "tici-tizi" --labels "tici" --work "$BUILDS_DIR" --unattended
-}
-
-set_directory_permissions() {
-    sudo chown -R ${RUNNER_USER}:comma "$BASE_DIR"
-    sudo chmod g+rwx "$BASE_DIR"
-    sudo chmod g+s "$BASE_DIR"
-}
-
-modify_service_template() {
-    cat <<EOL > "$RUNNER_DIR/bin/actions.runner.service.template"
-[Unit]
-Description={{Description}}
-After=network-online.target nss-lookup.target time-sync.target
-Wants=network-online.target nss-lookup.target time-sync.target
-StartLimitInterval=5
-StartLimitBurst=10
-
-[Service]
-Type=simple
-User=root
-ExecStart=/usr/bin/unshare -m -- /bin/bash -c 'mount --bind ${OPENPILOT_DIR} /data/openpilot && setpriv --reuid={{User}} --regid={{User}} --init-groups env HOME=${BASE_DIR} USER={{User}} LOGNAME={{User}} MAIL=/var/mail/{{User}} {{RunnerRoot}}/runsvc.sh'
-WorkingDirectory={{RunnerRoot}}
-KillMode=process
-KillSignal=SIGTERM
-TimeoutStopSec=5min
-Restart=always
-RestartSec=120
-
-[Install]
-WantedBy=multi-user.target
-EOL
-}
-
-# Make filesystem writable
-sudo mount -o remount,rw /
-
-# Ensure filesystem is remounted as read-only on script exit
-trap "sudo mount -o remount,ro /" EXIT
-
-# Execute installation steps
-setup_runner_user
-create_sudoers_entry
-create_directories
-download_and_setup_runner
-modify_service_template
-configure_runner
-set_directory_permissions
-
-# Install and start service using built-in installer
-cd "$RUNNER_DIR"
-sudo ./svc.sh install $RUNNER_USER
-
-# Handle auto-start configuration
-if [ "$START_AT_BOOT" = false ]; then
-    sudo systemctl disable actions.runner.sunnypilot.$(uname -n)
-fi
-
-sudo ./svc.sh start

release/ci/publish.sh

@@ -1,78 +0,0 @@
-#!/usr/bin/env bash
-
-set -e
-
-DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" >/dev/null && pwd)"
-cd $DIR
-
-# Take parameters as arguments
-SOURCE_DIR=$1
-OUTPUT_DIR=$2
-DEV_BRANCH=$3
-VERSION=$4
-GIT_ORIGIN=$5
-EXTRA_VERSION_IDENTIFIER=$6
-
-# Check parameters
-if [ -z "$SOURCE_DIR" ] || [ -z "$OUTPUT_DIR" ]; then
-    echo "Error: No source or output directory provided."
-    exit 1
-fi
-
-if [ -z "$DEV_BRANCH" ] || [ -z "$VERSION" ]; then
-    echo "Error: No dev branch or version provided."
-    exit 1
-fi
-
-if [ -z "$GIT_ORIGIN" ]; then
-    echo "Error: No GIT_ORIGIN provided"
-    exit 1
-fi
-
-# "Tagging"
-echo "#define COMMA_VERSION \"$VERSION\"" > ${OUTPUT_DIR}/common/version.h
-
-## set git identity
-#source $DIR/identity.sh
-#export GIT_SSH_COMMAND="ssh -i /data/gitkey"
-
-echo "[-] Setting up repo T=$SECONDS"
-cd $OUTPUT_DIR
-git init
-
-# set git username/password
-#source /data/identity.sh
-
-git rm -rf $OUTPUT_DIR/.git || true # Doing cleanup, but it might fail if the .git doesn't exist or not allowed to delete
-git remote remove origin || true # ensure cleanup
-git remote add origin $GIT_ORIGIN
-#git push origin -d $DEV_BRANCH || true # Ensuring we delete the remote branch if it exists as we are wiping it out
-git fetch origin $DEV_BRANCH || (git checkout -b $DEV_BRANCH && git commit --allow-empty -m "sunnypilot v$VERSION release" && git push -u origin $DEV_BRANCH)
-
-echo "[-] committing version $VERSION T=$SECONDS"
-git add -f .
-git commit -a -m "sunnypilot v$VERSION release"
-git branch --set-upstream-to=origin/$DEV_BRANCH
-
-# include source commit hash and build date in commit
-GIT_HASH=$(git --git-dir=$SOURCE_DIR/.git rev-parse HEAD)
-DATETIME=$(date '+%Y-%m-%dT%H:%M:%S')
-SP_VERSION=$(cat $SOURCE_DIR/common/version.h | awk -F\" '{print $2}')
-
-# Add built files to git
-git add -f .
-if [ "$EXTRA_VERSION_IDENTIFIER" = "-release" ] || [ "$EXTRA_VERSION_IDENTIFIER" = "-staging" ]; then
-  export VERSION=${VERSION%"$EXTRA_VERSION_IDENTIFIER"}
-  git commit --amend -m "sunnypilot v$VERSION"
-else
-  git commit --amend -m "sunnypilot v$VERSION
-  version: sunnypilot v$SP_VERSION release
-  date: $DATETIME
-  master commit: $GIT_HASH
-  "
-fi
-git branch -m $DEV_BRANCH
-
-# Push!
-echo "[-] pushing T=$SECONDS"
-git push -f origin $DEV_BRANCH

release/ci/uninstall_github_runner.sh

@@ -1,66 +0,0 @@
-#!/usr/bin/env bash
-# Determine BASE_DIR based on mount point
-if mountpoint -q /data/media; then
-    GITHUB_BASE_DIR="/data/media/0/github"
-else
-    GITHUB_BASE_DIR="/data/github"
-fi
-
-# Define directories and user
-BIN_DIR="$GITHUB_BASE_DIR/bin"
-BUILDS_DIR="$GITHUB_BASE_DIR/builds"
-OPENPILOT_DIR="$GITHUB_BASE_DIR/openpilot"
-LOGS_DIR="$GITHUB_BASE_DIR/logs"
-CACHE_DIR="$GITHUB_BASE_DIR/cache"
-RUNNER_USERNAME="github-runner"
-# Define the systemd service name
-SERVICE_NAME="github-runner"
-USER_GROUPS="comma,gpu,gpio,sudo"
-
-# Function to stop and disable the systemd service
-stop_and_uninstall_service() {
-    cd $GITHUB_BASE_DIR/runner
-    sudo ./svc.sh stop
-    sudo ./svc.sh uninstall
-}
-
-# Function to remove the systemd service file
-remove_runner() {
-    cd $GITHUB_BASE_DIR/runner
-    sudo rm .runner
-    sudo su -c './config.sh remove' github-runner
-}
-
-# Function to delete the Github Runner directories
-delete_directories() {
-    sudo rm -rf "$BIN_DIR/github-runner"
-    sudo rm -rf "$GITHUB_BASE_DIR" "$BIN_DIR" "$BUILDS_DIR" "$LOGS_DIR" "$CACHE_DIR" "$OPENPILOT_DIR"
-}
-
-# Function to remove the Github Runner user
-delete_user() {
-    for group in ${USER_GROUPS//,/ }
-    do
-       sudo gpasswd -d ${RUNNER_USERNAME} ${group}
-    done
-    sudo userdel -r ${RUNNER_USERNAME}
-}
-
-# Function to remove sudoers entry
-remove_sudoers_entry() {
-    sudo sed -i.bak "/${RUNNER_USERNAME} ALL=(ALL) NOPASSWD: ALL/d" /etc/sudoers
-}
-
-# Make filesystem writable
-sudo mount -o remount rw /
-
-# Ensure filesystem is remounted as read-only on script exit
-trap "sudo mount -o remount ro /" EXIT
-
-# Call functions
-stop_and_uninstall_service
-remove_runner
-delete_directories
-delete_user
-remove_sudoers_entry
-# End of uninstall script

release/release_files.py

@@ -32,7 +32,6 @@ blacklist = [
 
   ".git/",
   ".github/",
-  ".devcontainer/",
   "Darwin/",
   ".vscode",
 
@@ -48,42 +47,6 @@ blacklist = [
   ".gitmodules",
 ]
 
-# Sunnypilot blacklist
-sunnypilot_blacklist = [
-  "system/loggerd/sunnylink_uploader.py",  # Temporarily, until we are ready to roll it out widely
-  ".idea/",
-  ".run/",
-  ".*__pycache__/.*",
-  ".*\\.pyc",
-  "teleoprtc/*",
-  "third_party/snpe/x86_64/*",
-  "body/board/canloader.py",
-  "body/board/flash_base.sh",
-  "body/board/flash_knee.sh",
-  "body/board/recover.sh",
-  ".*/test/",
-  ".*/tests/",
-  ".*tinygrad_repo/tinygrad/renderer/",
-  "README.md",
-  ".*internal/",
-  "docs/.*",
-  ".sconsign.dblite",
-  "release/ci/scons_cache/",
-  ".gitlab-ci.yml",
-  ".clang-tidy",
-  ".dockerignore",
-  ".editorconfig",
-  ".python-version",
-  "SECURITY.md",
-  "codecov.yml",
-  "conftest.py",
-  "poetry.lock",
-  ".venv/",
-]
-
-# Merge the blacklists
-blacklist += sunnypilot_blacklist
-
 # gets you through the blacklist
 whitelist = [
   "tools/lib/",
@@ -91,7 +54,7 @@ whitelist = [
   "tools/joystick/",
   "tools/longitudinal_maneuvers/",
 
-  "tinygrad_repo/examples/openpilot/compile3.py",
+  "tinygrad_repo/openpilot/compile2.py",
   "tinygrad_repo/extra/onnx.py",
   "tinygrad_repo/extra/onnx_ops.py",
   "tinygrad_repo/extra/thneed.py",
@@ -156,45 +119,8 @@ whitelist = [
   "opendbc_repo/dbc/toyota_tss2_adas.dbc",
   "opendbc_repo/dbc/vw_golf_mk4.dbc",
   "opendbc_repo/dbc/vw_mqb_2010.dbc",
-  "opendbc_repo/dbc/tesla_can.dbc",
-  "opendbc_repo/dbc/tesla_radar_bosch_generated.dbc",
-  "opendbc_repo/dbc/tesla_radar_continental_generated.dbc",
-  "opendbc_repo/dbc/tesla_powertrain.dbc",
 ]
 
-# Sunnypilot whitelist
-sunnypilot_whitelist = [
-  "^README.md",
-  ".*selfdrive/test/fuzzy_generation.py",
-  ".*selfdrive/test/helpers.py",
-  ".*selfdrive/test/__init__.py",
-  ".*selfdrive/test/setup_device_ci.sh",
-  ".*selfdrive/test/test_time_to_onroad.py",
-  ".*selfdrive/test/test_onroad.py",
-  ".*system/manager/test/test_manager.py",
-  ".*system/manager/test/__init__.py",
-  ".*system/qcomgpsd/tests/test_qcomgpsd.py",
-  ".*system/updated/casync/tests/test_casync.py",
-  ".*system/updated/tests/test_git.py",
-  ".*system/updated/tests/test_base.py",
-  ".*selfdrive/ui/tests/test_translations.py",
-  ".*selfdrive/car/tests/__init__.py",
-  ".*selfdrive/car/tests/test_car_interfaces.py",
-  ".*selfdrive/navd/tests/test_navd.py",
-  ".*selfdrive/navd/tests/test_map_renderer.py",
-  ".*selfdrive/boardd/tests/test_boardd_loopback.py",
-  ".*INTEGRATION.md",
-  ".*HOW-TOS.md",
-  ".*CARS.md",
-  ".*LIMITATIONS.md",
-  ".*CONTRIBUTING.md",
-  ".*sunnyhaibin0850_qrcode_paypal.me.png",
-  "opendbc/.*.dbc",
-]
-
-# Merge the whitelists
-whitelist += sunnypilot_whitelist
-
 
 if __name__ == "__main__":
   for f in Path(ROOT).rglob("**/*"):

scripts/lint/lint.sh

@@ -13,7 +13,7 @@ cd $ROOT
 
 FAILED=0
 
-IGNORED_FILES="uv\.lock|docs\/CARS.md|LICENSE\.md|.*\.zst"
+IGNORED_FILES="uv\.lock|docs\/CARS.md"
 IGNORED_DIRS="^third_party.*|^msgq.*|^msgq_repo.*|^opendbc.*|^opendbc_repo.*|^cereal.*|^panda.*|^rednose.*|^rednose_repo.*|^tinygrad.*|^tinygrad_repo.*|^teleoprtc.*|^teleoprtc_repo.*"
 
 function run() {

selfdrive/car/car_specific.py

@@ -148,8 +148,7 @@ class CarSpecificEvents:
       # To avoid re-engaging when openpilot cancels, check user engagement intention via buttons
       # Main button also can trigger an engagement on these cars
       self.cruise_buttons.append(any(ev.type in HYUNDAI_ENABLE_BUTTONS for ev in CS.buttonEvents))
-      events = self.create_common_events(CS, CS_prev, extra_gears=(GearShifter.sport, GearShifter.manumatic),
-                                         pcm_enable=self.CP.pcmCruise, allow_enable=any(self.cruise_buttons))
+      events = self.create_common_events(CS, CS_prev, pcm_enable=self.CP.pcmCruise, allow_enable=any(self.cruise_buttons))
 
       # low speed steer alert hysteresis logic (only for cars with steer cut off above 10 m/s)
       if CS.vEgo < (self.CP.minSteerSpeed + 2.) and self.CP.minSteerSpeed > 10.:

selfdrive/controls/lib/drive_helpers.py

@@ -5,15 +5,12 @@ from openpilot.common.realtime import DT_CTRL
 MIN_SPEED = 1.0
 CONTROL_N = 17
 CAR_ROTATION_RADIUS = 0.0
-# This is a turn radius smaller than most cars can achieve
-MAX_CURVATURE = 0.2
 
 # EU guidelines
 MAX_LATERAL_JERK = 5.0
 MAX_VEL_ERR = 5.0
 
 def clip_curvature(v_ego, prev_curvature, new_curvature):
-  new_curvature = clip(new_curvature, -MAX_CURVATURE, MAX_CURVATURE)
   v_ego = max(MIN_SPEED, v_ego)
   max_curvature_rate = MAX_LATERAL_JERK / (v_ego**2) # inexact calculation, check https://github.com/commaai/openpilot/pull/24755
   safe_desired_curvature = clip(new_curvature,

selfdrive/controls/lib/longitudinal_planner.py

@@ -50,20 +50,24 @@ def limit_accel_in_turns(v_ego, angle_steers, a_target, CP):
   return [a_target[0], min(a_target[1], a_x_allowed)]
 
 
-def get_accel_from_plan(speeds, accels, action_t=DT_MDL, vEgoStopping=0.05):
+def get_accel_from_plan(CP, speeds, accels):
   if len(speeds) == CONTROL_N:
-    v_now = speeds[0]
-    a_now = accels[0]
+    v_target_now = interp(DT_MDL, CONTROL_N_T_IDX, speeds)
+    a_target_now = interp(DT_MDL, CONTROL_N_T_IDX, accels)
 
-    v_target = interp(action_t, CONTROL_N_T_IDX, speeds)
-    a_target = 2 * (v_target - v_now) / (action_t) - a_now
-    v_target_1sec = interp(action_t + 1.0, CONTROL_N_T_IDX, speeds)
+    v_target = interp(CP.longitudinalActuatorDelay + DT_MDL, CONTROL_N_T_IDX, speeds)
+    if v_target != v_target_now:
+      a_target = 2 * (v_target - v_target_now) / CP.longitudinalActuatorDelay - a_target_now
+    else:
+      a_target = a_target_now
+
+    v_target_1sec = interp(CP.longitudinalActuatorDelay + DT_MDL + 1.0, CONTROL_N_T_IDX, speeds)
   else:
     v_target = 0.0
     v_target_1sec = 0.0
     a_target = 0.0
-  should_stop = (v_target < vEgoStopping and
-                 v_target_1sec < vEgoStopping)
+  should_stop = (v_target < CP.vEgoStopping and
+                 v_target_1sec < CP.vEgoStopping)
   return a_target, should_stop
 
 
@@ -197,9 +201,7 @@ class LongitudinalPlanner:
     longitudinalPlan.longitudinalPlanSource = self.mpc.source
     longitudinalPlan.fcw = self.fcw
 
-    action_t =  self.CP.longitudinalActuatorDelay + DT_MDL
-    a_target, should_stop = get_accel_from_plan(longitudinalPlan.speeds, longitudinalPlan.accels,
-                                                action_t=action_t, vEgoStopping=self.CP.vEgoStopping)
+    a_target, should_stop = get_accel_from_plan(self.CP, longitudinalPlan.speeds, longitudinalPlan.accels)
     longitudinalPlan.aTarget = a_target
     longitudinalPlan.shouldStop = should_stop
     longitudinalPlan.allowBrake = True

selfdrive/debug/touch_replay.py

@@ -1,54 +0,0 @@
-#!/usr/bin/env python3
-import argparse
-
-import numpy as np
-import matplotlib.pyplot as plt
-
-from openpilot.tools.lib.logreader import LogReader
-
-if __name__ == '__main__':
-  parser = argparse.ArgumentParser()
-  parser.add_argument('--width', default=2160, type=int)
-  parser.add_argument('--height', default=1080, type=int)
-  parser.add_argument('--route', default='rlog', type=str)
-  args = parser.parse_args()
-
-  w = args.width
-  h = args.height
-  route = args.route
-
-  fingers = [[-1, -1]] * 5
-  touch_points = []
-  current_slot = 0
-
-  lr = list(LogReader(route))
-  for msg in lr:
-    if msg.which() == 'touch':
-      for event in msg.touch:
-        if event.type == 3 and event.code == 47:
-          current_slot = event.value
-        elif event.type == 3 and event.code == 57 and event.value == -1:
-          fingers[current_slot] = [-1, -1]
-        elif event.type == 3 and event.code == 53:
-          fingers[current_slot][1] = h - (h - event.value)
-          if fingers[current_slot][0] != -1:
-            touch_points.append(fingers[current_slot].copy())
-        elif event.type == 3 and event.code == 54:
-          fingers[current_slot][0] = w - event.value
-          if fingers[current_slot][1] != -1:
-            touch_points.append(fingers[current_slot].copy())
-
-  if not touch_points:
-    print(f'No touch events found for {route}')
-    quit()
-
-  unique_points, counts = np.unique(touch_points, axis=0, return_counts=True)
-
-  plt.figure(figsize=(10, 3))
-  plt.scatter(unique_points[:, 0], unique_points[:, 1], c=counts, s=counts * 20, edgecolors='red')
-  plt.colorbar()
-  plt.title(f'Touches for {route}')
-  plt.xlim(0, w)
-  plt.ylim(0, h)
-  plt.grid(True)
-  plt.show()

selfdrive/modeld/SConscript

@@ -13,6 +13,20 @@ common_src = [
   "transforms/transform.cc",
 ]
 
+thneed_src_common = [
+  "thneed/thneed_common.cc",
+  "thneed/serialize.cc",
+]
+
+thneed_src_qcom = thneed_src_common + ["thneed/thneed_qcom2.cc"]
+thneed_src_pc = thneed_src_common + ["thneed/thneed_pc.cc"]
+thneed_src = thneed_src_qcom if arch == "larch64" else thneed_src_pc
+
+# SNPE except on Mac and ARM Linux
+snpe_lib = []
+if arch != "Darwin" and arch != "aarch64":
+  common_src += ['runners/snpemodel.cc']
+  snpe_lib += ['SNPE']
 
 # OpenCL is a framework on Mac
 if arch == "Darwin":
@@ -31,24 +45,34 @@ snpe_rpath_pc = f"{Dir('#').abspath}/third_party/snpe/x86_64-linux-clang"
 snpe_rpath = lenvCython['RPATH'] + [snpe_rpath_qcom if arch == "larch64" else snpe_rpath_pc]
 
 cython_libs = envCython["LIBS"] + libs
+snpemodel_lib = lenv.Library('snpemodel', ['runners/snpemodel.cc'])
 commonmodel_lib = lenv.Library('commonmodel', common_src)
+
+lenvCython.Program('runners/runmodel_pyx.so', 'runners/runmodel_pyx.pyx', LIBS=cython_libs, FRAMEWORKS=frameworks)
+lenvCython.Program('runners/snpemodel_pyx.so', 'runners/snpemodel_pyx.pyx', LIBS=[snpemodel_lib, snpe_lib, *cython_libs], FRAMEWORKS=frameworks, RPATH=snpe_rpath)
 lenvCython.Program('models/commonmodel_pyx.so', 'models/commonmodel_pyx.pyx', LIBS=[commonmodel_lib, *cython_libs], FRAMEWORKS=frameworks)
-tinygrad_files = ["#"+x for x in glob.glob(env.Dir("#tinygrad_repo").relpath + "/**", recursive=True, root_dir=env.Dir("#").abspath) if 'pycache' not in x]
+
+tinygrad_files = ["#"+x for x in glob.glob(env.Dir("#tinygrad_repo").relpath + "/**", recursive=True, root_dir=env.Dir("#").abspath)]
 
 # Get model metadata
 fn = File("models/supercombo").abspath
 cmd = f'python3 {Dir("#selfdrive/modeld").abspath}/get_model_metadata.py {fn}.onnx'
 lenv.Command(fn + "_metadata.pkl", [fn + ".onnx"] + tinygrad_files, cmd)
 
-# Compile tinygrad model
-pythonpath_string = 'PYTHONPATH="${PYTHONPATH}:' + env.Dir("#tinygrad_repo").abspath + '"'
-if arch == 'larch64':
-  device_string = 'QCOM=1'
-else:
-  device_string = 'CLANG=1 IMAGE=0'
+# Build thneed model
+if arch == "larch64" or GetOption('pc_thneed'):
+  tinygrad_opts = []
+  if not GetOption('pc_thneed'):
+    # use FLOAT16 on device for speed + don't cache the CL kernels for space
+    tinygrad_opts += ["FLOAT16=1", "PYOPENCL_NO_CACHE=1"]
+  cmd = f"cd {Dir('#').abspath}/tinygrad_repo && " + ' '.join(tinygrad_opts) + f" python3 openpilot/compile2.py {fn}.onnx {fn}.thneed"
 
-for model_name in ['supercombo', 'dmonitoring_model']:
-  fn = File(f"models/{model_name}").abspath
-  cmd = f'{pythonpath_string} {device_string} python3 {Dir("#tinygrad_repo").abspath}/examples/openpilot/compile3.py {fn}.onnx {fn}_tinygrad.pkl'
-  lenv.Command(fn + "_tinygrad.pkl", [fn + ".onnx"] + tinygrad_files, cmd)
+  lenv.Command(fn + ".thneed", [fn + ".onnx"] + tinygrad_files, cmd)
 
+  fn_dm = File("models/dmonitoring_model").abspath
+  cmd = f"cd {Dir('#').abspath}/tinygrad_repo && " + ' '.join(tinygrad_opts) + f" python3 openpilot/compile2.py {fn_dm}.onnx {fn_dm}.thneed"
+  lenv.Command(fn_dm + ".thneed", [fn_dm + ".onnx"] + tinygrad_files, cmd)
+
+  thneed_lib = env.SharedLibrary('thneed', thneed_src, LIBS=[gpucommon, common, 'OpenCL', 'dl'])
+  thneedmodel_lib = env.Library('thneedmodel', ['runners/thneedmodel.cc'])
+  lenvCython.Program('runners/thneedmodel_pyx.so', 'runners/thneedmodel_pyx.pyx', LIBS=envCython["LIBS"]+[thneedmodel_lib, thneed_lib, gpucommon, common, 'dl', 'OpenCL'])

selfdrive/modeld/dmonitoringmodeld

@@ -1,4 +1,10 @@
 #!/usr/bin/env bash
 
 DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" >/dev/null && pwd)"
+cd "$DIR/../../"
+
+if [ -f "$DIR/libthneed.so" ]; then
+  export LD_PRELOAD="$DIR/libthneed.so"
+fi
+
 exec "$DIR/dmonitoringmodeld.py" "$@"

selfdrive/modeld/dmonitoringmodeld.py

@@ -1,17 +1,8 @@
 #!/usr/bin/env python3
 import os
-from openpilot.system.hardware import TICI
-if TICI:
-  from tinygrad.tensor import Tensor
-  from tinygrad.dtype import dtypes
-  from openpilot.selfdrive.modeld.runners.tinygrad_helpers import qcom_tensor_from_opencl_address
-  os.environ['QCOM'] = '1'
-else:
-  from openpilot.selfdrive.modeld.runners.ort_helpers import make_onnx_cpu_runner
 import gc
 import math
 import time
-import pickle
 import ctypes
 import numpy as np
 from pathlib import Path
@@ -22,20 +13,21 @@ from cereal.messaging import PubMaster, SubMaster
 from msgq.visionipc import VisionIpcClient, VisionStreamType, VisionBuf
 from openpilot.common.swaglog import cloudlog
 from openpilot.common.realtime import set_realtime_priority
-from openpilot.common.transformations.model import dmonitoringmodel_intrinsics, DM_INPUT_SIZE
-from openpilot.common.transformations.camera import _ar_ox_fisheye, _os_fisheye
-from openpilot.selfdrive.modeld.models.commonmodel_pyx import CLContext, MonitoringModelFrame
+from openpilot.selfdrive.modeld.runners import ModelRunner, Runtime
+from openpilot.selfdrive.modeld.models.commonmodel_pyx import CLContext
 from openpilot.selfdrive.modeld.parse_model_outputs import sigmoid
 
-MODEL_WIDTH, MODEL_HEIGHT = DM_INPUT_SIZE
 CALIB_LEN = 3
+MODEL_WIDTH = 1440
+MODEL_HEIGHT = 960
 FEATURE_LEN = 512
 OUTPUT_SIZE = 84 + FEATURE_LEN
 
 PROCESS_NAME = "selfdrive.modeld.dmonitoringmodeld"
 SEND_RAW_PRED = os.getenv('SEND_RAW_PRED')
-MODEL_PATH = Path(__file__).parent / 'models/dmonitoring_model.onnx'
-MODEL_PKL_PATH = Path(__file__).parent / 'models/dmonitoring_model_tinygrad.pkl'
+MODEL_PATHS = {
+  ModelRunner.THNEED: Path(__file__).parent / 'models/dmonitoring_model.thneed',
+  ModelRunner.ONNX: Path(__file__).parent / 'models/dmonitoring_model.onnx'}
 
 class DriverStateResult(ctypes.Structure):
   _fields_ = [
@@ -66,42 +58,33 @@ class DMonitoringModelResult(ctypes.Structure):
 class ModelState:
   inputs: dict[str, np.ndarray]
   output: np.ndarray
+  model: ModelRunner
 
   def __init__(self, cl_ctx):
     assert ctypes.sizeof(DMonitoringModelResult) == OUTPUT_SIZE * ctypes.sizeof(ctypes.c_float)
+    self.output = np.zeros(OUTPUT_SIZE, dtype=np.float32)
+    self.inputs = {
+      'input_img': np.zeros(MODEL_HEIGHT * MODEL_WIDTH, dtype=np.uint8),
+      'calib': np.zeros(CALIB_LEN, dtype=np.float32)}
 
-    self.frame = MonitoringModelFrame(cl_ctx)
-    self.numpy_inputs = {
-      'calib': np.zeros((1, CALIB_LEN), dtype=np.float32),
-    }
+    self.model = ModelRunner(MODEL_PATHS, self.output, Runtime.GPU, False, cl_ctx)
+    self.model.addInput("input_img", None)
+    self.model.addInput("calib", self.inputs['calib'])
 
-    if TICI:
-      self.tensor_inputs = {k: Tensor(v, device='NPY').realize() for k,v in self.numpy_inputs.items()}
-      with open(MODEL_PKL_PATH, "rb") as f:
-        self.model_run = pickle.load(f)
-    else:
-      self.onnx_cpu_runner = make_onnx_cpu_runner(MODEL_PATH)
+  def run(self, buf:VisionBuf, calib:np.ndarray) -> tuple[np.ndarray, float]:
+    self.inputs['calib'][:] = calib
 
-  def run(self, buf:VisionBuf, calib:np.ndarray, transform:np.ndarray) -> tuple[np.ndarray, float]:
-    self.numpy_inputs['calib'][0,:] = calib
+    v_offset = buf.height - MODEL_HEIGHT
+    h_offset = (buf.width - MODEL_WIDTH) // 2
+    buf_data = buf.data.reshape(-1, buf.stride)
+    input_data = self.inputs['input_img'].reshape(MODEL_HEIGHT, MODEL_WIDTH)
+    input_data[:] = buf_data[v_offset:v_offset+MODEL_HEIGHT, h_offset:h_offset+MODEL_WIDTH]
 
+    self.model.setInputBuffer("input_img", self.inputs['input_img'].view(np.float32))
     t1 = time.perf_counter()
-
-    input_img_cl = self.frame.prepare(buf, transform.flatten())
-    if TICI:
-      # The imgs tensors are backed by opencl memory, only need init once
-      if 'input_img' not in self.tensor_inputs:
-        self.tensor_inputs['input_img'] = qcom_tensor_from_opencl_address(input_img_cl.mem_address, (1, MODEL_WIDTH*MODEL_HEIGHT), dtype=dtypes.uint8)
-    else:
-      self.numpy_inputs['input_img'] = self.frame.buffer_from_cl(input_img_cl).reshape((1, MODEL_WIDTH*MODEL_HEIGHT))
-
-    if TICI:
-      output = self.model_run(**self.tensor_inputs).numpy().flatten()
-    else:
-      output = self.onnx_cpu_runner.run(None, self.numpy_inputs)[0].flatten()
-
+    self.model.execute()
     t2 = time.perf_counter()
-    return output, t2 - t1
+    return self.output, t2 - t1
 
 
 def fill_driver_state(msg, ds_result: DriverStateResult):
@@ -154,23 +137,18 @@ def main():
   pm = PubMaster(["driverStateV2"])
 
   calib = np.zeros(CALIB_LEN, dtype=np.float32)
-  model_transform = None
 
   while True:
     buf = vipc_client.recv()
     if buf is None:
       continue
 
-    if model_transform is None:
-      cam = _os_fisheye if buf.width == _os_fisheye.width else _ar_ox_fisheye
-      model_transform = np.linalg.inv(np.dot(dmonitoringmodel_intrinsics, np.linalg.inv(cam.intrinsics))).astype(np.float32)
-
     sm.update(0)
     if sm.updated["liveCalibration"]:
       calib[:] = np.array(sm["liveCalibration"].rpyCalib)
 
     t1 = time.perf_counter()
-    model_output, gpu_execution_time = model.run(buf, calib, model_transform)
+    model_output, gpu_execution_time = model.run(buf, calib)
     t2 = time.perf_counter()
 
     pm.send("driverStateV2", get_driverstate_packet(model_output, vipc_client.frame_id, vipc_client.timestamp_sof, t2 - t1, gpu_execution_time))

selfdrive/modeld/fill_model_msg.py

@@ -3,22 +3,11 @@ import capnp
 import numpy as np
 from cereal import log
 from openpilot.selfdrive.modeld.constants import ModelConstants, Plan, Meta
-from openpilot.selfdrive.controls.lib.drive_helpers import MIN_SPEED
 
 SEND_RAW_PRED = os.getenv('SEND_RAW_PRED')
 
 ConfidenceClass = log.ModelDataV2.ConfidenceClass
 
-def curv_from_psis(psi_target, psi_rate, vego, delay):
-  vego = np.clip(vego, MIN_SPEED, np.inf)
-  curv_from_psi = psi_target / (vego * delay)  # epsilon to prevent divide-by-zero
-  return 2*curv_from_psi - psi_rate / vego
-
-def get_curvature_from_plan(plan, vego, delay):
-  psi_target = np.interp(delay, ModelConstants.T_IDXS, plan[:, Plan.T_FROM_CURRENT_EULER][:, 2])
-  psi_rate = plan[:, Plan.ORIENTATION_RATE][0, 2]
-  return curv_from_psis(psi_target, psi_rate, vego, delay)
-
 class PublishState:
   def __init__(self):
     self.disengage_buffer = np.zeros(ModelConstants.CONFIDENCE_BUFFER_LEN*ModelConstants.DISENGAGE_WIDTH, dtype=np.float32)
@@ -66,17 +55,14 @@ def fill_lane_line_meta(builder, lane_lines, lane_line_probs):
   builder.rightProb = lane_line_probs[2]
 
 def fill_model_msg(base_msg: capnp._DynamicStructBuilder, extended_msg: capnp._DynamicStructBuilder,
-                   net_output_data: dict[str, np.ndarray], v_ego: float, delay: float,
-                   publish_state: PublishState, vipc_frame_id: int, vipc_frame_id_extra: int,
-                   frame_id: int, frame_drop: float, timestamp_eof: int, model_execution_time: float,
-                   valid: bool) -> None:
+                   net_output_data: dict[str, np.ndarray], publish_state: PublishState,
+                   vipc_frame_id: int, vipc_frame_id_extra: int, frame_id: int, frame_drop: float,
+                   timestamp_eof: int, model_execution_time: float, valid: bool) -> None:
   frame_age = frame_id - vipc_frame_id if frame_id > vipc_frame_id else 0
   frame_drop_perc = frame_drop * 100
   extended_msg.valid = valid
   base_msg.valid = valid
 
-  desired_curv = float(get_curvature_from_plan(net_output_data['plan'][0], v_ego, delay))
-
   driving_model_data = base_msg.drivingModelData
 
   driving_model_data.frameId = vipc_frame_id
@@ -85,7 +71,7 @@ def fill_model_msg(base_msg: capnp._DynamicStructBuilder, extended_msg: capnp._D
   driving_model_data.modelExecutionTime = model_execution_time
 
   action = driving_model_data.action
-  action.desiredCurvature = desired_curv
+  action.desiredCurvature = float(net_output_data['desired_curvature'][0,0])
 
   modelV2 = extended_msg.modelV2
   modelV2.frameId = vipc_frame_id
@@ -120,7 +106,7 @@ def fill_model_msg(base_msg: capnp._DynamicStructBuilder, extended_msg: capnp._D
 
   # lateral planning
   action = modelV2.action
-  action.desiredCurvature = desired_curv
+  action.desiredCurvature = float(net_output_data['desired_curvature'][0,0])
 
   # times at X_IDXS according to model plan
   PLAN_T_IDXS = [np.nan] * ModelConstants.IDX_N

selfdrive/modeld/modeld

@@ -1,4 +1,10 @@
 #!/usr/bin/env bash
 
 DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" >/dev/null && pwd)"
+cd "$DIR/../../"
+
+if [ -f "$DIR/libthneed.so" ]; then
+  export LD_PRELOAD="$DIR/libthneed.so"
+fi
+
 exec "$DIR/modeld.py" "$@"

selfdrive/modeld/modeld.py

@@ -1,16 +1,5 @@
 #!/usr/bin/env python3
 import os
-from openpilot.system.hardware import TICI
-
-#
-USE_TINYGRAD = os.getenv('USE_TINYGRAD', True) or TICI
-if USE_TINYGRAD:
-  from tinygrad.tensor import Tensor
-  from tinygrad.dtype import dtypes
-  from openpilot.selfdrive.modeld.runners.tinygrad_helpers import qcom_tensor_from_opencl_address
-  os.environ['QCOM'] = '1'
-else:
-  from openpilot.selfdrive.modeld.runners.ort_helpers import make_onnx_cpu_runner, ORT_TYPES_TO_NP_TYPES
 import time
 import pickle
 import numpy as np
@@ -29,19 +18,22 @@ from openpilot.common.transformations.camera import DEVICE_CAMERAS
 from openpilot.common.transformations.model import get_warp_matrix
 from openpilot.system import sentry
 from openpilot.selfdrive.controls.lib.desire_helper import DesireHelper
+from openpilot.selfdrive.modeld.runners import ModelRunner, Runtime
 from openpilot.selfdrive.modeld.parse_model_outputs import Parser
 from openpilot.selfdrive.modeld.fill_model_msg import fill_model_msg, fill_pose_msg, PublishState
 from openpilot.selfdrive.modeld.constants import ModelConstants
-from openpilot.selfdrive.modeld.models.commonmodel_pyx import DrivingModelFrame, CLContext
-
+from openpilot.selfdrive.modeld.models.commonmodel_pyx import ModelFrame, CLContext
 
 PROCESS_NAME = "selfdrive.modeld.modeld"
 SEND_RAW_PRED = os.getenv('SEND_RAW_PRED')
 
-MODEL_PATH = Path(__file__).parent / 'models/supercombo.onnx'
-MODEL_PKL_PATH = Path(__file__).parent / 'models/supercombo_tinygrad.pkl'
+MODEL_PATHS = {
+  ModelRunner.THNEED: Path(__file__).parent / 'models/supercombo.thneed',
+  ModelRunner.ONNX: Path(__file__).parent / 'models/supercombo.onnx'}
+
 METADATA_PATH = Path(__file__).parent / 'models/supercombo_metadata.pkl'
 
+
 class FrameMeta:
   frame_id: int = 0
   timestamp_sof: int = 0
@@ -52,45 +44,43 @@ class FrameMeta:
       self.frame_id, self.timestamp_sof, self.timestamp_eof = vipc.frame_id, vipc.timestamp_sof, vipc.timestamp_eof
 
 class ModelState:
-  frames: dict[str, DrivingModelFrame]
+  frame: ModelFrame
+  wide_frame: ModelFrame
   inputs: dict[str, np.ndarray]
   output: np.ndarray
   prev_desire: np.ndarray  # for tracking the rising edge of the pulse
+  model: ModelRunner
 
   def __init__(self, context: CLContext):
-    self.frames = {'input_imgs': DrivingModelFrame(context), 'big_input_imgs': DrivingModelFrame(context)}
+    self.frame = ModelFrame(context)
+    self.wide_frame = ModelFrame(context)
     self.prev_desire = np.zeros(ModelConstants.DESIRE_LEN, dtype=np.float32)
     self.full_features_20Hz = np.zeros((ModelConstants.FULL_HISTORY_BUFFER_LEN, ModelConstants.FEATURE_LEN), dtype=np.float32)
-    self.desire_20Hz = np.zeros((ModelConstants.FULL_HISTORY_BUFFER_LEN + 1, ModelConstants.DESIRE_LEN), dtype=np.float32)
+    self.desire_20Hz =  np.zeros((ModelConstants.FULL_HISTORY_BUFFER_LEN + 1, ModelConstants.DESIRE_LEN), dtype=np.float32)
+    self.prev_desired_curv_20hz = np.zeros((ModelConstants.FULL_HISTORY_BUFFER_LEN + 1, ModelConstants.PREV_DESIRED_CURV_LEN), dtype=np.float32)
 
     # img buffers are managed in openCL transform code
-    self.numpy_inputs = {}
+    self.inputs = {
+      'desire': np.zeros(ModelConstants.DESIRE_LEN * (ModelConstants.HISTORY_BUFFER_LEN+1), dtype=np.float32),
+      'traffic_convention': np.zeros(ModelConstants.TRAFFIC_CONVENTION_LEN, dtype=np.float32),
+      'lateral_control_params': np.zeros(ModelConstants.LATERAL_CONTROL_PARAMS_LEN, dtype=np.float32),
+      'prev_desired_curv': np.zeros(ModelConstants.PREV_DESIRED_CURV_LEN * (ModelConstants.HISTORY_BUFFER_LEN+1), dtype=np.float32),
+      'features_buffer': np.zeros(ModelConstants.HISTORY_BUFFER_LEN * ModelConstants.FEATURE_LEN, dtype=np.float32),
+    }
 
     with open(METADATA_PATH, 'rb') as f:
       model_metadata = pickle.load(f)
-    self.input_shapes =  model_metadata['input_shapes']
-
-    for key, shape in self.input_shapes.items():
-      if key not in self.frames: # Managed by opencl
-        self.numpy_inputs[key] = np.zeros(shape, dtype=np.float32)
 
     self.output_slices = model_metadata['output_slices']
     net_output_size = model_metadata['output_shapes']['outputs'][1]
     self.output = np.zeros(net_output_size, dtype=np.float32)
     self.parser = Parser()
 
-    if USE_TINYGRAD:
-      self.tensor_inputs = {k: Tensor(v, device='NPY').realize() for k,v in self.numpy_inputs.items()}
-      with open(MODEL_PKL_PATH, "rb") as f:
-        self.model_run = pickle.load(f)
-    else:
-      self.onnx_cpu_runner = make_onnx_cpu_runner(MODEL_PATH)
-      self.onnx_model_metadata = {input.name: input.type for input in self.onnx_cpu_runner.get_inputs()}
-
-    num_elements = self.numpy_inputs['features_buffer'].shape[1]
-    step_size = int(-100 / num_elements)
-    self.full_features_20Hz_idxs = np.arange(step_size, step_size * (num_elements + 1), step_size)[::-1]
-    self.desire_reshape_dims = (self.numpy_inputs['desire'].shape[0], self.numpy_inputs['desire'].shape[1], -1, self.numpy_inputs['desire'].shape[2])
+    self.model = ModelRunner(MODEL_PATHS, self.output, Runtime.GPU, False, context)
+    self.model.addInput("input_imgs", None)
+    self.model.addInput("big_input_imgs", None)
+    for k,v in self.inputs.items():
+      self.model.addInput(k, v)
 
   def slice_outputs(self, model_outputs: np.ndarray) -> dict[str, np.ndarray]:
     parsed_model_outputs = {k: model_outputs[np.newaxis, v] for k,v in self.output_slices.items()}
@@ -107,50 +97,30 @@ class ModelState:
 
     self.desire_20Hz[:-1] = self.desire_20Hz[1:]
     self.desire_20Hz[-1] = new_desire
-    self.numpy_inputs['desire'][:] = self.desire_20Hz.reshape(self.desire_reshape_dims).max(axis=2)
+    self.inputs['desire'][:] = self.desire_20Hz.reshape((25,4,-1)).max(axis=1).flatten()
 
-    self.numpy_inputs['traffic_convention'][:] = inputs['traffic_convention']
-    imgs_cl = {'input_imgs': self.frames['input_imgs'].prepare(buf, transform.flatten()),
-               'big_input_imgs': self.frames['big_input_imgs'].prepare(wbuf, transform_wide.flatten())}
+    self.inputs['traffic_convention'][:] = inputs['traffic_convention']
+    self.inputs['lateral_control_params'][:] = inputs['lateral_control_params']
 
-    if USE_TINYGRAD:
-      # The imgs tensors are backed by opencl memory, only need init once
-      for key in imgs_cl:
-        if not TICI or key not in self.tensor_inputs:
-          index = self.model_run.captured.expected_names.index(key)
-          _, _, dtype, device = self.model_run.captured.expected_st_vars_dtype_device[index]
-          if TICI:
-            self.tensor_inputs[key] = qcom_tensor_from_opencl_address(imgs_cl[key].mem_address, self.input_shapes[key], dtype=dtype)
-          else:
-            shape = self.frames[key].buffer_from_cl(imgs_cl[key]).reshape(self.input_shapes[key])
-            self.tensor_inputs[key] = Tensor(shape, device=device, dtype=dtype).realize()
-    else:
-      for key in imgs_cl:
-        dtype = self.onnx_model_metadata[key]
-        self.numpy_inputs[key] = self.frames[key].buffer_from_cl(imgs_cl[key]).astype(ORT_TYPES_TO_NP_TYPES[dtype]).reshape(self.input_shapes[key])
+    self.model.setInputBuffer("input_imgs", self.frame.prepare(buf, transform.flatten(), self.model.getCLBuffer("input_imgs")))
+    self.model.setInputBuffer("big_input_imgs", self.wide_frame.prepare(wbuf, transform_wide.flatten(), self.model.getCLBuffer("big_input_imgs")))
 
     if prepare_only:
       return None
 
-    if USE_TINYGRAD:
-      self.output = self.model_run(**self.tensor_inputs).numpy().flatten()
-    else:
-      self.output = self.onnx_cpu_runner.run(None, self.numpy_inputs)[0].flatten()
-
+    self.model.execute()
     outputs = self.parser.parse_outputs(self.slice_outputs(self.output))
 
     self.full_features_20Hz[:-1] = self.full_features_20Hz[1:]
     self.full_features_20Hz[-1] = outputs['hidden_state'][0, :]
 
-    self.numpy_inputs['features_buffer'][:] = self.full_features_20Hz[self.full_features_20Hz_idxs]
-    if "desired_curvature" in outputs:
-      if "prev_desired_curvs" in self.numpy_inputs.keys():
-        self.numpy_inputs['prev_desired_curvs'][:-1] = self.numpy_inputs['prev_desired_curvs'][1:]
-        self.numpy_inputs['prev_desired_curvs'][-1] = outputs['desired_curvature'][:, 0:1, None]  # Reshape to (1,1,1)
-      if "prev_desired_curv" in self.numpy_inputs.keys():
-        # First shift everything
-        self.numpy_inputs['prev_desired_curv'][:-ModelConstants.PREV_DESIRED_CURV_LEN] = self.numpy_inputs['prev_desired_curv'][ModelConstants.PREV_DESIRED_CURV_LEN:]
-        self.numpy_inputs['prev_desired_curv'][-ModelConstants.PREV_DESIRED_CURV_LEN:] = outputs['desired_curvature'][:, :1].reshape(1, -1, 1)
+    self.prev_desired_curv_20hz[:-1] = self.prev_desired_curv_20hz[1:]
+    self.prev_desired_curv_20hz[-1] = outputs['desired_curvature'][0, :]
+
+    idxs = np.arange(-4,-100,-4)[::-1]
+    self.inputs['features_buffer'][:] = self.full_features_20Hz[idxs].flatten()
+    # TODO model only uses last value now, once that changes we need to input strided action history buffer
+    self.inputs['prev_desired_curv'][-ModelConstants.PREV_DESIRED_CURV_LEN:] = 0. * self.prev_desired_curv_20hz[-4, :]
     return outputs
 
 
@@ -261,10 +231,7 @@ def main(demo=False):
     is_rhd = sm["driverMonitoringState"].isRHD
     frame_id = sm["roadCameraState"].frameId
     v_ego = max(sm["carState"].vEgo, 0.)
-    lateral_control_params = None #TODO-SP: hardcoded ,this shouldnt' be here this way. We should do it more dynamically
-    if "lateral_control_params" in model.numpy_inputs.keys(): #TODO-SP: hardcoded ,this shouldnt' be here this way. We should do it more dynamically
-      lateral_control_params = np.array([sm["carState"].vEgo, steer_delay], dtype=np.float32)
-
+    lateral_control_params = np.array([v_ego, steer_delay], dtype=np.float32)
     if sm.updated["liveCalibration"] and sm.seen['roadCameraState'] and sm.seen['deviceState']:
       device_from_calib_euler = np.array(sm["liveCalibration"].rpyCalib, dtype=np.float32)
       dc = DEVICE_CAMERAS[(str(sm['deviceState'].deviceType), str(sm['roadCameraState'].sensor))]
@@ -295,9 +262,8 @@ def main(demo=False):
     inputs:dict[str, np.ndarray] = {
       'desire': vec_desire,
       'traffic_convention': traffic_convention,
+      'lateral_control_params': lateral_control_params,
       }
-    if "lateral_control_params" in model.numpy_inputs.keys():
-      inputs['lateral_control_params'] = lateral_control_params
 
     mt1 = time.perf_counter()
     model_output = model.run(buf_main, buf_extra, model_transform_main, model_transform_extra, inputs, prepare_only)
@@ -308,8 +274,7 @@ def main(demo=False):
       modelv2_send = messaging.new_message('modelV2')
       drivingdata_send = messaging.new_message('drivingModelData')
       posenet_send = messaging.new_message('cameraOdometry')
-      fill_model_msg(drivingdata_send, modelv2_send, model_output, v_ego, steer_delay,
-                     publish_state, meta_main.frame_id, meta_extra.frame_id, frame_id,
+      fill_model_msg(drivingdata_send, modelv2_send, model_output, publish_state, meta_main.frame_id, meta_extra.frame_id, frame_id,
                      frame_drop_ratio, meta_main.timestamp_eof, model_execution_time, live_calib_seen)
 
       desire_state = modelv2_send.modelV2.meta.desireState
@@ -326,6 +291,7 @@ def main(demo=False):
       pm.send('modelV2', modelv2_send)
       pm.send('drivingModelData', drivingdata_send)
       pm.send('cameraOdometry', posenet_send)
+
     last_vipc_frame_id = meta_main.frame_id
 
 

selfdrive/modeld/models/commonmodel.cc

@@ -1,61 +1,58 @@
 #include "selfdrive/modeld/models/commonmodel.h"
 
+#include <cassert>
 #include <cmath>
 #include <cstring>
 
 #include "common/clutil.h"
 
-DrivingModelFrame::DrivingModelFrame(cl_device_id device_id, cl_context context) : ModelFrame(device_id, context) {
+ModelFrame::ModelFrame(cl_device_id device_id, cl_context context) {
   input_frames = std::make_unique<uint8_t[]>(buf_size);
-  input_frames_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, buf_size, NULL, &err));
+
+  q = CL_CHECK_ERR(clCreateCommandQueue(context, device_id, 0, &err));
+  y_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, MODEL_WIDTH * MODEL_HEIGHT, NULL, &err));
+  u_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, (MODEL_WIDTH / 2) * (MODEL_HEIGHT / 2), NULL, &err));
+  v_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, (MODEL_WIDTH / 2) * (MODEL_HEIGHT / 2), NULL, &err));
   img_buffer_20hz_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, 5*frame_size_bytes, NULL, &err));
   region.origin = 4 * frame_size_bytes;
   region.size = frame_size_bytes;
   last_img_cl = CL_CHECK_ERR(clCreateSubBuffer(img_buffer_20hz_cl, CL_MEM_READ_WRITE, CL_BUFFER_CREATE_TYPE_REGION, &region, &err));
 
+  transform_init(&transform, context, device_id);
   loadyuv_init(&loadyuv, context, device_id, MODEL_WIDTH, MODEL_HEIGHT);
-  init_transform(device_id, context, MODEL_WIDTH, MODEL_HEIGHT);
 }
 
-cl_mem* DrivingModelFrame::prepare(cl_mem yuv_cl, int frame_width, int frame_height, int frame_stride, int frame_uv_offset, const mat3& projection) {
-  run_transform(yuv_cl, MODEL_WIDTH, MODEL_HEIGHT, frame_width, frame_height, frame_stride, frame_uv_offset, projection);
+uint8_t* ModelFrame::prepare(cl_mem yuv_cl, int frame_width, int frame_height, int frame_stride, int frame_uv_offset, const mat3 &projection, cl_mem *output) {
+  transform_queue(&this->transform, q,
+                yuv_cl, frame_width, frame_height, frame_stride, frame_uv_offset,
+                y_cl, u_cl, v_cl, MODEL_WIDTH, MODEL_HEIGHT, projection);
 
   for (int i = 0; i < 4; i++) {
     CL_CHECK(clEnqueueCopyBuffer(q, img_buffer_20hz_cl, img_buffer_20hz_cl, (i+1)*frame_size_bytes, i*frame_size_bytes, frame_size_bytes, 0, nullptr, nullptr));
   }
   loadyuv_queue(&loadyuv, q, y_cl, u_cl, v_cl, last_img_cl);
+  if (output == NULL) {
+    CL_CHECK(clEnqueueReadBuffer(q, img_buffer_20hz_cl, CL_TRUE, 0, frame_size_bytes, &input_frames[0], 0, nullptr, nullptr));
+    CL_CHECK(clEnqueueReadBuffer(q, last_img_cl, CL_TRUE, 0, frame_size_bytes, &input_frames[MODEL_FRAME_SIZE], 0, nullptr, nullptr));
+    clFinish(q);
+    return &input_frames[0];
+  } else {
+    copy_queue(&loadyuv, q, img_buffer_20hz_cl, *output, 0, 0, frame_size_bytes);
+    copy_queue(&loadyuv, q, last_img_cl, *output, 0, frame_size_bytes, frame_size_bytes);
 
-  copy_queue(&loadyuv, q, img_buffer_20hz_cl, input_frames_cl, 0, 0, frame_size_bytes);
-  copy_queue(&loadyuv, q, last_img_cl, input_frames_cl, 0, frame_size_bytes, frame_size_bytes);
-
-  // NOTE: Since thneed is using a different command queue, this clFinish is needed to ensure the image is ready.
-  clFinish(q);
-  return &input_frames_cl;
+    // NOTE: Since thneed is using a different command queue, this clFinish is needed to ensure the image is ready.
+    clFinish(q);
+    return NULL;
+  }
 }
 
-DrivingModelFrame::~DrivingModelFrame() {
-  deinit_transform();
+ModelFrame::~ModelFrame() {
+  transform_destroy(&transform);
   loadyuv_destroy(&loadyuv);
   CL_CHECK(clReleaseMemObject(img_buffer_20hz_cl));
   CL_CHECK(clReleaseMemObject(last_img_cl));
+  CL_CHECK(clReleaseMemObject(v_cl));
+  CL_CHECK(clReleaseMemObject(u_cl));
+  CL_CHECK(clReleaseMemObject(y_cl));
   CL_CHECK(clReleaseCommandQueue(q));
-}
-
-
-MonitoringModelFrame::MonitoringModelFrame(cl_device_id device_id, cl_context context) : ModelFrame(device_id, context) {
-  input_frames = std::make_unique<uint8_t[]>(buf_size);
-  input_frame_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, buf_size, NULL, &err));
-
-  init_transform(device_id, context, MODEL_WIDTH, MODEL_HEIGHT);
-}
-
-cl_mem* MonitoringModelFrame::prepare(cl_mem yuv_cl, int frame_width, int frame_height, int frame_stride, int frame_uv_offset, const mat3& projection) {
-  run_transform(yuv_cl, MODEL_WIDTH, MODEL_HEIGHT, frame_width, frame_height, frame_stride, frame_uv_offset, projection);
-  clFinish(q);
-  return &y_cl;
-}
-
-MonitoringModelFrame::~MonitoringModelFrame() {
-  deinit_transform();
-  CL_CHECK(clReleaseCommandQueue(q));
-}
+}

selfdrive/modeld/models/commonmodel.h

@@ -2,7 +2,6 @@
 
 #include <cfloat>
 #include <cstdlib>
-#include <cassert>
 
 #include <memory>
 
@@ -19,54 +18,9 @@
 
 class ModelFrame {
 public:
-  ModelFrame(cl_device_id device_id, cl_context context) {
-    q = CL_CHECK_ERR(clCreateCommandQueue(context, device_id, 0, &err));
-  }
-  virtual ~ModelFrame() {}
-  virtual cl_mem* prepare(cl_mem yuv_cl, int frame_width, int frame_height, int frame_stride, int frame_uv_offset, const mat3& projection) { return NULL; }
-  uint8_t* buffer_from_cl(cl_mem *in_frames, int buffer_size) {
-    CL_CHECK(clEnqueueReadBuffer(q, *in_frames, CL_TRUE, 0, buffer_size, input_frames.get(), 0, nullptr, nullptr));
-    clFinish(q);
-    return &input_frames[0];
-  }
-
-  int MODEL_WIDTH;
-  int MODEL_HEIGHT;
-  int MODEL_FRAME_SIZE;
-  int buf_size;
-
-protected:
-  cl_mem y_cl, u_cl, v_cl;
-  Transform transform;
-  cl_command_queue q;
-  std::unique_ptr<uint8_t[]> input_frames;
-
-  void init_transform(cl_device_id device_id, cl_context context, int model_width, int model_height) {
-    y_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, model_width * model_height, NULL, &err));
-    u_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, (model_width / 2) * (model_height / 2), NULL, &err));
-    v_cl = CL_CHECK_ERR(clCreateBuffer(context, CL_MEM_READ_WRITE, (model_width / 2) * (model_height / 2), NULL, &err));
-    transform_init(&transform, context, device_id);
-  }
-
-  void deinit_transform() {
-    transform_destroy(&transform);
-    CL_CHECK(clReleaseMemObject(v_cl));
-    CL_CHECK(clReleaseMemObject(u_cl));
-    CL_CHECK(clReleaseMemObject(y_cl));
-  }
-
-  void run_transform(cl_mem yuv_cl, int model_width, int model_height, int frame_width, int frame_height, int frame_stride, int frame_uv_offset, const mat3& projection) {
-    transform_queue(&transform, q,
-        yuv_cl, frame_width, frame_height, frame_stride, frame_uv_offset,
-        y_cl, u_cl, v_cl, model_width, model_height, projection);
-  }
-};
-
-class DrivingModelFrame : public ModelFrame {
-public:
-  DrivingModelFrame(cl_device_id device_id, cl_context context);
-  ~DrivingModelFrame();
-  cl_mem* prepare(cl_mem yuv_cl, int frame_width, int frame_height, int frame_stride, int frame_uv_offset, const mat3& projection);
+  ModelFrame(cl_device_id device_id, cl_context context);
+  ~ModelFrame();
+  uint8_t* prepare(cl_mem yuv_cl, int width, int height, int frame_stride, int frame_uv_offset, const mat3& transform, cl_mem *output);
 
   const int MODEL_WIDTH = 512;
   const int MODEL_HEIGHT = 256;
@@ -75,22 +29,10 @@ public:
   const size_t frame_size_bytes = MODEL_FRAME_SIZE * sizeof(uint8_t);
 
 private:
+  Transform transform;
   LoadYUVState loadyuv;
-  cl_mem img_buffer_20hz_cl, last_img_cl, input_frames_cl;
+  cl_command_queue q;
+  cl_mem y_cl, u_cl, v_cl, img_buffer_20hz_cl, last_img_cl;
   cl_buffer_region region;
-};
-
-class MonitoringModelFrame : public ModelFrame {
-public:
-  MonitoringModelFrame(cl_device_id device_id, cl_context context);
-  ~MonitoringModelFrame();
-  cl_mem* prepare(cl_mem yuv_cl, int frame_width, int frame_height, int frame_stride, int frame_uv_offset, const mat3& projection);
-
-  const int MODEL_WIDTH = 1440;
-  const int MODEL_HEIGHT = 960;
-  const int MODEL_FRAME_SIZE = MODEL_WIDTH * MODEL_HEIGHT;
-  const int buf_size = MODEL_FRAME_SIZE;
-
-private:
-  cl_mem input_frame_cl;
-};
+  std::unique_ptr<uint8_t[]> input_frames;
+};

selfdrive/modeld/models/commonmodel.pxd

@@ -14,13 +14,5 @@ cdef extern from "common/clutil.h":
 cdef extern from "selfdrive/modeld/models/commonmodel.h":
   cppclass ModelFrame:
     int buf_size
-    unsigned char * buffer_from_cl(cl_mem*, int);
-    cl_mem * prepare(cl_mem, int, int, int, int, mat3)
-
-  cppclass DrivingModelFrame:
-    int buf_size
-    DrivingModelFrame(cl_device_id, cl_context)
-
-  cppclass MonitoringModelFrame:
-    int buf_size
-    MonitoringModelFrame(cl_device_id, cl_context)
+    ModelFrame(cl_device_id, cl_context)
+    unsigned char * prepare(cl_mem, int, int, int, int, mat3, cl_mem*)

selfdrive/modeld/models/commonmodel_pyx.pyx

@@ -4,12 +4,11 @@
 import numpy as np
 cimport numpy as cnp
 from libc.string cimport memcpy
-from libc.stdint cimport uintptr_t
 
 from msgq.visionipc.visionipc cimport cl_mem
 from msgq.visionipc.visionipc_pyx cimport VisionBuf, CLContext as BaseCLContext
 from .commonmodel cimport CL_DEVICE_TYPE_DEFAULT, cl_get_device_id, cl_create_context
-from .commonmodel cimport mat3, ModelFrame as cppModelFrame, DrivingModelFrame as cppDrivingModelFrame, MonitoringModelFrame as cppMonitoringModelFrame
+from .commonmodel cimport mat3, ModelFrame as cppModelFrame
 
 
 cdef class CLContext(BaseCLContext):
@@ -24,47 +23,23 @@ cdef class CLMem:
     mem.mem = <cl_mem*> cmem
     return mem
 
-  @property
-  def mem_address(self):
-    return <uintptr_t>(self.mem)
-
-def cl_from_visionbuf(VisionBuf buf):
-  return CLMem.create(<void*>&buf.buf.buf_cl)
-
-
 cdef class ModelFrame:
   cdef cppModelFrame * frame
-  cdef int buf_size
+
+  def __cinit__(self, CLContext context):
+    self.frame = new cppModelFrame(context.device_id, context.context)
 
   def __dealloc__(self):
     del self.frame
 
-  def prepare(self, VisionBuf buf, float[:] projection):
+  def prepare(self, VisionBuf buf, float[:] projection, CLMem output):
     cdef mat3 cprojection
     memcpy(cprojection.v, &projection[0], 9*sizeof(float))
-    cdef cl_mem * data
-    data = self.frame.prepare(buf.buf.buf_cl, buf.width, buf.height, buf.stride, buf.uv_offset, cprojection)
-    return CLMem.create(data)
-
-  def buffer_from_cl(self, CLMem in_frames):
-    cdef unsigned char * data2
-    data2 = self.frame.buffer_from_cl(in_frames.mem, self.buf_size)
-    return np.asarray(<cnp.uint8_t[:self.buf_size]> data2)
-
-
-cdef class DrivingModelFrame(ModelFrame):
-  cdef cppDrivingModelFrame * _frame
-
-  def __cinit__(self, CLContext context):
-    self._frame = new cppDrivingModelFrame(context.device_id, context.context)
-    self.frame = <cppModelFrame*>(self._frame)
-    self.buf_size = self._frame.buf_size
-
-cdef class MonitoringModelFrame(ModelFrame):
-  cdef cppMonitoringModelFrame * _frame
-
-  def __cinit__(self, CLContext context):
-    self._frame = new cppMonitoringModelFrame(context.device_id, context.context)
-    self.frame = <cppModelFrame*>(self._frame)
-    self.buf_size = self._frame.buf_size
-
+    cdef unsigned char * data
+    if output is None:
+      data = self.frame.prepare(buf.buf.buf_cl, buf.width, buf.height, buf.stride, buf.uv_offset, cprojection, NULL)
+    else:
+      data = self.frame.prepare(buf.buf.buf_cl, buf.width, buf.height, buf.stride, buf.uv_offset, cprojection, output.mem)
+    if not data:
+      return None
+    return np.asarray(<cnp.uint8_t[:self.frame.buf_size]> data)

selfdrive/modeld/parse_model_outputs.py

@@ -96,6 +96,8 @@ class Parser:
                    out_shape=(ModelConstants.LEAD_TRAJ_LEN,ModelConstants.LEAD_WIDTH))
     if 'lat_planner_solution' in outs:
       self.parse_mdn('lat_planner_solution', outs, in_N=0, out_N=0, out_shape=(ModelConstants.IDX_N,ModelConstants.LAT_PLANNER_SOLUTION_WIDTH))
+    if 'desired_curvature' in outs:
+      self.parse_mdn('desired_curvature', outs, in_N=0, out_N=0, out_shape=(ModelConstants.DESIRED_CURV_WIDTH,))
     for k in ['lead_prob', 'lane_lines_prob', 'meta']:
       self.parse_binary_crossentropy(k, outs)
     self.parse_categorical_crossentropy('desire_state', outs, out_shape=(ModelConstants.DESIRE_PRED_WIDTH,))

selfdrive/modeld/runners/__init__.py

@@ -0,0 +1,27 @@
+import os
+from openpilot.system.hardware import TICI
+from openpilot.selfdrive.modeld.runners.runmodel_pyx import RunModel, Runtime
+assert Runtime
+
+USE_THNEED = int(os.getenv('USE_THNEED', str(int(TICI))))
+USE_SNPE = int(os.getenv('USE_SNPE', str(int(TICI))))
+
+class ModelRunner(RunModel):
+  THNEED = 'THNEED'
+  SNPE = 'SNPE'
+  ONNX = 'ONNX'
+
+  def __new__(cls, paths, *args, **kwargs):
+    if ModelRunner.THNEED in paths and USE_THNEED:
+      from openpilot.selfdrive.modeld.runners.thneedmodel_pyx import ThneedModel as Runner
+      runner_type = ModelRunner.THNEED
+    elif ModelRunner.SNPE in paths and USE_SNPE:
+      from openpilot.selfdrive.modeld.runners.snpemodel_pyx import SNPEModel as Runner
+      runner_type = ModelRunner.SNPE
+    elif ModelRunner.ONNX in paths:
+      from openpilot.selfdrive.modeld.runners.onnxmodel import ONNXModel as Runner
+      runner_type = ModelRunner.ONNX
+    else:
+      raise Exception("Couldn't select a model runner, make sure to pass at least one valid model path")
+
+    return Runner(str(paths[runner_type]), *args, **kwargs)

selfdrive/modeld/runners/onnxmodel.py

@@ -0,0 +1,98 @@
+import onnx
+import itertools
+import os
+import sys
+import numpy as np
+from typing import Any
+
+from openpilot.selfdrive.modeld.runners.runmodel_pyx import RunModel
+
+ORT_TYPES_TO_NP_TYPES = {'tensor(float16)': np.float16, 'tensor(float)': np.float32, 'tensor(uint8)': np.uint8}
+
+def attributeproto_fp16_to_fp32(attr):
+  float32_list = np.frombuffer(attr.raw_data, dtype=np.float16)
+  attr.data_type = 1
+  attr.raw_data = float32_list.astype(np.float32).tobytes()
+
+def convert_fp16_to_fp32(onnx_path_or_bytes):
+  if isinstance(onnx_path_or_bytes, bytes):
+    model = onnx.load_from_string(onnx_path_or_bytes)
+  elif isinstance(onnx_path_or_bytes, str):
+    model = onnx.load(onnx_path_or_bytes)
+
+  for i in model.graph.initializer:
+    if i.data_type == 10:
+      attributeproto_fp16_to_fp32(i)
+  for i in itertools.chain(model.graph.input, model.graph.output):
+    if i.type.tensor_type.elem_type == 10:
+      i.type.tensor_type.elem_type = 1
+  for i in model.graph.node:
+    if i.op_type == 'Cast' and i.attribute[0].i == 10:
+      i.attribute[0].i = 1
+    for a in i.attribute:
+      if hasattr(a, 't'):
+        if a.t.data_type == 10:
+          attributeproto_fp16_to_fp32(a.t)
+  return model.SerializeToString()
+
+def create_ort_session(path, fp16_to_fp32):
+  os.environ["OMP_NUM_THREADS"] = "4"
+  os.environ["OMP_WAIT_POLICY"] = "PASSIVE"
+
+  import onnxruntime as ort
+  print("Onnx available providers: ", ort.get_available_providers(), file=sys.stderr)
+  options = ort.SessionOptions()
+  options.graph_optimization_level = ort.GraphOptimizationLevel.ORT_DISABLE_ALL
+
+  provider: str | tuple[str, dict[Any, Any]]
+  if 'OpenVINOExecutionProvider' in ort.get_available_providers() and 'ONNXCPU' not in os.environ:
+    provider = 'OpenVINOExecutionProvider'
+  elif 'CUDAExecutionProvider' in ort.get_available_providers() and 'ONNXCPU' not in os.environ:
+    options.intra_op_num_threads = 2
+    provider = ('CUDAExecutionProvider', {'cudnn_conv_algo_search': 'DEFAULT'})
+  else:
+    options.intra_op_num_threads = 2
+    options.execution_mode = ort.ExecutionMode.ORT_SEQUENTIAL
+    options.graph_optimization_level = ort.GraphOptimizationLevel.ORT_ENABLE_ALL
+    provider = 'CPUExecutionProvider'
+
+  model_data = convert_fp16_to_fp32(path) if fp16_to_fp32 else path
+  print("Onnx selected provider: ", [provider], file=sys.stderr)
+  ort_session = ort.InferenceSession(model_data, options, providers=[provider])
+  print("Onnx using ", ort_session.get_providers(), file=sys.stderr)
+  return ort_session
+
+
+class ONNXModel(RunModel):
+  def __init__(self, path, output, runtime, use_tf8, cl_context):
+    self.inputs = {}
+    self.output = output
+
+    self.session = create_ort_session(path, fp16_to_fp32=True)
+    self.input_names = [x.name for x in self.session.get_inputs()]
+    self.input_shapes = {x.name: [1, *x.shape[1:]] for x in self.session.get_inputs()}
+    self.input_dtypes = {x.name: ORT_TYPES_TO_NP_TYPES[x.type] for x in self.session.get_inputs()}
+
+    # run once to initialize CUDA provider
+    if "CUDAExecutionProvider" in self.session.get_providers():
+      self.session.run(None, {k: np.zeros(self.input_shapes[k], dtype=self.input_dtypes[k]) for k in self.input_names})
+    print("ready to run onnx model", self.input_shapes, file=sys.stderr)
+
+  def addInput(self, name, buffer):
+    assert name in self.input_names
+    self.inputs[name] = buffer
+
+  def setInputBuffer(self, name, buffer):
+    assert name in self.inputs
+    self.inputs[name] = buffer
+
+  def getCLBuffer(self, name):
+    return None
+
+  def execute(self):
+    inputs = {k: v.view(self.input_dtypes[k]) for k,v in self.inputs.items()}
+    inputs = {k: v.reshape(self.input_shapes[k]).astype(self.input_dtypes[k]) for k,v in inputs.items()}
+    outputs = self.session.run(None, inputs)
+    assert len(outputs) == 1, "Only single model outputs are supported"
+    self.output[:] = outputs[0]
+    return self.output

selfdrive/modeld/runners/ort_helpers.py

@@ -1,36 +0,0 @@
-import onnx
-import onnxruntime as ort
-import numpy as np
-import itertools
-
-ORT_TYPES_TO_NP_TYPES = {'tensor(float16)': np.float16, 'tensor(float)': np.float32, 'tensor(uint8)': np.uint8}
-
-def attributeproto_fp16_to_fp32(attr):
-  float32_list = np.frombuffer(attr.raw_data, dtype=np.float16)
-  attr.data_type = 1
-  attr.raw_data = float32_list.astype(np.float32).tobytes()
-
-def convert_fp16_to_fp32(model):
-  for i in model.graph.initializer:
-    if i.data_type == 10:
-      attributeproto_fp16_to_fp32(i)
-  for i in itertools.chain(model.graph.input, model.graph.output):
-    if i.type.tensor_type.elem_type == 10:
-      i.type.tensor_type.elem_type = 1
-  for i in model.graph.node:
-    if i.op_type == 'Cast' and i.attribute[0].i == 10:
-      i.attribute[0].i = 1
-    for a in i.attribute:
-      if hasattr(a, 't'):
-        if a.t.data_type == 10:
-          attributeproto_fp16_to_fp32(a.t)
-  return model.SerializeToString()
-
-
-def make_onnx_cpu_runner(model_path):
-  options = ort.SessionOptions()
-  options.intra_op_num_threads = 4
-  options.execution_mode = ort.ExecutionMode.ORT_SEQUENTIAL
-  options.graph_optimization_level = ort.GraphOptimizationLevel.ORT_ENABLE_ALL
-  model_data = convert_fp16_to_fp32(onnx.load(model_path))
-  return ort.InferenceSession(model_data,  options, providers=['CPUExecutionProvider'])

selfdrive/modeld/runners/run.h

@@ -0,0 +1,4 @@
+#pragma once
+
+#include "selfdrive/modeld/runners/runmodel.h"
+#include "selfdrive/modeld/runners/snpemodel.h"

selfdrive/modeld/runners/runmodel.h

@@ -0,0 +1,49 @@
+#pragma once
+
+#include <string>
+#include <vector>
+#include <memory>
+#include <cassert>
+
+#include "common/clutil.h"
+#include "common/swaglog.h"
+
+#define USE_CPU_RUNTIME 0
+#define USE_GPU_RUNTIME 1
+#define USE_DSP_RUNTIME 2
+
+struct ModelInput {
+  const std::string name;
+  float *buffer;
+  int size;
+
+  ModelInput(const std::string _name, float *_buffer, int _size) : name(_name), buffer(_buffer), size(_size) {}
+  virtual void setBuffer(float *_buffer, int _size) {
+    assert(size == _size || size == 0);
+    buffer = _buffer;
+    size = _size;
+  }
+};
+
+class RunModel {
+public:
+  std::vector<std::unique_ptr<ModelInput>> inputs;
+
+  virtual ~RunModel() {}
+  virtual void execute() {}
+  virtual void* getCLBuffer(const std::string name) { return nullptr; }
+
+  virtual void addInput(const std::string name, float *buffer, int size) {
+    inputs.push_back(std::unique_ptr<ModelInput>(new ModelInput(name, buffer, size)));
+  }
+  virtual void setInputBuffer(const std::string name, float *buffer, int size) {
+    for (auto &input : inputs) {
+      if (name == input->name) {
+        input->setBuffer(buffer, size);
+        return;
+      }
+    }
+    LOGE("Tried to update input `%s` but no input with this name exists", name.c_str());
+    assert(false);
+  }
+};

selfdrive/modeld/runners/runmodel.pxd

@@ -0,0 +1,14 @@
+# distutils: language = c++
+
+from libcpp.string cimport string
+
+cdef extern from "selfdrive/modeld/runners/runmodel.h":
+  cdef int USE_CPU_RUNTIME
+  cdef int USE_GPU_RUNTIME
+  cdef int USE_DSP_RUNTIME
+
+  cdef cppclass RunModel:
+    void addInput(string, float*, int)
+    void setInputBuffer(string, float*, int)
+    void * getCLBuffer(string)
+    void execute()

selfdrive/modeld/runners/runmodel_pyx.pxd

@@ -0,0 +1,6 @@
+# distutils: language = c++
+
+from .runmodel cimport RunModel as cppRunModel
+
+cdef class RunModel:
+  cdef cppRunModel * model

selfdrive/modeld/runners/runmodel_pyx.pyx

@@ -0,0 +1,37 @@
+# distutils: language = c++
+# cython: c_string_encoding=ascii, language_level=3
+
+from libcpp.string cimport string
+
+from .runmodel cimport USE_CPU_RUNTIME, USE_GPU_RUNTIME, USE_DSP_RUNTIME
+from selfdrive.modeld.models.commonmodel_pyx cimport CLMem
+
+class Runtime:
+  CPU = USE_CPU_RUNTIME
+  GPU = USE_GPU_RUNTIME
+  DSP = USE_DSP_RUNTIME
+
+cdef class RunModel:
+  def __dealloc__(self):
+    del self.model
+
+  def addInput(self, string name, float[:] buffer):
+    if buffer is not None:
+      self.model.addInput(name, &buffer[0], len(buffer))
+    else:
+      self.model.addInput(name, NULL, 0)
+
+  def setInputBuffer(self, string name, float[:] buffer):
+    if buffer is not None:
+      self.model.setInputBuffer(name, &buffer[0], len(buffer))
+    else:
+      self.model.setInputBuffer(name, NULL, 0)
+
+  def getCLBuffer(self, string name):
+    cdef void * cl_buf = self.model.getCLBuffer(name)
+    if not cl_buf:
+      return None
+    return CLMem.create(cl_buf)
+
+  def execute(self):
+    self.model.execute()

selfdrive/modeld/runners/snpemodel.cc

@@ -0,0 +1,116 @@
+#pragma clang diagnostic ignored "-Wexceptions"
+
+#include "selfdrive/modeld/runners/snpemodel.h"
+
+#include <cstring>
+#include <memory>
+#include <string>
+#include <utility>
+#include <vector>
+
+#include "common/util.h"
+#include "common/timing.h"
+
+void PrintErrorStringAndExit() {
+  std::cerr << zdl::DlSystem::getLastErrorString() << std::endl;
+  std::exit(EXIT_FAILURE);
+}
+
+SNPEModel::SNPEModel(const std::string path, float *_output, size_t _output_size, int runtime, bool _use_tf8, cl_context context) {
+  output = _output;
+  output_size = _output_size;
+  use_tf8 = _use_tf8;
+
+#ifdef QCOM2
+  if (runtime == USE_GPU_RUNTIME) {
+    snpe_runtime = zdl::DlSystem::Runtime_t::GPU;
+  } else if (runtime == USE_DSP_RUNTIME) {
+    snpe_runtime = zdl::DlSystem::Runtime_t::DSP;
+  } else {
+    snpe_runtime = zdl::DlSystem::Runtime_t::CPU;
+  }
+  assert(zdl::SNPE::SNPEFactory::isRuntimeAvailable(snpe_runtime));
+#endif
+  model_data = util::read_file(path);
+  assert(model_data.size() > 0);
+
+  // load model
+  std::unique_ptr<zdl::DlContainer::IDlContainer> container = zdl::DlContainer::IDlContainer::open((uint8_t*)model_data.data(), model_data.size());
+  if (!container) { PrintErrorStringAndExit(); }
+  LOGW("loaded model with size: %lu", model_data.size());
+
+  // create model runner
+  zdl::SNPE::SNPEBuilder snpe_builder(container.get());
+  while (!snpe) {
+#ifdef QCOM2
+    snpe = snpe_builder.setOutputLayers({})
+                       .setRuntimeProcessor(snpe_runtime)
+                       .setUseUserSuppliedBuffers(true)
+                       .setPerformanceProfile(zdl::DlSystem::PerformanceProfile_t::HIGH_PERFORMANCE)
+                       .build();
+#else
+    snpe = snpe_builder.setOutputLayers({})
+                       .setUseUserSuppliedBuffers(true)
+                       .setPerformanceProfile(zdl::DlSystem::PerformanceProfile_t::HIGH_PERFORMANCE)
+                       .build();
+#endif
+    if (!snpe) std::cerr << zdl::DlSystem::getLastErrorString() << std::endl;
+  }
+
+  // create output buffer
+  zdl::DlSystem::UserBufferEncodingFloat ub_encoding_float;
+  zdl::DlSystem::IUserBufferFactory &ub_factory = zdl::SNPE::SNPEFactory::getUserBufferFactory();
+
+  const auto &output_tensor_names_opt = snpe->getOutputTensorNames();
+  if (!output_tensor_names_opt) throw std::runtime_error("Error obtaining output tensor names");
+  const auto &output_tensor_names = *output_tensor_names_opt;
+  assert(output_tensor_names.size() == 1);
+  const char *output_tensor_name = output_tensor_names.at(0);
+  const zdl::DlSystem::TensorShape &buffer_shape = snpe->getInputOutputBufferAttributes(output_tensor_name)->getDims();
+  if (output_size != 0) {
+    assert(output_size == buffer_shape[1]);
+  } else {
+    output_size = buffer_shape[1];
+  }
+  std::vector<size_t> output_strides = {output_size * sizeof(float), sizeof(float)};
+  output_buffer = ub_factory.createUserBuffer(output, output_size * sizeof(float), output_strides, &ub_encoding_float);
+  output_map.add(output_tensor_name, output_buffer.get());
+}
+
+void SNPEModel::addInput(const std::string name, float *buffer, int size) {
+  const int idx = inputs.size();
+  const auto &input_tensor_names_opt = snpe->getInputTensorNames();
+  if (!input_tensor_names_opt) throw std::runtime_error("Error obtaining input tensor names");
+  const auto &input_tensor_names = *input_tensor_names_opt;
+  const char *input_tensor_name = input_tensor_names.at(idx);
+  const bool input_tf8 = use_tf8 && strcmp(input_tensor_name, "input_img") == 0;  // TODO: This is a terrible hack, get rid of this name check both here and in onnx_runner.py
+  LOGW("adding index %d: %s", idx, input_tensor_name);
+
+  zdl::DlSystem::UserBufferEncodingFloat ub_encoding_float;
+  zdl::DlSystem::UserBufferEncodingTf8 ub_encoding_tf8(0, 1./255); // network takes 0-1
+  zdl::DlSystem::IUserBufferFactory &ub_factory = zdl::SNPE::SNPEFactory::getUserBufferFactory();
+  zdl::DlSystem::UserBufferEncoding *input_encoding = input_tf8 ? (zdl::DlSystem::UserBufferEncoding*)&ub_encoding_tf8 : (zdl::DlSystem::UserBufferEncoding*)&ub_encoding_float;
+
+  const auto &buffer_shape_opt = snpe->getInputDimensions(input_tensor_name);
+  const zdl::DlSystem::TensorShape &buffer_shape = *buffer_shape_opt;
+  size_t size_of_input = input_tf8 ? sizeof(uint8_t) : sizeof(float);
+  std::vector<size_t> strides(buffer_shape.rank());
+  strides[strides.size() - 1] = size_of_input;
+  size_t product = 1;
+  for (size_t i = 0; i < buffer_shape.rank(); i++) product *= buffer_shape[i];
+  size_t stride = strides[strides.size() - 1];
+  for (size_t i = buffer_shape.rank() - 1; i > 0; i--) {
+    stride *= buffer_shape[i];
+    strides[i-1] = stride;
+  }
+
+  auto input_buffer = ub_factory.createUserBuffer(buffer, product*size_of_input, strides, input_encoding);
+  input_map.add(input_tensor_name, input_buffer.get());
+  inputs.push_back(std::unique_ptr<SNPEModelInput>(new SNPEModelInput(name, buffer, size, std::move(input_buffer))));
+}
+
+void SNPEModel::execute() {
+  if (!snpe->execute(input_map, output_map)) {
+    PrintErrorStringAndExit();
+  }
+}

selfdrive/modeld/runners/snpemodel.h

@@ -0,0 +1,52 @@
+#pragma once
+#pragma clang diagnostic ignored "-Wdeprecated-declarations"
+
+#include <memory>
+#include <string>
+#include <utility>
+
+#include <DlContainer/IDlContainer.hpp>
+#include <DlSystem/DlError.hpp>
+#include <DlSystem/ITensor.hpp>
+#include <DlSystem/ITensorFactory.hpp>
+#include <DlSystem/IUserBuffer.hpp>
+#include <DlSystem/IUserBufferFactory.hpp>
+#include <SNPE/SNPE.hpp>
+#include <SNPE/SNPEBuilder.hpp>
+#include <SNPE/SNPEFactory.hpp>
+
+#include "selfdrive/modeld/runners/runmodel.h"
+
+struct SNPEModelInput : public ModelInput {
+  std::unique_ptr<zdl::DlSystem::IUserBuffer> snpe_buffer;
+
+  SNPEModelInput(const std::string _name, float *_buffer, int _size, std::unique_ptr<zdl::DlSystem::IUserBuffer> _snpe_buffer) : ModelInput(_name, _buffer, _size), snpe_buffer(std::move(_snpe_buffer)) {}
+  void setBuffer(float *_buffer, int _size) {
+    ModelInput::setBuffer(_buffer, _size);
+    assert(snpe_buffer->setBufferAddress(_buffer) == true);
+  }
+};
+
+class SNPEModel : public RunModel {
+public:
+  SNPEModel(const std::string path, float *_output, size_t _output_size, int runtime, bool use_tf8 = false, cl_context context = NULL);
+  void addInput(const std::string name, float *buffer, int size);
+  void execute();
+
+private:
+  std::string model_data;
+
+#ifdef QCOM2
+  zdl::DlSystem::Runtime_t snpe_runtime;
+#endif
+
+  // snpe model stuff
+  std::unique_ptr<zdl::SNPE::SNPE> snpe;
+  zdl::DlSystem::UserBufferMap input_map;
+  zdl::DlSystem::UserBufferMap output_map;
+  std::unique_ptr<zdl::DlSystem::IUserBuffer> output_buffer;
+
+  bool use_tf8;
+  float *output;
+  size_t output_size;
+};

selfdrive/modeld/runners/snpemodel.pxd

@@ -0,0 +1,9 @@
+# distutils: language = c++
+
+from libcpp.string cimport string
+
+from msgq.visionipc.visionipc cimport cl_context
+
+cdef extern from "selfdrive/modeld/runners/snpemodel.h":
+  cdef cppclass SNPEModel:
+    SNPEModel(string, float*, size_t, int, bool, cl_context)

selfdrive/modeld/runners/snpemodel_pyx.pyx

@@ -0,0 +1,17 @@
+# distutils: language = c++
+# cython: c_string_encoding=ascii, language_level=3
+
+import os
+from libcpp cimport bool
+from libcpp.string cimport string
+
+from .snpemodel cimport SNPEModel as cppSNPEModel
+from selfdrive.modeld.models.commonmodel_pyx cimport CLContext
+from selfdrive.modeld.runners.runmodel_pyx cimport RunModel
+from selfdrive.modeld.runners.runmodel cimport RunModel as cppRunModel
+
+os.environ['ADSP_LIBRARY_PATH'] = "/data/pythonpath/third_party/snpe/dsp/"
+
+cdef class SNPEModel(RunModel):
+  def __cinit__(self, string path, float[:] output, int runtime, bool use_tf8, CLContext context):
+    self.model = <cppRunModel *> new cppSNPEModel(path, &output[0], len(output), runtime, use_tf8, context.context)

selfdrive/modeld/runners/thneedmodel.cc

@@ -0,0 +1,58 @@
+#include "selfdrive/modeld/runners/thneedmodel.h"
+
+#include <string>
+
+#include "common/swaglog.h"
+
+ThneedModel::ThneedModel(const std::string path, float *_output, size_t _output_size, int runtime, bool luse_tf8, cl_context context) {
+  thneed = new Thneed(true, context);
+  thneed->load(path.c_str());
+  thneed->clexec();
+
+  recorded = false;
+  output = _output;
+}
+
+void* ThneedModel::getCLBuffer(const std::string name) {
+  int index = -1;
+  for (int i = 0; i < inputs.size(); i++) {
+    if (name == inputs[i]->name) {
+      index = i;
+      break;
+    }
+  }
+
+  if (index == -1) {
+    LOGE("Tried to get CL buffer for input `%s` but no input with this name exists", name.c_str());
+    assert(false);
+  }
+
+  if (thneed->input_clmem.size() >= inputs.size()) {
+    return &thneed->input_clmem[inputs.size() - index - 1];
+  } else {
+    return nullptr;
+  }
+}
+
+void ThneedModel::execute() {
+  if (!recorded) {
+    thneed->record = true;
+    float *input_buffers[inputs.size()];
+    for (int i = 0; i < inputs.size(); i++) {
+      input_buffers[inputs.size() - i - 1] = inputs[i]->buffer;
+    }
+
+    thneed->copy_inputs(input_buffers);
+    thneed->clexec();
+    thneed->copy_output(output);
+    thneed->stop();
+
+    recorded = true;
+  } else {
+    float *input_buffers[inputs.size()];
+    for (int i = 0; i < inputs.size(); i++) {
+      input_buffers[inputs.size() - i - 1] = inputs[i]->buffer;
+    }
+    thneed->execute(input_buffers, output);
+  }
+}

selfdrive/modeld/runners/thneedmodel.h

@@ -0,0 +1,17 @@
+#pragma once
+
+#include <string>
+
+#include "selfdrive/modeld/runners/runmodel.h"
+#include "selfdrive/modeld/thneed/thneed.h"
+
+class ThneedModel : public RunModel {
+public:
+  ThneedModel(const std::string path, float *_output, size_t _output_size, int runtime, bool use_tf8 = false, cl_context context = NULL);
+  void *getCLBuffer(const std::string name);
+  void execute();
+private:
+  Thneed *thneed = NULL;
+  bool recorded;
+  float *output;
+};

selfdrive/modeld/runners/thneedmodel.pxd

@@ -0,0 +1,9 @@
+# distutils: language = c++
+
+from libcpp.string cimport string
+
+from msgq.visionipc.visionipc cimport cl_context
+
+cdef extern from "selfdrive/modeld/runners/thneedmodel.h":
+  cdef cppclass ThneedModel:
+    ThneedModel(string, float*, size_t, int, bool, cl_context)

selfdrive/modeld/runners/thneedmodel_pyx.pyx

@@ -0,0 +1,14 @@
+# distutils: language = c++
+# cython: c_string_encoding=ascii, language_level=3
+
+from libcpp cimport bool
+from libcpp.string cimport string
+
+from .thneedmodel cimport ThneedModel as cppThneedModel
+from selfdrive.modeld.models.commonmodel_pyx cimport CLContext
+from selfdrive.modeld.runners.runmodel_pyx cimport RunModel
+from selfdrive.modeld.runners.runmodel cimport RunModel as cppRunModel
+
+cdef class ThneedModel(RunModel):
+  def __cinit__(self, string path, float[:] output, int runtime, bool use_tf8, CLContext context):
+    self.model = <cppRunModel *> new cppThneedModel(path, &output[0], len(output), runtime, use_tf8, context.context)

selfdrive/modeld/runners/tinygrad_helpers.py

@@ -1,8 +0,0 @@
-
-from tinygrad.tensor import Tensor
-from tinygrad.helpers import to_mv
-
-def qcom_tensor_from_opencl_address(opencl_address, shape, dtype):
-  cl_buf_desc_ptr = to_mv(opencl_address, 8).cast('Q')[0]
-  rawbuf_ptr = to_mv(cl_buf_desc_ptr, 0x100).cast('Q')[20] # offset 0xA0 is a raw gpu pointer.
-  return Tensor.from_blob(rawbuf_ptr, shape, dtype=dtype, device='QCOM')

selfdrive/modeld/thneed/README

@@ -0,0 +1,8 @@
+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.
+

selfdrive/modeld/thneed/serialize.cc

@@ -0,0 +1,154 @@
+#include <cassert>
+#include <set>
+
+#include "third_party/json11/json11.hpp"
+#include "common/util.h"
+#include "common/clutil.h"
+#include "common/swaglog.h"
+#include "selfdrive/modeld/thneed/thneed.h"
+using namespace json11;
+
+extern map<cl_program, string> g_program_source;
+
+void Thneed::load(const char *filename) {
+  LOGD("Thneed::load: loading from %s\n", filename);
+
+  string buf = util::read_file(filename);
+  int jsz = *(int *)buf.data();
+  string jsonerr;
+  string jj(buf.data() + sizeof(int), jsz);
+  Json jdat = Json::parse(jj, jsonerr);
+
+  map<cl_mem, cl_mem> real_mem;
+  real_mem[NULL] = NULL;
+
+  int ptr = sizeof(int)+jsz;
+  for (auto &obj : jdat["objects"].array_items()) {
+    auto mobj = obj.object_items();
+    int sz = mobj["size"].int_value();
+    cl_mem clbuf = NULL;
+
+    if (mobj["buffer_id"].string_value().size() > 0) {
+      // image buffer must already be allocated
+      clbuf = real_mem[*(cl_mem*)(mobj["buffer_id"].string_value().data())];
+      assert(mobj["needs_load"].bool_value() == false);
+    } else {
+      if (mobj["needs_load"].bool_value()) {
+        clbuf = clCreateBuffer(context, CL_MEM_COPY_HOST_PTR | CL_MEM_READ_WRITE, sz, &buf[ptr], NULL);
+        if (debug >= 1) printf("loading %p %d @ 0x%X\n", clbuf, sz, ptr);
+        ptr += sz;
+      } else {
+        // TODO: is there a faster way to init zeroed out buffers?
+        void *host_zeros = calloc(sz, 1);
+        clbuf = clCreateBuffer(context, CL_MEM_COPY_HOST_PTR | CL_MEM_READ_WRITE, sz, host_zeros, NULL);
+        free(host_zeros);
+      }
+    }
+    assert(clbuf != NULL);
+
+    if (mobj["arg_type"] == "image2d_t" || mobj["arg_type"] == "image1d_t") {
+      cl_image_desc desc = {0};
+      desc.image_type = (mobj["arg_type"] == "image2d_t") ? CL_MEM_OBJECT_IMAGE2D : CL_MEM_OBJECT_IMAGE1D_BUFFER;
+      desc.image_width = mobj["width"].int_value();
+      desc.image_height = mobj["height"].int_value();
+      desc.image_row_pitch = mobj["row_pitch"].int_value();
+      assert(sz == desc.image_height*desc.image_row_pitch);
+#ifdef QCOM2
+      desc.buffer = clbuf;
+#else
+      // TODO: we are creating unused buffers on PC
+      clReleaseMemObject(clbuf);
+#endif
+      cl_image_format format = {0};
+      format.image_channel_order = CL_RGBA;
+      format.image_channel_data_type = mobj["float32"].bool_value() ? CL_FLOAT : CL_HALF_FLOAT;
+
+      cl_int errcode;
+
+#ifndef QCOM2
+      if (mobj["needs_load"].bool_value()) {
+        clbuf = clCreateImage(context, CL_MEM_COPY_HOST_PTR | CL_MEM_READ_WRITE, &format, &desc, &buf[ptr-sz], &errcode);
+      } else {
+        clbuf = clCreateImage(context, CL_MEM_READ_WRITE, &format, &desc, NULL, &errcode);
+      }
+#else
+      clbuf = clCreateImage(context, CL_MEM_READ_WRITE, &format, &desc, NULL, &errcode);
+#endif
+      if (clbuf == NULL) {
+        LOGE("clError: %s create image %zux%zu rp %zu with buffer %p\n", cl_get_error_string(errcode),
+             desc.image_width, desc.image_height, desc.image_row_pitch, desc.buffer);
+      }
+      assert(clbuf != NULL);
+    }
+
+    real_mem[*(cl_mem*)(mobj["id"].string_value().data())] = clbuf;
+  }
+
+  map<string, cl_program> g_programs;
+  for (const auto &[name, source] : jdat["programs"].object_items()) {
+    if (debug >= 1) printf("building %s with size %zu\n", name.c_str(), source.string_value().size());
+    g_programs[name] = cl_program_from_source(context, device_id, source.string_value());
+  }
+
+  for (auto &obj : jdat["inputs"].array_items()) {
+    auto mobj = obj.object_items();
+    int sz = mobj["size"].int_value();
+    cl_mem aa = real_mem[*(cl_mem*)(mobj["buffer_id"].string_value().data())];
+    input_clmem.push_back(aa);
+    input_sizes.push_back(sz);
+    LOGD("Thneed::load: adding input %s with size %d\n", mobj["name"].string_value().data(), sz);
+
+    cl_int cl_err;
+    void *ret = clEnqueueMapBuffer(command_queue, aa, CL_TRUE, CL_MAP_WRITE, 0, sz, 0, NULL, NULL, &cl_err);
+    if (cl_err != CL_SUCCESS) LOGE("clError: %s map %p %d\n", cl_get_error_string(cl_err), aa, sz);
+    assert(cl_err == CL_SUCCESS);
+    inputs.push_back(ret);
+  }
+
+  for (auto &obj : jdat["outputs"].array_items()) {
+    auto mobj = obj.object_items();
+    int sz = mobj["size"].int_value();
+    LOGD("Thneed::save: adding output with size %d\n", sz);
+    // TODO: support multiple outputs
+    output = real_mem[*(cl_mem*)(mobj["buffer_id"].string_value().data())];
+    assert(output != NULL);
+  }
+
+  for (auto &obj : jdat["binaries"].array_items()) {
+    string name = obj["name"].string_value();
+    size_t length = obj["length"].int_value();
+    if (debug >= 1) printf("binary %s with size %zu\n", name.c_str(), length);
+    g_programs[name] = cl_program_from_binary(context, device_id, (const uint8_t*)&buf[ptr], length);
+    ptr += length;
+  }
+
+  for (auto &obj : jdat["kernels"].array_items()) {
+    auto gws = obj["global_work_size"];
+    auto lws = obj["local_work_size"];
+    auto kk = shared_ptr<CLQueuedKernel>(new CLQueuedKernel(this));
+
+    kk->name = obj["name"].string_value();
+    kk->program = g_programs[kk->name];
+    kk->work_dim = obj["work_dim"].int_value();
+    for (int i = 0; i < kk->work_dim; i++) {
+      kk->global_work_size[i] = gws[i].int_value();
+      kk->local_work_size[i] = lws[i].int_value();
+    }
+    kk->num_args = obj["num_args"].int_value();
+    for (int i = 0; i < kk->num_args; i++) {
+      string arg = obj["args"].array_items()[i].string_value();
+      int arg_size = obj["args_size"].array_items()[i].int_value();
+      kk->args_size.push_back(arg_size);
+      if (arg_size == 8) {
+        cl_mem val = *(cl_mem*)(arg.data());
+        val = real_mem[val];
+        kk->args.push_back(string((char*)&val, sizeof(val)));
+      } else {
+        kk->args.push_back(arg);
+      }
+    }
+    kq.push_back(kk);
+  }
+
+  clFinish(command_queue);
+}

selfdrive/modeld/thneed/thneed.h

@@ -0,0 +1,133 @@
+#pragma once
+
+#ifndef __user
+#define __user __attribute__(())
+#endif
+
+#include <cstdint>
+#include <cstdlib>
+#include <memory>
+#include <string>
+#include <vector>
+
+#include <CL/cl.h>
+
+#include "third_party/linux/include/msm_kgsl.h"
+
+using namespace std;
+
+cl_int thneed_clSetKernelArg(cl_kernel kernel, cl_uint arg_index, size_t arg_size, const void *arg_value);
+
+namespace json11 {
+  class Json;
+}
+class Thneed;
+
+class GPUMalloc {
+  public:
+    GPUMalloc(int size, int fd);
+    ~GPUMalloc();
+    void *alloc(int size);
+  private:
+    uint64_t base;
+    int remaining;
+};
+
+class CLQueuedKernel {
+  public:
+    CLQueuedKernel(Thneed *lthneed) { thneed = lthneed; }
+    CLQueuedKernel(Thneed *lthneed,
+                   cl_kernel _kernel,
+                   cl_uint _work_dim,
+                   const size_t *_global_work_size,
+                   const size_t *_local_work_size);
+    cl_int exec();
+    void debug_print(bool verbose);
+    int get_arg_num(const char *search_arg_name);
+    cl_program program;
+    string name;
+    cl_uint num_args;
+    vector<string> arg_names;
+    vector<string> arg_types;
+    vector<string> args;
+    vector<int> args_size;
+    cl_kernel kernel = NULL;
+    json11::Json to_json() const;
+
+    cl_uint work_dim;
+    size_t global_work_size[3] = {0};
+    size_t local_work_size[3] = {0};
+  private:
+    Thneed *thneed;
+};
+
+class CachedIoctl {
+  public:
+    virtual void exec() {}
+};
+
+class CachedSync: public CachedIoctl {
+  public:
+    CachedSync(Thneed *lthneed, string ldata) { thneed = lthneed; data = ldata; }
+    void exec();
+  private:
+    Thneed *thneed;
+    string data;
+};
+
+class CachedCommand: public CachedIoctl {
+  public:
+    CachedCommand(Thneed *lthneed, struct kgsl_gpu_command *cmd);
+    void exec();
+  private:
+    void disassemble(int cmd_index);
+    struct kgsl_gpu_command cache;
+    unique_ptr<kgsl_command_object[]> cmds;
+    unique_ptr<kgsl_command_object[]> objs;
+    Thneed *thneed;
+    vector<shared_ptr<CLQueuedKernel> > kq;
+};
+
+class Thneed {
+  public:
+    Thneed(bool do_clinit=false, cl_context _context = NULL);
+    void stop();
+    void execute(float **finputs, float *foutput, bool slow=false);
+    void wait();
+
+    vector<cl_mem> input_clmem;
+    vector<void *> inputs;
+    vector<size_t> input_sizes;
+    cl_mem output = NULL;
+
+    cl_context context = NULL;
+    cl_command_queue command_queue;
+    cl_device_id device_id;
+    int context_id;
+
+    // protected?
+    bool record = false;
+    int debug;
+    int timestamp;
+
+#ifdef QCOM2
+    unique_ptr<GPUMalloc> ram;
+    vector<unique_ptr<CachedIoctl> > cmds;
+    int fd;
+#endif
+
+    // all CL kernels
+    void copy_inputs(float **finputs, bool internal=false);
+    void copy_output(float *foutput);
+    cl_int clexec();
+    vector<shared_ptr<CLQueuedKernel> > kq;
+
+    // pending CL kernels
+    vector<shared_ptr<CLQueuedKernel> > ckq;
+
+    // loading
+    void load(const char *filename);
+  private:
+    void clinit();
+};
+

selfdrive/modeld/thneed/thneed_common.cc

@@ -0,0 +1,216 @@
+#include "selfdrive/modeld/thneed/thneed.h"
+
+#include <cassert>
+#include <cstring>
+#include <map>
+
+#include "common/clutil.h"
+#include "common/timing.h"
+
+map<pair<cl_kernel, int>, string> g_args;
+map<pair<cl_kernel, int>, int> g_args_size;
+map<cl_program, string> g_program_source;
+
+void Thneed::stop() {
+  //printf("Thneed::stop: recorded %lu commands\n", cmds.size());
+  record = false;
+}
+
+void Thneed::clinit() {
+  device_id = cl_get_device_id(CL_DEVICE_TYPE_DEFAULT);
+  if (context == NULL) context = CL_CHECK_ERR(clCreateContext(NULL, 1, &device_id, NULL, NULL, &err));
+  //cl_command_queue_properties props[3] = {CL_QUEUE_PROPERTIES, CL_QUEUE_PROFILING_ENABLE, 0};
+  cl_command_queue_properties props[3] = {CL_QUEUE_PROPERTIES, 0, 0};
+  command_queue = CL_CHECK_ERR(clCreateCommandQueueWithProperties(context, device_id, props, &err));
+  printf("Thneed::clinit done\n");
+}
+
+cl_int Thneed::clexec() {
+  if (debug >= 1) printf("Thneed::clexec: running %lu queued kernels\n", kq.size());
+  for (auto &k : kq) {
+    if (record) ckq.push_back(k);
+    cl_int ret = k->exec();
+    assert(ret == CL_SUCCESS);
+  }
+  return clFinish(command_queue);
+}
+
+void Thneed::copy_inputs(float **finputs, bool internal) {
+  for (int idx = 0; idx < inputs.size(); ++idx) {
+    if (debug >= 1) printf("copying %lu -- %p -> %p (cl %p)\n", input_sizes[idx], finputs[idx], inputs[idx], input_clmem[idx]);
+
+    if (internal) {
+      // if it's internal, using memcpy is fine since the buffer sync is cached in the ioctl layer
+      if (finputs[idx] != NULL) memcpy(inputs[idx], finputs[idx], input_sizes[idx]);
+    } else {
+      if (finputs[idx] != NULL) CL_CHECK(clEnqueueWriteBuffer(command_queue, input_clmem[idx], CL_TRUE, 0, input_sizes[idx], finputs[idx], 0, NULL, NULL));
+    }
+  }
+}
+
+void Thneed::copy_output(float *foutput) {
+  if (output != NULL) {
+    size_t sz;
+    clGetMemObjectInfo(output, CL_MEM_SIZE, sizeof(sz), &sz, NULL);
+    if (debug >= 1) printf("copying %lu for output %p -> %p\n", sz, output, foutput);
+    CL_CHECK(clEnqueueReadBuffer(command_queue, output, CL_TRUE, 0, sz, foutput, 0, NULL, NULL));
+  } else {
+    printf("CAUTION: model output is NULL, does it have no outputs?\n");
+  }
+}
+
+// *********** CLQueuedKernel ***********
+
+CLQueuedKernel::CLQueuedKernel(Thneed *lthneed,
+                               cl_kernel _kernel,
+                               cl_uint _work_dim,
+                               const size_t *_global_work_size,
+                               const size_t *_local_work_size) {
+  thneed = lthneed;
+  kernel = _kernel;
+  work_dim = _work_dim;
+  assert(work_dim <= 3);
+  for (int i = 0; i < work_dim; i++) {
+    global_work_size[i] = _global_work_size[i];
+    local_work_size[i] = _local_work_size[i];
+  }
+
+  char _name[0x100];
+  clGetKernelInfo(kernel, CL_KERNEL_FUNCTION_NAME, sizeof(_name), _name, NULL);
+  name = string(_name);
+  clGetKernelInfo(kernel, CL_KERNEL_NUM_ARGS, sizeof(num_args), &num_args, NULL);
+
+  // get args
+  for (int i = 0; i < num_args; i++) {
+    char arg_name[0x100] = {0};
+    clGetKernelArgInfo(kernel, i, CL_KERNEL_ARG_NAME, sizeof(arg_name), arg_name, NULL);
+    arg_names.push_back(string(arg_name));
+    clGetKernelArgInfo(kernel, i, CL_KERNEL_ARG_TYPE_NAME, sizeof(arg_name), arg_name, NULL);
+    arg_types.push_back(string(arg_name));
+
+    args.push_back(g_args[make_pair(kernel, i)]);
+    args_size.push_back(g_args_size[make_pair(kernel, i)]);
+  }
+
+  // get program
+  clGetKernelInfo(kernel, CL_KERNEL_PROGRAM, sizeof(program), &program, NULL);
+}
+
+int CLQueuedKernel::get_arg_num(const char *search_arg_name) {
+  for (int i = 0; i < num_args; i++) {
+    if (arg_names[i] == search_arg_name) return i;
+  }
+  printf("failed to find %s in %s\n", search_arg_name, name.c_str());
+  assert(false);
+}
+
+cl_int CLQueuedKernel::exec() {
+  if (kernel == NULL) {
+    kernel = clCreateKernel(program, name.c_str(), NULL);
+    arg_names.clear();
+    arg_types.clear();
+
+    for (int j = 0; j < num_args; j++) {
+      char arg_name[0x100] = {0};
+      clGetKernelArgInfo(kernel, j, CL_KERNEL_ARG_NAME, sizeof(arg_name), arg_name, NULL);
+      arg_names.push_back(string(arg_name));
+      clGetKernelArgInfo(kernel, j, CL_KERNEL_ARG_TYPE_NAME, sizeof(arg_name), arg_name, NULL);
+      arg_types.push_back(string(arg_name));
+
+      cl_int ret;
+      if (args[j].size() != 0) {
+        assert(args[j].size() == args_size[j]);
+        ret = thneed_clSetKernelArg(kernel, j, args[j].size(), args[j].data());
+      } else {
+        ret = thneed_clSetKernelArg(kernel, j, args_size[j], NULL);
+      }
+      assert(ret == CL_SUCCESS);
+    }
+  }
+
+  if (thneed->debug >= 1) {
+    debug_print(thneed->debug >= 2);
+  }
+
+  return clEnqueueNDRangeKernel(thneed->command_queue,
+    kernel, work_dim, NULL, global_work_size, local_work_size, 0, NULL, NULL);
+}
+
+void CLQueuedKernel::debug_print(bool verbose) {
+  printf("%p %56s -- ", kernel, name.c_str());
+  for (int i = 0; i < work_dim; i++) {
+    printf("%4zu ", global_work_size[i]);
+  }
+  printf(" -- ");
+  for (int i = 0; i < work_dim; i++) {
+    printf("%4zu ", local_work_size[i]);
+  }
+  printf("\n");
+
+  if (verbose) {
+    for (int i = 0; i < num_args; i++) {
+      string arg = args[i];
+      printf("  %s %s", arg_types[i].c_str(), arg_names[i].c_str());
+      void *arg_value = (void*)arg.data();
+      int arg_size = arg.size();
+      if (arg_size == 0) {
+        printf(" (size) %d", args_size[i]);
+      } else if (arg_size == 1) {
+        printf(" = %d", *((char*)arg_value));
+      } else if (arg_size == 2) {
+        printf(" = %d", *((short*)arg_value));
+      } else if (arg_size == 4) {
+        if (arg_types[i] == "float") {
+          printf(" = %f", *((float*)arg_value));
+        } else {
+          printf(" = %d", *((int*)arg_value));
+        }
+      } else if (arg_size == 8) {
+        cl_mem val = (cl_mem)(*((uintptr_t*)arg_value));
+        printf(" = %p", val);
+        if (val != NULL) {
+          cl_mem_object_type obj_type;
+          clGetMemObjectInfo(val, CL_MEM_TYPE, sizeof(obj_type), &obj_type, NULL);
+          if (arg_types[i] == "image2d_t" || arg_types[i] == "image1d_t" || obj_type == CL_MEM_OBJECT_IMAGE2D) {
+            cl_image_format format;
+            size_t width, height, depth, array_size, row_pitch, slice_pitch;
+            cl_mem buf;
+            clGetImageInfo(val, CL_IMAGE_FORMAT, sizeof(format), &format, NULL);
+            assert(format.image_channel_order == CL_RGBA);
+            assert(format.image_channel_data_type == CL_HALF_FLOAT || format.image_channel_data_type == CL_FLOAT);
+            clGetImageInfo(val, CL_IMAGE_WIDTH, sizeof(width), &width, NULL);
+            clGetImageInfo(val, CL_IMAGE_HEIGHT, sizeof(height), &height, NULL);
+            clGetImageInfo(val, CL_IMAGE_ROW_PITCH, sizeof(row_pitch), &row_pitch, NULL);
+            clGetImageInfo(val, CL_IMAGE_DEPTH, sizeof(depth), &depth, NULL);
+            clGetImageInfo(val, CL_IMAGE_ARRAY_SIZE, sizeof(array_size), &array_size, NULL);
+            clGetImageInfo(val, CL_IMAGE_SLICE_PITCH, sizeof(slice_pitch), &slice_pitch, NULL);
+            assert(depth == 0);
+            assert(array_size == 0);
+            assert(slice_pitch == 0);
+
+            clGetImageInfo(val, CL_IMAGE_BUFFER, sizeof(buf), &buf, NULL);
+            size_t sz = 0;
+            if (buf != NULL) clGetMemObjectInfo(buf, CL_MEM_SIZE, sizeof(sz), &sz, NULL);
+            printf(" image %zu x %zu rp %zu @ %p buffer %zu", width, height, row_pitch, buf, sz);
+          } else {
+            size_t sz;
+            clGetMemObjectInfo(val, CL_MEM_SIZE, sizeof(sz), &sz, NULL);
+            printf(" buffer %zu", sz);
+          }
+        }
+      }
+      printf("\n");
+    }
+  }
+}
+
+cl_int thneed_clSetKernelArg(cl_kernel kernel, cl_uint arg_index, size_t arg_size, const void *arg_value) {
+  g_args_size[make_pair(kernel, arg_index)] = arg_size;
+  if (arg_value != NULL) {
+    g_args[make_pair(kernel, arg_index)] = string((char*)arg_value, arg_size);
+  } else {
+    g_args[make_pair(kernel, arg_index)] = string("");
+  }
+  cl_int ret = clSetKernelArg(kernel, arg_index, arg_size, arg_value);
+  return ret;
+}

selfdrive/modeld/thneed/thneed_pc.cc

@@ -0,0 +1,32 @@
+#include "selfdrive/modeld/thneed/thneed.h"
+
+#include <cassert>
+
+#include "common/clutil.h"
+#include "common/timing.h"
+
+Thneed::Thneed(bool do_clinit, cl_context _context) {
+  context = _context;
+  if (do_clinit) clinit();
+  char *thneed_debug_env = getenv("THNEED_DEBUG");
+  debug = (thneed_debug_env != NULL) ? atoi(thneed_debug_env) : 0;
+}
+
+void Thneed::execute(float **finputs, float *foutput, bool slow) {
+  uint64_t tb, te;
+  if (debug >= 1) tb = nanos_since_boot();
+
+  // ****** copy inputs
+  copy_inputs(finputs);
+
+  // ****** run commands
+  clexec();
+
+  // ****** copy outputs
+  copy_output(foutput);
+
+  if (debug >= 1) {
+    te = nanos_since_boot();
+    printf("model exec in %lu us\n", (te-tb)/1000);
+  }
+}

selfdrive/modeld/thneed/thneed_qcom2.cc

@@ -0,0 +1,258 @@
+#include "selfdrive/modeld/thneed/thneed.h"
+
+#include <dlfcn.h>
+#include <sys/mman.h>
+
+#include <cassert>
+#include <cerrno>
+#include <cstring>
+#include <map>
+#include <string>
+
+#include "common/clutil.h"
+#include "common/timing.h"
+
+Thneed *g_thneed = NULL;
+int g_fd = -1;
+
+void hexdump(uint8_t *d, int len) {
+  assert((len%4) == 0);
+  printf("  dumping %p len 0x%x\n", d, len);
+  for (int i = 0; i < len/4; i++) {
+    if (i != 0 && (i%0x10) == 0) printf("\n");
+    printf("%8x ", d[i]);
+  }
+  printf("\n");
+}
+
+// *********** ioctl interceptor ***********
+
+extern "C" {
+
+int (*my_ioctl)(int filedes, unsigned long request, void *argp) = NULL;
+#undef ioctl
+int ioctl(int filedes, unsigned long request, void *argp) {
+  request &= 0xFFFFFFFF;  // needed on QCOM2
+  if (my_ioctl == NULL) my_ioctl = reinterpret_cast<decltype(my_ioctl)>(dlsym(RTLD_NEXT, "ioctl"));
+  Thneed *thneed = g_thneed;
+
+  // save the fd
+  if (request == IOCTL_KGSL_GPUOBJ_ALLOC) g_fd = filedes;
+
+  // note that this runs always, even without a thneed object
+  if (request == IOCTL_KGSL_DRAWCTXT_CREATE) {
+    struct kgsl_drawctxt_create *create = (struct kgsl_drawctxt_create *)argp;
+    create->flags &= ~KGSL_CONTEXT_PRIORITY_MASK;
+    create->flags |= 6 << KGSL_CONTEXT_PRIORITY_SHIFT;   // priority from 1-15, 1 is max priority
+    printf("IOCTL_KGSL_DRAWCTXT_CREATE: creating context with flags 0x%x\n", create->flags);
+  }
+
+  if (thneed != NULL) {
+    if (request == IOCTL_KGSL_GPU_COMMAND) {
+      struct kgsl_gpu_command *cmd = (struct kgsl_gpu_command *)argp;
+      if (thneed->record) {
+        thneed->timestamp = cmd->timestamp;
+        thneed->context_id = cmd->context_id;
+        thneed->cmds.push_back(unique_ptr<CachedCommand>(new CachedCommand(thneed, cmd)));
+      }
+      if (thneed->debug >= 1) {
+        printf("IOCTL_KGSL_GPU_COMMAND(%2zu): flags: 0x%lx    context_id: %u  timestamp: %u  numcmds: %d  numobjs: %d\n",
+            thneed->cmds.size(),
+            cmd->flags,
+            cmd->context_id, cmd->timestamp, cmd->numcmds, cmd->numobjs);
+      }
+    } else if (request == IOCTL_KGSL_GPUOBJ_SYNC) {
+      struct kgsl_gpuobj_sync *cmd = (struct kgsl_gpuobj_sync *)argp;
+      struct kgsl_gpuobj_sync_obj *objs = (struct kgsl_gpuobj_sync_obj *)(cmd->objs);
+
+      if (thneed->debug >= 2) {
+        printf("IOCTL_KGSL_GPUOBJ_SYNC count:%d ", cmd->count);
+        for (int i = 0; i < cmd->count; i++) {
+          printf(" -- offset:0x%lx len:0x%lx id:%d op:%d  ", objs[i].offset, objs[i].length, objs[i].id, objs[i].op);
+        }
+        printf("\n");
+      }
+
+      if (thneed->record) {
+        thneed->cmds.push_back(unique_ptr<CachedSync>(new
+              CachedSync(thneed, string((char *)objs, sizeof(struct kgsl_gpuobj_sync_obj)*cmd->count))));
+      }
+    } else if (request == IOCTL_KGSL_DEVICE_WAITTIMESTAMP_CTXTID) {
+      struct kgsl_device_waittimestamp_ctxtid *cmd = (struct kgsl_device_waittimestamp_ctxtid *)argp;
+      if (thneed->debug >= 1) {
+        printf("IOCTL_KGSL_DEVICE_WAITTIMESTAMP_CTXTID: context_id: %d  timestamp: %d  timeout: %d\n",
+            cmd->context_id, cmd->timestamp, cmd->timeout);
+      }
+    } else if (request == IOCTL_KGSL_SETPROPERTY) {
+      if (thneed->debug >= 1) {
+        struct kgsl_device_getproperty *prop = (struct kgsl_device_getproperty *)argp;
+        printf("IOCTL_KGSL_SETPROPERTY: 0x%x sizebytes:%zu\n", prop->type, prop->sizebytes);
+        if (thneed->debug >= 2) {
+          hexdump((uint8_t *)prop->value, prop->sizebytes);
+          if (prop->type == KGSL_PROP_PWR_CONSTRAINT) {
+            struct kgsl_device_constraint *constraint = (struct kgsl_device_constraint *)prop->value;
+            hexdump((uint8_t *)constraint->data, constraint->size);
+          }
+        }
+      }
+    } else if (request == IOCTL_KGSL_DRAWCTXT_CREATE || request == IOCTL_KGSL_DRAWCTXT_DESTROY) {
+      // this happens
+    } else if (request == IOCTL_KGSL_GPUOBJ_ALLOC || request == IOCTL_KGSL_GPUOBJ_FREE) {
+      // this happens
+    } else {
+      if (thneed->debug >= 1) {
+        printf("other ioctl %lx\n", request);
+      }
+    }
+  }
+
+  int ret = my_ioctl(filedes, request, argp);
+  // NOTE: This error message goes into stdout and messes up pyenv
+  // if (ret != 0) printf("ioctl returned %d with errno %d\n", ret, errno);
+  return ret;
+}
+
+}
+
+// *********** GPUMalloc ***********
+
+GPUMalloc::GPUMalloc(int size, int fd) {
+  struct kgsl_gpuobj_alloc alloc;
+  memset(&alloc, 0, sizeof(alloc));
+  alloc.size = size;
+  alloc.flags = 0x10000a00;
+  ioctl(fd, IOCTL_KGSL_GPUOBJ_ALLOC, &alloc);
+  void *addr = mmap64(NULL, alloc.mmapsize, 0x3, 0x1, fd, alloc.id*0x1000);
+  assert(addr != MAP_FAILED);
+
+  base = (uint64_t)addr;
+  remaining = size;
+}
+
+GPUMalloc::~GPUMalloc() {
+  // TODO: free the GPU malloced area
+}
+
+void *GPUMalloc::alloc(int size) {
+  void *ret = (void*)base;
+  size = (size+0xff) & (~0xFF);
+  assert(size <= remaining);
+  remaining -= size;
+  base += size;
+  return ret;
+}
+
+// *********** CachedSync, at the ioctl layer ***********
+
+void CachedSync::exec() {
+  struct kgsl_gpuobj_sync cmd;
+
+  cmd.objs = (uint64_t)data.data();
+  cmd.obj_len = data.length();
+  cmd.count = data.length() / sizeof(struct kgsl_gpuobj_sync_obj);
+
+  int ret = ioctl(thneed->fd, IOCTL_KGSL_GPUOBJ_SYNC, &cmd);
+  assert(ret == 0);
+}
+
+// *********** CachedCommand, at the ioctl layer ***********
+
+CachedCommand::CachedCommand(Thneed *lthneed, struct kgsl_gpu_command *cmd) {
+  thneed = lthneed;
+  assert(cmd->numsyncs == 0);
+
+  memcpy(&cache, cmd, sizeof(cache));
+
+  if (cmd->numcmds > 0) {
+    cmds = make_unique<struct kgsl_command_object[]>(cmd->numcmds);
+    memcpy(cmds.get(), (void *)cmd->cmdlist, sizeof(struct kgsl_command_object)*cmd->numcmds);
+    cache.cmdlist = (uint64_t)cmds.get();
+    for (int i = 0; i < cmd->numcmds; i++) {
+      void *nn = thneed->ram->alloc(cmds[i].size);
+      memcpy(nn, (void*)cmds[i].gpuaddr, cmds[i].size);
+      cmds[i].gpuaddr = (uint64_t)nn;
+    }
+  }
+
+  if (cmd->numobjs > 0) {
+    objs = make_unique<struct kgsl_command_object[]>(cmd->numobjs);
+    memcpy(objs.get(), (void *)cmd->objlist, sizeof(struct kgsl_command_object)*cmd->numobjs);
+    cache.objlist = (uint64_t)objs.get();
+    for (int i = 0; i < cmd->numobjs; i++) {
+      void *nn = thneed->ram->alloc(objs[i].size);
+      memset(nn, 0, objs[i].size);
+      objs[i].gpuaddr = (uint64_t)nn;
+    }
+  }
+
+  kq = thneed->ckq;
+  thneed->ckq.clear();
+}
+
+void CachedCommand::exec() {
+  cache.timestamp = ++thneed->timestamp;
+  int ret = ioctl(thneed->fd, IOCTL_KGSL_GPU_COMMAND, &cache);
+
+  if (thneed->debug >= 1) printf("CachedCommand::exec got %d\n", ret);
+
+  if (thneed->debug >= 2) {
+    for (auto &it : kq) {
+      it->debug_print(false);
+    }
+  }
+
+  assert(ret == 0);
+}
+
+// *********** Thneed ***********
+
+Thneed::Thneed(bool do_clinit, cl_context _context) {
+  // TODO: QCOM2 actually requires a different context
+  //context = _context;
+  if (do_clinit) clinit();
+  assert(g_fd != -1);
+  fd = g_fd;
+  ram = make_unique<GPUMalloc>(0x80000, fd);
+  timestamp = -1;
+  g_thneed = this;
+  char *thneed_debug_env = getenv("THNEED_DEBUG");
+  debug = (thneed_debug_env != NULL) ? atoi(thneed_debug_env) : 0;
+}
+
+void Thneed::wait() {
+  struct kgsl_device_waittimestamp_ctxtid wait;
+  wait.context_id = context_id;
+  wait.timestamp = timestamp;
+  wait.timeout = -1;
+
+  uint64_t tb = nanos_since_boot();
+  int wret = ioctl(fd, IOCTL_KGSL_DEVICE_WAITTIMESTAMP_CTXTID, &wait);
+  uint64_t te = nanos_since_boot();
+
+  if (debug >= 1) printf("wait %d after %lu us\n", wret, (te-tb)/1000);
+}
+
+void Thneed::execute(float **finputs, float *foutput, bool slow) {
+  uint64_t tb, te;
+  if (debug >= 1) tb = nanos_since_boot();
+
+  // ****** copy inputs
+  copy_inputs(finputs, true);
+
+  // ****** run commands
+  int i = 0;
+  for (auto &it : cmds) {
+    ++i;
+    if (debug >= 1) printf("run %2d @ %7lu us: ", i, (nanos_since_boot()-tb)/1000);
+    it->exec();
+    if ((i == cmds.size()) || slow) wait();
+  }
+
+  // ****** copy outputs
+  copy_output(foutput);
+
+  if (debug >= 1) {
+    te = nanos_since_boot();
+    printf("model exec in %lu us\n", (te-tb)/1000);
+  }
+}

selfdrive/pandad/panda_comms.cc

@@ -36,7 +36,7 @@ PandaUsbHandle::PandaUsbHandle(std::string serial) : PandaCommsHandle(serial) {
   for (size_t i = 0; i < num_devices; ++i) {
     libusb_device_descriptor desc;
     libusb_get_device_descriptor(dev_list[i], &desc);
-    if (desc.idVendor == 0x3801 && desc.idProduct == 0xddcc) {
+    if (desc.idVendor == 0xbbaa && desc.idProduct == 0xddcc) {
       int ret = libusb_open(dev_list[i], &dev_handle);
       if (dev_handle == NULL || ret < 0) { goto fail; }
 
@@ -110,7 +110,7 @@ std::vector<std::string> PandaUsbHandle::list() {
     libusb_device *device = dev_list[i];
     libusb_device_descriptor desc;
     libusb_get_device_descriptor(device, &desc);
-    if (desc.idVendor == 0x3801 && desc.idProduct == 0xddcc) {
+    if (desc.idVendor == 0xbbaa && desc.idProduct == 0xddcc) {
       libusb_device_handle *handle = NULL;
       int ret = libusb_open(device, &handle);
       if (ret < 0) { goto finish; }

selfdrive/pandad/pandad.cc

@@ -416,7 +416,6 @@ void process_peripheral_state(Panda *panda, PubMaster *pm, bool no_fan_control)
 
     if (ir_pwr != prev_ir_pwr || sm.frame % 100 == 0 || ir_pwr >= 50.0) {
       panda->set_ir_pwr(ir_pwr);
-      Hardware::set_ir_power(ir_pwr);
       prev_ir_pwr = ir_pwr;
     }
   }

selfdrive/test/process_replay/.gitignore

@@ -1 +1 @@
-!fakedata/*
+fakedata/