Skip to content

Enable iterative initialization for target='opencl'

Just by looking on it: I guess here 'self._gpu' should be used to (maybe) also work with 'opencl'.

Merge request reports