1
1
openmpi/opal/datatype
Ralph Castain 552c9ca5a0 George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT:    Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL

All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies.  This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP.  Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose.  UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs.  A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.

This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
..
Makefile.am The final step of the RFC: convert the <foo>libdir and friends to fit their respective code areas, and equate them all at the top. Note that we can't entirely separate things as the opal_install_dirs framework can't handle separated locations for the various trees. 2014-05-08 02:01:35 +00:00
opal_convertor_internal.h George pointed out that r29991 wasn't quite right. This patch is 2013-12-19 22:54:27 +00:00
opal_convertor_raw.c George pointed out that r29991 wasn't quite right. This patch is 2013-12-19 22:54:27 +00:00
opal_convertor.c Correctly deal with partially converted datatypes both 2014-05-29 21:53:44 +00:00
opal_convertor.h Improve CUDA GPU transfers over openib BTL. Use aynchronous copies. 2013-01-17 22:34:43 +00:00
opal_copy_functions_heterogeneous.c Get rid of all unused functions. 2009-07-16 19:53:07 +00:00
opal_copy_functions.c - Split the datatype engine into two parts: an MPI specific part in 2009-07-13 04:56:31 +00:00
opal_datatype_add.c Datatype with count of zero do not generate traffic in the type map, nor 2014-05-05 21:49:18 +00:00
opal_datatype_checksum.h - Split the datatype engine into two parts: an MPI specific part in 2009-07-13 04:56:31 +00:00
opal_datatype_clone.c Correct the pointer math when we copy the opal_datatype_t object. In addition 2009-09-18 20:05:22 +00:00
opal_datatype_copy.c George pointed out that r29991 wasn't quite right. This patch is 2013-12-19 22:54:27 +00:00
opal_datatype_copy.h Promote from int32_t to size_t. 2012-03-05 14:08:47 +00:00
opal_datatype_create_contiguous.c Reshape the datatype engine. The basic types are built down in OPAL. MPI types are 2011-01-13 06:08:54 +00:00
opal_datatype_create.c Init all fields (valgrind cleanup). 2013-12-21 23:24:29 +00:00
opal_datatype_cuda.c George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-) 2014-07-26 00:47:28 +00:00
opal_datatype_cuda.h Change one argument to use const. Matches better how function is called. 2014-05-13 18:55:41 +00:00
opal_datatype_destroy.c - Split the datatype engine into two parts: an MPI specific part in 2009-07-13 04:56:31 +00:00
opal_datatype_dump.c George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-) 2014-07-26 00:47:28 +00:00
opal_datatype_fake_stack.c - Split the datatype engine into two parts: an MPI specific part in 2009-07-13 04:56:31 +00:00
opal_datatype_get_count.c MPI-3: add support for large counts using derived datatypes 2013-07-23 15:35:14 +00:00
opal_datatype_internal.h George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-) 2014-07-26 00:47:28 +00:00
opal_datatype_memcpy.h Add a second version of the datatype copy function using memmove instead of memcpy. 2010-11-12 23:22:35 +00:00
opal_datatype_module.c Correctly initialize and finalize all the datatype classes. No memory leaks on the 2013-12-20 15:57:10 +00:00
opal_datatype_optimize.c Remove warnings identified by clang 3.4 2014-03-12 13:17:54 +00:00
opal_datatype_pack.c Correctly save the displacement for the case where the convertor is not 2014-04-14 04:46:58 +00:00
opal_datatype_pack.h Improve CUDA GPU transfers over openib BTL. Use aynchronous copies. 2013-01-17 22:34:43 +00:00
opal_datatype_position.c Correctly deal with partially converted datatypes both 2014-05-29 21:53:44 +00:00
opal_datatype_prototypes.h Missing C_DECLS. 2009-07-13 17:31:42 +00:00
opal_datatype_resize.c - Split the datatype engine into two parts: an MPI specific part in 2009-07-13 04:56:31 +00:00
opal_datatype_unpack.c Correctly deal with partially converted datatypes both 2014-05-29 21:53:44 +00:00
opal_datatype_unpack.h Improve CUDA GPU transfers over openib BTL. Use aynchronous copies. 2013-01-17 22:34:43 +00:00
opal_datatype.h rearrange the fields to remove a gap in the datatype. 2013-12-20 15:57:56 +00:00