Software and Hardware Requirements
PyTorch
At this moment, only Linux distributions with glibc >= v2.31 and python >= v3.9 are supported. An Nvidia GPU of compute capability at least 6.0 and drivers at least 535 are required for the hardware accelerated functionality of the Library.
TensorFlow
The TensorFlow version is not available yet.
Free Library
PyTorch
pip3 install geondptfree==2.9.0.1.0.post1
v2.7.0 - v2.8.0:
pip3 install geondptfree==2.7.0.1.0.post1
v2.5.1 - v2.6.0:
pip3 install geondptfree==2.5.1.1.0.post1
Known Issues
PyTorch
- undefined symbol (not LIBFFI related)
- Likely cause: mismatch between the PyTorch version and the GeoND Library version.
- Solution: refer above to install the correct GeoND Library version based on your PyTorch version. - undefined symbol (LIBFFI related)
- Likely cause: conda virtual environment has a different libffi version than the system.
- Solution: resolve the discrepancy OR (temporary fix) find the file libffi.so.7 (or libffi.so.8), it is typically in either /lib/x86_64-linux-gnu/ or /usr/lib/x86_64-linux-gnu/ and runexport LD_PRELOAD=/lib/x86_64-linux-gnu/libffi.so.7(or the path to libffi.so on your system) before running your program. - Not listed: Submit a Bug Report.
