XXX failure of TensorRT X.Y when running XXX on GPU XXX

[Ocean-ZHY]

Description

The latest version of tensorrt creates a docker image. When running trtexec in docker to convert the onnx model into a trt model, a cuda error occurs. How to solve this?

    &&&& RUNNING TensorRT.trtexec [TensorRT v100001] # trtexec --onnx=./weights/StandWater_Seg_efsam_240508.onnx --saveEngine=./weights/StandWater_Seg_efsam_240508.trt --fp16
    [05/27/2024-07:46:47] [I] === Model Options ===
    [05/27/2024-07:46:47] [I] Format: ONNX
    [05/27/2024-07:46:47] [I] Model: ./weights/StandWater_Seg_efsam_240508.onnx
    [05/27/2024-07:46:47] [I] Output:
    [05/27/2024-07:46:47] [I] === Build Options ===
    [05/27/2024-07:46:47] [I] Memory Pools: workspace: default, dlaSRAM: default, dlaLocalDRAM: default, dlaGlobalDRAM: default, tacticSharedMem: default
    [05/27/2024-07:46:47] [I] avgTiming: 8
    [05/27/2024-07:46:47] [I] Precision: FP32+FP16
    [05/27/2024-07:46:47] [I] LayerPrecisions:
    [05/27/2024-07:46:47] [I] Layer Device Types:
    [05/27/2024-07:46:47] [I] Calibration:
    [05/27/2024-07:46:47] [I] Refit: Disabled
    [05/27/2024-07:46:47] [I] Strip weights: Disabled
    [05/27/2024-07:46:47] [I] Version Compatible: Disabled
    [05/27/2024-07:46:47] [I] ONNX Plugin InstanceNorm: Disabled
    [05/27/2024-07:46:47] [I] TensorRT runtime: full
    [05/27/2024-07:46:47] [I] Lean DLL Path:
    [05/27/2024-07:46:47] [I] Tempfile Controls: { in_memory: allow, temporary: allow }
    [05/27/2024-07:46:47] [I] Exclude Lean Runtime: Disabled
    [05/27/2024-07:46:47] [I] Sparsity: Disabled
    [05/27/2024-07:46:47] [I] Safe mode: Disabled
    [05/27/2024-07:46:47] [I] Build DLA standalone loadable: Disabled
    [05/27/2024-07:46:47] [I] Allow GPU fallback for DLA: Disabled
    [05/27/2024-07:46:47] [I] DirectIO mode: Disabled
    [05/27/2024-07:46:47] [I] Restricted mode: Disabled
    [05/27/2024-07:46:47] [I] Skip inference: Disabled
    [05/27/2024-07:46:47] [I] Save engine: ./weights/StandWater_Seg_efsam_240508.trt
    [05/27/2024-07:46:47] [I] Load engine:
    [05/27/2024-07:46:47] [I] Profiling verbosity: 0
    [05/27/2024-07:46:47] [I] Tactic sources: Using default tactic sources
    [05/27/2024-07:46:47] [I] timingCacheMode: local
    [05/27/2024-07:46:47] [I] timingCacheFile:
    [05/27/2024-07:46:47] [I] Enable Compilation Cache: Enabled
    [05/27/2024-07:46:47] [I] errorOnTimingCacheMiss: Disabled
    [05/27/2024-07:46:47] [I] Preview Features: Use default preview flags.
    [05/27/2024-07:46:47] [I] MaxAuxStreams: -1
    [05/27/2024-07:46:47] [I] BuilderOptimizationLevel: -1
    [05/27/2024-07:46:47] [I] Calibration Profile Index: 0
    [05/27/2024-07:46:47] [I] Weight Streaming: Disabled
    [05/27/2024-07:46:47] [I] Debug Tensors:
    [05/27/2024-07:46:47] [I] Input(s)s format: fp32:CHW
    [05/27/2024-07:46:47] [I] Output(s)s format: fp32:CHW
    [05/27/2024-07:46:47] [I] Input build shapes: model
    [05/27/2024-07:46:47] [I] Input calibration shapes: model
    [05/27/2024-07:46:47] [I] === System Options ===
    [05/27/2024-07:46:47] [I] Device: 0
    [05/27/2024-07:46:47] [I] DLACore:
    [05/27/2024-07:46:47] [I] Plugins:
    [05/27/2024-07:46:47] [I] setPluginsToSerialize:
    [05/27/2024-07:46:47] [I] dynamicPlugins:
    [05/27/2024-07:46:47] [I] ignoreParsedPluginLibs: 0
    [05/27/2024-07:46:47] [I]
    [05/27/2024-07:46:47] [I] === Inference Options ===
    [05/27/2024-07:46:47] [I] Batch: Explicit
    [05/27/2024-07:46:47] [I] Input inference shapes: model
    [05/27/2024-07:46:47] [I] Iterations: 10
    [05/27/2024-07:46:47] [I] Duration: 3s (+ 200ms warm up)
    [05/27/2024-07:46:47] [I] Sleep time: 0ms
    [05/27/2024-07:46:47] [I] Idle time: 0ms
    [05/27/2024-07:46:47] [I] Inference Streams: 1
    [05/27/2024-07:46:47] [I] ExposeDMA: Disabled
    [05/27/2024-07:46:47] [I] Data transfers: Enabled
    [05/27/2024-07:46:47] [I] Spin-wait: Disabled
    [05/27/2024-07:46:47] [I] Multithreading: Disabled
    [05/27/2024-07:46:47] [I] CUDA Graph: Disabled
    [05/27/2024-07:46:47] [I] Separate profiling: Disabled
    [05/27/2024-07:46:47] [I] Time Deserialize: Disabled
    [05/27/2024-07:46:47] [I] Time Refit: Disabled
    [05/27/2024-07:46:47] [I] NVTX verbosity: 0
    [05/27/2024-07:46:47] [I] Persistent Cache Ratio: 0
    [05/27/2024-07:46:47] [I] Optimization Profile Index: 0
    [05/27/2024-07:46:47] [I] Weight Streaming Budget: Disabled
    [05/27/2024-07:46:47] [I] Inputs:
    [05/27/2024-07:46:47] [I] Debug Tensor Save Destinations:
    [05/27/2024-07:46:47] [I] === Reporting Options ===
    [05/27/2024-07:46:47] [I] Verbose: Disabled
    [05/27/2024-07:46:47] [I] Averages: 10 inferences
    [05/27/2024-07:46:47] [I] Percentiles: 90,95,99
    [05/27/2024-07:46:47] [I] Dump refittable layers:Disabled
    [05/27/2024-07:46:47] [I] Dump output: Disabled
    [05/27/2024-07:46:47] [I] Profile: Disabled
    [05/27/2024-07:46:47] [I] Export timing to JSON file:
    [05/27/2024-07:46:47] [I] Export output to JSON file:
    [05/27/2024-07:46:47] [I] Export profile to JSON file:
    [05/27/2024-07:46:47] [I]
    [05/27/2024-07:46:47] [I] === Device Information ===
    Cuda failure: named symbol not found

Environment

TensorRT Version:10.0

NVIDIA GPU:4060Ti

NVIDIA Driver Version:12.5

CUDA Version:12.4

CUDNN Version:8.9.7