Requirements for DriveOS LLM SDK x86 PC?

Autonomous Vehicles DRIVE AGX Thor DRIVE AGX Thor General
1

DRIVE OS Version: 7.0.3

Issue Description: I am trying to use the Nvidia DriveOS LLM SDK as recommended on my Razer Blade 18 with RTX 5090 Laptop GPU. Unfortunately the build of DriveOS LLM SDK fails on this machine. I get the following logs:

cd build
cmake .. -DTRT_PACKAGE_DIR=/usr/aarch64-linux-gnu -DCMAKE_TOOLCHAIN_FILE=cmake/aarch64_cross_toolchain.cmake -DAUTO_TARGET=thor
make
– The CXX compiler identification is GNU 13.3.0
– The CUDA compiler identification is NVIDIA 13.2.51
– Detecting CXX compiler ABI info
– Detecting CXX compiler ABI info - done
– Check for working CXX compiler: /usr/bin/aarch64-linux-gnu-g++ - skipped
– Detecting CXX compile features
– Detecting CXX compile features - donebut
– Configurable variable CUDA_VERSION set to 12.8
– Configurable variable CUDA_DIR set to /usr/local/cuda/targets/aarch64-linux
CMake Error at CMakeLists.txt:75 (message):
NvOnnxParser.h not found in TensorRT headers, please specify the
-DONNX_PARSER_INCLUDE_DIR when invoking CMake

-- Configuring incomplete, errors occurred!
make: *** No targets specified and no makefile found. Stop.

I tried to install different versions of TensorRT already, but none of them worked yet. Either it seems to be too old for my GPU or too new for the DriveOS LLM SDK. Do you have an idea, why it fails?

Building of TensorRT Edge LLM works fine on my laptop btw, but I can’t use it on my Drive AGX Thor currently due to DriveOS 7.0.3.

4

Dear @adas.developer ,
Are you trying to build DRIVE OS LLM SDK on 7.0.3 docker?

5

No I just unpacked the tar.gz and tried to build it locally on my x86 laptop. I need it to export the llm to onnx format. This doesn’t work. On my Drive AGX Thor with DriveOS 7.0.3 building works.

6

It does not work on local host. It needs DOS 7.0.3 environment