NVIDIA
trtexec failed with cuda driver 570
Description
Environment
TensorRT Version: 10.10
NVIDIA GPU: RTX3090
NVIDIA Driver Version: NVIDIA-SMI 570.133.07 Driver Version: 570.133.07 CUDA Version: 12.8
CUDA Version: 12.4
CUDNN Version:
Operating System: ubuntu
Python Version (if applicable):
Tensorflow Version (if applicable):
PyTorch Version (if applicable):
Baremetal or Container (if so, version):
Steps To Reproduce
Commands or scripts:
[11/07/2025-06:59:06] [I] === Device Information === [11/07/2025-06:59:06] [I] Available Devices: [11/07/2025-06:59:06] [I] Device 0: "NVIDIA GeForce RTX 3090" UUID: GPU-9196782c-2fae-91c7-c24f-e9a05e4369a7 [11/07/2025-06:59:06] [I] Device 1: "NVIDIA GeForce RTX 3090" UUID: GPU-fa724978-a973-65f4-7a78-26d847638556 [11/07/2025-06:59:06] [I] Selected Device: NVIDIA GeForce RTX 3090 [11/07/2025-06:59:06] [I] Selected Device ID: 0 [11/07/2025-06:59:06] [I] Selected Device UUID: GPU-9196782c-2fae-91c7-c24f-e9a05e4369a7 [11/07/2025-06:59:06] [I] Compute Capability: 8.6 [11/07/2025-06:59:06] [I] SMs: 82 [11/07/2025-06:59:06] [I] Device Global Memory: 24135 MiB [11/07/2025-06:59:06] [I] Shared Memory per SM: 100 KiB [11/07/2025-06:59:06] [I] Memory Bus Width: 384 bits (ECC disabled) [11/07/2025-06:59:06] [I] Application Compute Clock Rate: 1.695 GHz [11/07/2025-06:59:06] [I] Application Memory Clock Rate: 9.751 GHz [11/07/2025-06:59:06] [I] [11/07/2025-06:59:06] [I] Note: The application clock rates do not reflect the actual clock rates that the GPU is currently running at. [11/07/2025-06:59:06] [I] [11/07/2025-06:59:06] [I] TensorRT version: 10.10.0 [11/07/2025-06:59:06] [I] Loading standard plugins [11/07/2025-06:59:10] [W] [TRT] Unable to determine GPU memory usage: In getGpuMemStatsInBytes at /_src/common/extended/resources.cpp:1175 [11/07/2025-06:59:10] [E] [TRT] createInferBuilder: Error Code 6: API Usage Error (CUDA initialization failure with error: 35. Please check your CUDA installation: http://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html In operator() at /_src/optimizer/api/builder.cpp:1360) [11/07/2025-06:59:10] [E] [TRT] [checkMacros.cpp::catchCudaError::229] Error Code 1: Cuda Runtime (In catchCudaError at /_src/common/dispatch/checkMacros.cpp:229) [11/07/2025-06:59:10] [E] Builder creation failed [11/07/2025-06:59:10] [E] Failed to create engine from model or file. [11/07/2025-06:59:10] [E] Engine set up failed &&&& FAILED TensorRT.trtexec [TensorRT v101000] [b31] # ./build_10.10/trtexec --onnx=./samples/sampleNamedDimensions/concat_layer.onnx --saveEngine=test.trt
===============================
=== Help === --help, -h Print this message [11/07/2025-07:20:11] [E] Model missing or format not recognized &&&& FAILED TensorRT.trtexec [TensorRT v101000] [b31] # ./build_10.10/trtexec --version
==============================
/usr/src/tensorrt/bin/trtexec --version &&&& RUNNING TensorRT.trtexec [TensorRT v101401] [b48] # /usr/src/tensorrt/bin/trtexec --version Cuda failure at /_src/samples/common/common.h:1038: CUDA driver version is insufficient for CUDA runtime version
driver 570 support cuda library up to 12.8, and my cuda lib version is 12.4.