madhavakrishna / clu Goto Github PK

This is the OpenCL Utility (CLU) library, a set of functions and C
macros to make the host side of OpenCL programming less tedious.

The library provides wrappers for common complex or repetitive tasks
such as platform and device selection, information gathering, command
queue management and memory management. This is achieved while keeping
perfect compatibility with the underlying OpenCL library, allowing code
and data mixing.

Some wrapper function will appear to be overly slower than the
lower-level OpenCL functions. This is generally due to CLU making an
effort to anticipate future requests and retrieving/caching additional
information. For example, all the platform/device information will be
available in appropriate structures when getting a platform/device using
the CLU functions, instead of requiring distinct (and often repetitive)
calls to the appropriate clGet*Info() OpenCL functions.

Some of the CLU functions are shamelessly inspired by the NVIDIA CUDA
APIs, although an effort has been made to keep the interface more
consistent with the one found in OpenCL, especially in terms of return
values and error management.