AMPI/CUDA: Synchronous and asynchronous invocation, some cleanup
authorPhil Miller <mille121@illinois.edu>
Mon, 17 May 2010 22:04:07 +0000 (17:04 -0500)
committerPhil Miller <mille121@illinois.edu>
Mon, 24 May 2010 20:57:37 +0000 (15:57 -0500)
commitf9adac4857522f6acdd9051867dd2b909b98c626
treebfbd52846d3bcb50a01f06fd36916b502c29b5fb
parent75b2cb515ae351e1e24df445b1316331aacd625b
AMPI/CUDA: Synchronous and asynchronous invocation, some cleanup

- Add a synchronous variant of the work request function
- Remove the MPI_Comm argument to the functions, and just use MPI_COMM_SELF internally
- Clean up comments and other nits
examples/ampi/cuda/simple/simple.c
src/libs/ck-libs/ampi/ampi.C
src/libs/ck-libs/ampi/ampi.h
src/libs/ck-libs/ampi/ampiimpl.h