I am currently planning to compile, deploy, and test OKVIS2-X on a cloud server powered by an ARM processor running the openEuler operating system.
Since many C++ VSLAM frameworks are primarily optimized for x86 architectures (e.g., relying on SSE/AVX instructions for vectorization) and standard Ubuntu environments, I would like to ask a few questions before I dive into the compilation process:
ARM Support: Does OKVIS2-X inherently support compilation and execution on ARM architectures (aarch64)? Are there any known hard dependencies on x86-specific instruction sets?
Compilation Flags: Are there any specific CMake flags, dependency versions (e.g., Eigen, Ceres), or code modifications you recommend to successfully build the project on ARM?
Community Experience: Has anyone in the team or community successfully tested it on openEuler (or similar Enterprise Linux distributions like CentOS/RHEL for ARM)?
For reference, here is my target hardware and software environment:
OS: openEuler 24.03 (Huawei Cloud EulerOS)
Architecture: aarch64 (ARMv8)
CPU: Kunpeng 920
NPU: Ascend 310P (Planning to explore if NPU acceleration is possible later)
Any guidance, tips, or pointers to relevant branches/commits would be greatly appreciated. I will also go ahead and try compiling it myself, and I can report back here with any specific compiler errors I encounter.
Thank you for your time and help!
I am currently planning to compile, deploy, and test OKVIS2-X on a cloud server powered by an ARM processor running the openEuler operating system.
Since many C++ VSLAM frameworks are primarily optimized for x86 architectures (e.g., relying on SSE/AVX instructions for vectorization) and standard Ubuntu environments, I would like to ask a few questions before I dive into the compilation process:
ARM Support: Does OKVIS2-X inherently support compilation and execution on ARM architectures (aarch64)? Are there any known hard dependencies on x86-specific instruction sets?
Compilation Flags: Are there any specific CMake flags, dependency versions (e.g., Eigen, Ceres), or code modifications you recommend to successfully build the project on ARM?
Community Experience: Has anyone in the team or community successfully tested it on openEuler (or similar Enterprise Linux distributions like CentOS/RHEL for ARM)?
For reference, here is my target hardware and software environment:
OS: openEuler 24.03 (Huawei Cloud EulerOS)
Architecture: aarch64 (ARMv8)
CPU: Kunpeng 920
NPU: Ascend 310P (Planning to explore if NPU acceleration is possible later)
Any guidance, tips, or pointers to relevant branches/commits would be greatly appreciated. I will also go ahead and try compiling it myself, and I can report back here with any specific compiler errors I encounter.
Thank you for your time and help!