Problem Description
Hello, I am encountering an issue with the ESS Engine provided by NVIDIA. The engine fails to generate. Additionally, my machine is limited to using JetPack 6.0 (JP6.0) and Isaac ROS release-3.1. Below are the details:
When executing the following command:
ros2 run isaac_ros_ess_models_install install_ess_models.sh --eula
2024-12-17 10:18:03 (10.7 MB/s) - ‘/workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx.tar.gz’ saved [129672015/129672015]
dnn_stereo_disparity_v4.1.0_onnx/
dnn_stereo_disparity_v4.1.0_onnx/ess.onnx
dnn_stereo_disparity_v4.1.0_onnx/plugins/
dnn_stereo_disparity_v4.1.0_onnx/plugins/aarch64/
dnn_stereo_disparity_v4.1.0_onnx/plugins/aarch64/ess_plugins.so
dnn_stereo_disparity_v4.1.0_onnx/plugins/x86_64/
dnn_stereo_disparity_v4.1.0_onnx/plugins/x86_64/ess_plugins.so
dnn_stereo_disparity_v4.1.0_onnx/light_ess.onnx
Converting ESS onnx file to engine file.
&&&& RUNNING TensorRT.trtexec [TensorRT v8602] # /usr/src/tensorrt/bin/trtexec --onnx=/workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/ess.onnx --saveEngine=/workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/ess.engine --fp16 --staticPlugins=/workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/plugins/aarch64/ess_plugins.so
[12/17/2024-10:18:04] [I] === Model Options ===
[12/17/2024-10:18:04] [I] Format: ONNX
[12/17/2024-10:18:04] [I] Model: /workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/ess.onnx
[12/17/2024-10:18:04] [I] Output:
[12/17/2024-10:18:04] [I] === Build Options ===
[12/17/2024-10:18:04] [I] Max batch: explicit batch
[12/17/2024-10:18:04] [I] Memory Pools: workspace: default, dlaSRAM: default, dlaLocalDRAM: default, dlaGlobalDRAM: default
[12/17/2024-10:18:04] [I] minTiming: 1
[12/17/2024-10:18:04] [I] avgTiming: 8
[12/17/2024-10:18:04] [I] Precision: FP32+FP16
[12/17/2024-10:18:04] [I] LayerPrecisions:
[12/17/2024-10:18:04] [I] Layer Device Types:
[12/17/2024-10:18:04] [I] Calibration:
[12/17/2024-10:18:04] [I] Refit: Disabled
[12/17/2024-10:18:04] [I] Version Compatible: Disabled
[12/17/2024-10:18:04] [I] ONNX Native InstanceNorm: Disabled
[12/17/2024-10:18:04] [I] TensorRT runtime: full
[12/17/2024-10:18:04] [I] Lean DLL Path:
[12/17/2024-10:18:04] [I] Tempfile Controls: { in_memory: allow, temporary: allow }
[12/17/2024-10:18:04] [I] Exclude Lean Runtime: Disabled
[12/17/2024-10:18:04] [I] Sparsity: Disabled
[12/17/2024-10:18:04] [I] Safe mode: Disabled
[12/17/2024-10:18:04] [I] Build DLA standalone loadable: Disabled
[12/17/2024-10:18:04] [I] Allow GPU fallback for DLA: Disabled
[12/17/2024-10:18:04] [I] DirectIO mode: Disabled
[12/17/2024-10:18:04] [I] Restricted mode: Disabled
[12/17/2024-10:18:04] [I] Skip inference: Disabled
[12/17/2024-10:18:04] [I] Save engine: /workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/ess.engine
[12/17/2024-10:18:04] [I] Load engine:
[12/17/2024-10:18:04] [I] Profiling verbosity: 0
[12/17/2024-10:18:04] [I] Tactic sources: Using default tactic sources
[12/17/2024-10:18:04] [I] timingCacheMode: local
[12/17/2024-10:18:04] [I] timingCacheFile:
[12/17/2024-10:18:04] [I] Heuristic: Disabled
[12/17/2024-10:18:04] [I] Preview Features: Use default preview flags.
[12/17/2024-10:18:04] [I] MaxAuxStreams: -1
[12/17/2024-10:18:04] [I] BuilderOptimizationLevel: -1
[12/17/2024-10:18:04] [I] Input(s)s format: fp32:CHW
[12/17/2024-10:18:04] [I] Output(s)s format: fp32:CHW
[12/17/2024-10:18:04] [I] Input build shapes: model
[12/17/2024-10:18:04] [I] Input calibration shapes: model
[12/17/2024-10:18:04] [I] === System Options ===
[12/17/2024-10:18:04] [I] Device: 0
[12/17/2024-10:18:04] [I] DLACore:
[12/17/2024-10:18:04] [I] Plugins: /workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/plugins/aarch64/ess_plugins.so
[12/17/2024-10:18:04] [I] setPluginsToSerialize:
[12/17/2024-10:18:04] [I] dynamicPlugins:
[12/17/2024-10:18:04] [I] ignoreParsedPluginLibs: 0
[12/17/2024-10:18:04] [I]
[12/17/2024-10:18:04] [I] === Inference Options ===
[12/17/2024-10:18:04] [I] Batch: Explicit
[12/17/2024-10:18:04] [I] Input inference shapes: model
[12/17/2024-10:18:04] [I] Iterations: 10
[12/17/2024-10:18:04] [I] Duration: 3s (+ 200ms warm up)
[12/17/2024-10:18:04] [I] Sleep time: 0ms
[12/17/2024-10:18:04] [I] Idle time: 0ms
[12/17/2024-10:18:04] [I] Inference Streams: 1
[12/17/2024-10:18:04] [I] ExposeDMA: Disabled
[12/17/2024-10:18:04] [I] Data transfers: Enabled
[12/17/2024-10:18:04] [I] Spin-wait: Disabled
[12/17/2024-10:18:04] [I] Multithreading: Disabled
[12/17/2024-10:18:04] [I] CUDA Graph: Disabled
[12/17/2024-10:18:04] [I] Separate profiling: Disabled
[12/17/2024-10:18:04] [I] Time Deserialize: Disabled
[12/17/2024-10:18:04] [I] Time Refit: Disabled
[12/17/2024-10:18:04] [I] NVTX verbosity: 0
[12/17/2024-10:18:04] [I] Persistent Cache Ratio: 0
[12/17/2024-10:18:04] [I] Inputs:
[12/17/2024-10:18:04] [I] === Reporting Options ===
[12/17/2024-10:18:04] [I] Verbose: Disabled
[12/17/2024-10:18:04] [I] Averages: 10 inferences
[12/17/2024-10:18:04] [I] Percentiles: 90,95,99
[12/17/2024-10:18:04] [I] Dump refittable layers:Disabled
[12/17/2024-10:18:04] [I] Dump output: Disabled
[12/17/2024-10:18:04] [I] Profile: Disabled
[12/17/2024-10:18:04] [I] Export timing to JSON file:
[12/17/2024-10:18:04] [I] Export output to JSON file:
[12/17/2024-10:18:04] [I] Export profile to JSON file:
[12/17/2024-10:18:04] [I]
[12/17/2024-10:18:04] [I] === Device Information ===
[12/17/2024-10:18:04] [I] Selected Device: Orin
[12/17/2024-10:18:04] [I] Compute Capability: 8.7
[12/17/2024-10:18:04] [I] SMs: 8
[12/17/2024-10:18:04] [I] Device Global Memory: 15656 MiB
[12/17/2024-10:18:04] [I] Shared Memory per SM: 164 KiB
[12/17/2024-10:18:04] [I] Memory Bus Width: 256 bits (ECC disabled)
[12/17/2024-10:18:04] [I] Application Compute Clock Rate: 0.918 GHz
[12/17/2024-10:18:04] [I] Application Memory Clock Rate: 0.918 GHz
[12/17/2024-10:18:04] [I]
[12/17/2024-10:18:04] [I] Note: The application clock rates do not reflect the actual clock rates that the GPU is currently running at.
[12/17/2024-10:18:04] [I]
[12/17/2024-10:18:04] [I] TensorRT version: 8.6.2
[12/17/2024-10:18:04] [I] Loading standard plugins
[12/17/2024-10:18:04] [E] Uncaught exception detected: Unable to open library: libnvinfer_plugin.so.8 due to libnvdla_compiler.so: cannot open shared object file: No such file or directory
&&&& FAILED TensorRT.trtexec [TensorRT v8602] # /usr/src/tensorrt/bin/trtexec --onnx=/workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/ess.onnx --saveEngine=/workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/ess.engine --fp16 --staticPlugins=/workspaces/isaac_ros-dev/isaac_ros_assets/models/dnn_stereo_disparity/dnn_stereo_disparity_v4.1.0_onnx/plugins/aarch64/ess_plugins.so
[ros2run]: Process exited with failure 1