• Malcolm Roberts's avatar
    Single-node multi-gpu functional implementation · d1c91135
    Malcolm Roberts authored
    Add a new experimental API for single-process multi-gpu transforms.  This is a functional implementation which is not designed for performance improvements.  Currently, only 2D/3D complex-to-complex transforms are supported using the interleaved format, and batch size must be 1.
    
    Authored by:
    Steve Leung
    Malcolm Roberts
    Alan Ayala
    d1c91135
To find the state of this project's repository at the time of any of these versions, check out the tags.