Conversation
* feat(all): adaption of apollo on cuda 12+trt10 * fix: fix build * fix: remove vtk * feat: remove paddle deps * fix: remove vtk * fix: fix ncut error * feat: update docker due to miss lib/fastrtps * chore: improve image processor * fix: check tensorrt --------- Co-authored-by: Your Name <you@example.com>
Contributor
There was a problem hiding this comment.
Pull request overview
This PR upgrades the build/tooling stack around TensorRT (towards TRT 10), adjusts CUDA/cuDNN detection to be more conditional, and removes several third-party integrations (VTK, PaddleInference, and associated perception ops) while adding a new NCut debug-proto publishing path.
Changes:
- Update TensorRT integration for TRT>=10 APIs (engine build/serialize and runtime execution) and adjust TensorRT repository configuration for v10 library/header layout.
- Make cuDNN/CUPTI handling more conditional in CUDA auto-configuration and bootstrap config flow.
- Remove VTK + PaddleInference (+ CADDN/CenterPoint infer ops) dependencies and wire in NCut debug info publishing + Docker environment updates (Ubuntu 22.04/aarch64 stack).
Reviewed changes
Copilot reviewed 96 out of 100 changed files in this pull request and generated 7 comments.
Show a summary per file
| File | Description |
|---|---|
| tools/workspace.bzl | Removes loading/initializing removed third_party repos (VTK, PaddleInference, infer ops). |
| tools/bootstrap.py | Updates CUDA compute caps defaults and changes CUDA/cuDNN detection flow to allow “no cuDNN” scenarios. |
| third_party/vtk/vtk_configure.bzl | Deletes VTK repository rule implementation. |
| third_party/vtk/init.bzl | Deletes VTK init macro. |
| third_party/vtk/BUILD.tpl | Deletes VTK BUILD template. |
| third_party/vtk/BUILD | Deletes VTK install packaging targets. |
| third_party/tensorrt/tensorrt_configure.bzl | Adjusts TRT libs/headers selection for TRT10 (e.g., drop nvparsers / parser headers). |
| third_party/paddleinference/workspace.bzl | Deletes PaddleInference repo definition. |
| third_party/paddleinference/paddleinference-x86_64.BUILD | Deletes PaddleInference x86_64 Bazel build. |
| third_party/paddleinference/paddleinference-aarch64.BUILD | Deletes PaddleInference aarch64 Bazel build. |
| third_party/paddleinference/init.bzl | Deletes PaddleInference init macro. |
| third_party/paddleinference/BUILD | Deletes PaddleInference install packaging targets. |
| third_party/gpus/cuda_configure.bzl | Makes cuDNN/CUPTI discovery and copying conditional; adds placeholder behavior when disabled. |
| third_party/centerpoint_infer_op/workspace.bzl | Deletes CenterPoint infer op repo definition. |
| third_party/centerpoint_infer_op/init.bzl | Deletes CenterPoint infer op init macro. |
| third_party/centerpoint_infer_op/BUILD | Deletes CenterPoint infer op install packaging targets. |
| third_party/caddn_infer_op/workspace.bzl | Deletes CADDN infer op repo definition. |
| third_party/caddn_infer_op/init.bzl | Deletes CADDN infer op init macro. |
| third_party/caddn_infer_op/BUILD | Deletes CADDN infer op install packaging targets. |
| third_party/ad_rss_lib/workspace.bzl | Applies a patch during ad_rss_lib fetch. |
| third_party/ad_rss_lib/ad_rss.patch | Adds missing standard include to fix build issue in upstream header. |
| third_party/ACKNOWLEDGEMENT.txt | Removes VTK acknowledgement entry. |
| modules/tools/visualizer/plane.cc | Formatting-only adjustments. |
| modules/perception/pipeline/pipeline.cc | Comments out BEV/CADDN/CenterPoint stage wiring. |
| modules/perception/pipeline/BUILD | Comments out deps for removed/disabled stages. |
| modules/perception/lidar/lib/detector/point_pillars_detection/point_pillars.cc | Updates TensorRT explicit-batch flag usage and adds TRT10 build/deserialize path. |
| modules/perception/lidar/lib/detector/ncut_segmentation/proto/ncut_param.proto | Adds publish_debug_info flag. |
| modules/perception/lidar/lib/detector/ncut_segmentation/proto/ncut_debug.proto | New proto for NCut debug pointcloud publishing. |
| modules/perception/lidar/lib/detector/ncut_segmentation/proto/BUILD | Adds cc/py proto targets for the new debug proto. |
| modules/perception/lidar/lib/detector/ncut_segmentation/ncut_segmentation.h | Adds debug writer/node members and helper for proto pointcloud filling. |
| modules/perception/lidar/lib/detector/ncut_segmentation/ncut_segmentation.cc | Publishes NCut debug info (topic writer) and removes old DEBUG_NCUT viewer code. |
| modules/perception/lidar/lib/detector/ncut_segmentation/BUILD | Drops VTK dep; adds new debug proto dependency. |
| modules/perception/lidar/lib/detector/center_point_detection/BUILD | Comments out center_point_detection build target. |
| modules/perception/lidar/app/BUILD | Comments out center_point_detection dependency. |
| modules/perception/inference/tensorrt/rt_net.h | Adds tensor-name -> buffer-index map for TRT>=10 binding. |
| modules/perception/inference/tensorrt/rt_net.cc | Adds TRT>=10 API paths (Nd ops, serialized build, enqueueV3 + setTensorAddress). |
| modules/perception/inference/tensorrt/rt_legacy.h | Adapts legacy dim helpers for TRT10 (int64 dims). |
| modules/perception/inference/tensorrt/rt_common_without_protobuf.h | Removes NvCaffeParser/cudnn/miopen includes; widens legacy include guard. |
| modules/perception/inference/tensorrt/rt_common.h | Removes NvCaffeParser/cudnn/miopen includes; widens legacy include guard. |
| modules/perception/inference/tensorrt/rt_common.cc | Widens legacy include guard. |
| modules/perception/inference/tensorrt/plugins/softmax_plugin.h | Adds cuDNN header include and adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/softmax_plugin.cu | Adjusts version guards / formatting. |
| modules/perception/inference/tensorrt/plugins/slice_plugin.h | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/slice_plugin.cu | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/rpn_proposal_ssd_plugin.h | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/rpn_proposal_ssd_plugin.cu | Adjusts version guards and formatting. |
| modules/perception/inference/tensorrt/plugins/rcnn_proposal_plugin.h | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/rcnn_proposal_plugin.cu | Adds <cfloat> include; adjusts version guards and formatting. |
| modules/perception/inference/tensorrt/plugins/leakyReLU_plugin.h | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/leakyReLU_plugin.cu | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/dfmb_psroi_align_plugin.h | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/dfmb_psroi_align_plugin.cu | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/argmax_plugin.h | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/argmax_plugin.cu | Adjusts version guards. |
| modules/perception/inference/tensorrt/plugins/BUILD | Updates deps and target list ordering; removes cudnn dep. |
| modules/perception/inference/tensorrt/entropy_calibrator.cc | Widens legacy include guard. |
| modules/perception/inference/tensorrt/batch_stream.h | Widens legacy include guard. |
| modules/perception/inference/tensorrt/batch_stream.cc | Widens legacy include guard; adds include and int cast fix. |
| modules/perception/inference/tensorrt/BUILD | Tweaks deps to include local TensorRT and remove cudnn_header. |
| modules/perception/inference/paddlepaddle/paddle_net.h | Deletes PaddleNet implementation header. |
| modules/perception/inference/paddlepaddle/paddle_net.cc | Deletes PaddleNet implementation source. |
| modules/perception/inference/paddlepaddle/BUILD | Deletes PaddleNet Bazel target. |
| modules/perception/inference/onnx/onnx_obstacle_detector.cc | Updates explicit-batch flag usage and adds TRT10 build/deserialize path. |
| modules/perception/inference/inference_factory.cc | Removes PaddleNet factory path. |
| modules/perception/inference/BUILD | Removes paddlepaddle inference dep. |
| modules/perception/camera/lib/obstacle/detector/caddn/BUILD | Comments out caddn_obstacle_detector build target. |
| modules/perception/camera/lib/obstacle/detector/bev_detection/BUILD | Comments out bev_obstacle_detector build target. |
| modules/perception/camera/app/BUILD | Comments out caddn detector dep. |
| modules/map/pnc_map/BUILD | Removes commented-out map data entries. |
| modules/map/BUILD | Removes commented-out install dep. |
| modules/drivers/camera/backend/image_processor.cc | Avoids unnecessary memcpy for YUYV conversion; adds input validation. |
| docker/scripts/dev_start.sh | Selects aarch64 docker tag by Ubuntu release; updates next-steps text. |
| docker/build/rcfiles/sources.list.tsinghua.x86_64.ubuntu.20.04 | Deletes duplicated sources list file. |
| docker/build/rcfiles/sources.list.tsinghua.aarch64.ubuntu.20.04 | Deletes duplicated sources list file. |
| docker/build/rcfiles/sources.list.local.x86_64.ubuntu.20.04 | Deletes local-only sources list file. |
| docker/build/rcfiles/sources.list.local.aarch64.ubuntu.20.04 | Deletes local-only sources list file. |
| docker/build/rcfiles/sources.list.cn.x86_64.ubuntu.20.04 | Replaces indirection with full sources list content. |
| docker/build/rcfiles/sources.list.cn.x86_64 | Replaces indirection with full sources list content. |
| docker/build/rcfiles/sources.list.cn.aarch64.ubuntu.22.04 | Adds Jammy sources list for aarch64. |
| docker/build/rcfiles/sources.list.cn.aarch64.ubuntu.20.04 | Replaces indirection with full sources list content. |
| docker/build/rcfiles/sources.list.cn.aarch64 | Replaces indirection with full sources list content. |
| docker/build/installers/install_vtk.sh | Deletes VTK installer. |
| docker/build/installers/install_modules_base.sh | Stops installing VTK. |
| docker/build/installers/install_minimal_environment.sh | Refactors minimal env install steps and pip mirror setup behavior. |
| docker/build/installers/install_magma.sh | Changes default install mode and updates MAGMA version. |
| docker/build/installers/install_llvm_clang.sh | Installs default clang packages instead of clang-10. |
| docker/build/installers/install_libtorch.sh | Major refactor; adds aarch64 wheel/source fallback and checksum validation. |
| docker/build/installers/install_geo_adjustment.sh | Simplifies apt-source selection and standardizes warning function usage. |
| docker/build/installers/install_ffmpeg.sh | Removes static pkg-config flag from configure invocation. |
| docker/build/installers/install_fast-rtps.sh | Updates git branch/tag reference. |
| docker/build/installers/install_cyber_deps.sh | Passes explicit build mode to fast-rtps installer. |
| docker/build/installers/install_cmake.sh | Improves robustness for replacing system cmake binary/symlink. |
| docker/build/dev.aarch64.nx.dockerfile | Adds a new aarch64 dev image based on CUDA12.6/cuDNN9/TRT10 (22.04). |
| docker/build/build_docker.sh | Switches default Ubuntu LTS to 22.04; updates aarch64 CUDA/cuDNN/TRT versions and build args. |
| docker/build/base.aarch64.l4t.dockerfile | Adds L4T base image for TRT10.3 + cuDNN 9 install flow. |
| BUILD | Removes VTK from install aggregation. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
* feat: add gnss conf tool (#153) * feat: add wheelos bazel registry url (#161) * 166 feature independent dreamview frontend (#168) * feat: Independent dreamview frontend * chore: remove frontend directory from git tracking * fix: fix dkit protocol (#134) * fix: fix dkit protocol * fix: call can_sender_->Update() in Emergency() * feat: add remote debugging tool (#178) * 177 feature remote debugging tool (#179) * feat: add remote debugging tool * chore: improve whl-remote * feat: Add auto-start feature (#84) * feat: Add auto-start feature * chore: Optimized implementation * fix: fix start/enter issues of whl.sh Signed-off-by: Pride Leong <lykling.lyk@gmail.com> * chore(docker/scripts/whl.sh): support custom names and tags Signed-off-by: Pride Leong <lykling.lyk@gmail.com> --------- Signed-off-by: Pride Leong <lykling.lyk@gmail.com> Co-authored-by: Pride Leong <lykling.lyk@gmail.com> --------- Signed-off-by: Pride Leong <lykling.lyk@gmail.com> Co-authored-by: zero <daohu527@gmail.com>
Signed-off-by: Pride Leong <lykling.lyk@gmail.com>
Signed-off-by: Pride Leong <lykling.lyk@gmail.com>
- replace local_config_cuda with rules_cuda - add cudnn and tenesorrt extension Signed-off-by: Pride Leong <lykling.lyk@gmail.com>
Signed-off-by: Pride Leong <lykling.lyk@gmail.com>
Signed-off-by: Pride Leong <lykling.lyk@gmail.com>
Signed-off-by: Pride Leong <lykling.lyk@gmail.com>
added 7 commits
March 3, 2026 14:38
…tpack 6.2 with specify O2 optimization
This was referenced Mar 8, 2026
Closed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Upgrade to Jetpack 6.2