Merge common nbnxn CUDA/OpenCL GPU wait code-paths

[alexxy/gromacs.git] / src / gromacs / mdlib / nbnxn_gpu.h

diff --git a/src/gromacs/mdlib/nbnxn_gpu.h b/src/gromacs/mdlib/nbnxn_gpu.h
index 1b903f13366e7c14a4025e078c981c3df2c9ea0b..cdb68597af3a8e5f1c80b48fdf2fead4f8279162 100644 (file)
--- a/src/gromacs/mdlib/nbnxn_gpu.h
+++ b/src/gromacs/mdlib/nbnxn_gpu.h
@@ -122,8 +122,19 @@ void nbnxn_gpu_launch_cpyback(gmx_nbnxn_gpu_t  gmx_unused              *nb,
                               int                    gmx_unused         aloc) GPU_FUNC_TERM
 
 /*! \brief
- * Wait for the asynchronously launched nonbonded calculations and data
+ * Wait for the asynchronously launched nonbonded tasks and data
  * transfers to finish.
+ *
+ * Also does timing accounting and reduction of the internal staging buffers.
+ * As this is called at the end of the step, it also resets the pair list and
+ * pruning flags.
+ *
+ * \param[in] nb The nonbonded data GPU structure
+ * \param[in] flags Force flags
+ * \param[in] aloc Atom locality identifier
+ * \param[out] e_lj Pointer to the LJ energy output to accumulate into
+ * \param[out] e_el Pointer to the electrostatics energy output to accumulate into
+ * \param[out] fshift Pointer to the shift force buffer to accumulate into
  */
 GPU_FUNC_QUALIFIER
 void nbnxn_gpu_wait_for_gpu(gmx_nbnxn_gpu_t gmx_unused *nb,