FastROCS TK 1.8.0¶
New features¶
FastROCS TK has been switched from an OpenCL backend to a CUDA backend. This should have no impact on FastROCS usage but may reduce the complexity of the installation process. The installation of CUDA development tools is not required to use FastROCS TK.
NVIDIA driver support has been added for driver versions
367.*
. Support has been dropped for340.*
. A list of currently supported NVIDIA drivers for FastROCS TK can be found in the Installation section.
Minor bug fixes¶
The method
OEShapeDatabase.Open
has been modified to give a warning rather than an error when a file contains some molecules with heavy atoms and some molecules without heavy atoms. This is to prevent an entire database upload from failing if only a handful of molecules are corrupt.ShapeSimilarityMatrix.py
has been renamedShapeDistanceMatrix.py
and the resultant matrix has been fixed to accurately reflect distance on the diagonal (i.e., 0.0).
Python-specific changes¶
The Python integration tests now check for the presence of a GPU and skip the FastROCS test if a GPU is not available. Previously, the integration tests would fail.