ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
272ef9f412
openmpi/ompi/datatype
История
George Bosilca 272ef9f412 Get rid of the storage in the convertor. It wasn't working as expected in all 
the cases. Instead replace it with a better solution, which work even for
fragments received not in order. However, this solution work only on the
current supported modes in ompi (homogeneous & heterogeneous with endianess).

The method is tricky. We will rely on 2 partial unpacks. First we will find
a byte that is not on the data to unpack, and we will pad the data with this
byte. Once we have the full length as expected, we will unpack the data, and
all the bytes in the unpacked form which do not match the unused byte will be
copied into the user buffer. This way we will reconstruct the unpacked data
in 2 times, once for the begining and once for the end.

This commit was SVN r10270.
2006-06-08 23:35:07 +00:00
..
.compile_files Rename src -> ompi 2005-07-02 13:43:57 +00:00
convertor_internal.h The master convertor is the one that knows if the peers are 2006-05-22 06:22:32 +00:00
convertor.c Get rid of the storage in the convertor. It wasn't working as expected in all 2006-06-08 23:35:07 +00:00
convertor.h Get rid of the storage in the convertor. It wasn't working as expected in all 2006-06-08 23:35:07 +00:00
copy_functions_heterogeneous.c * datatype_check might need to uptdate the value of count (if we received 2006-05-19 21:58:12 +00:00
copy_functions.c Add more knowledge into the master convertor. Now all the conversion 2006-05-05 04:30:11 +00:00
datatype_checksum.h - modified convertor copy_and_prepare routines to accept an addition 2006-03-17 18:46:48 +00:00
datatype_internal.h MPI standard state that MPI_LONG_LONG and MPI_LONG_LONG_INT are synonyms. Thanks to 2006-04-24 21:24:10 +00:00
datatype_memcpy.h Last step for the generation of 2 set of pack/unpack functions. One with 2006-03-17 08:04:59 +00:00
datatype_pack.c This one belong to the previous commit but it somehow get lost ... 2006-05-05 04:32:42 +00:00
datatype_pack.h First split of the datatype. More files and a cleaner distribution of functions 2006-03-16 21:04:34 +00:00
datatype_prototypes.h Keep only the best [the fastest] version of the unpack function. 2006-04-28 16:18:29 +00:00
datatype_unpack.c Get rid of the storage in the convertor. It wasn't working as expected in all 2006-06-08 23:35:07 +00:00
datatype_unpack.h First split of the datatype. More files and a cleaner distribution of functions 2006-03-16 21:04:34 +00:00
datatype.h Add another flag to mark the data that are really contiguous. Really here means that 2006-06-08 21:27:50 +00:00
dt_add.c Add another flag to mark the data that are really contiguous. Really here means that 2006-06-08 21:27:50 +00:00
dt_arch.c * Fix bug in determining size of C++ bool when creating the predefined 2006-03-25 02:53:41 +00:00
dt_arch.h * Fix bug in determining size of C++ bool when creating the predefined 2006-03-25 02:53:41 +00:00
dt_args.c - Get rid of compiler warning: init type to DT_MAX_PREDEFINED 2006-03-04 13:16:04 +00:00
dt_copy.c Move the code around to shorten the execution path for the most common data-types. 2006-04-28 16:17:28 +00:00
dt_create_array.c Make sure all datatype files have the correct updated copyright information. 2006-02-12 21:03:01 +00:00
dt_create_dup.c Allow zero count contiguous data-types. And be user friendly, set the ub, lb, true_lb 2006-06-05 21:57:28 +00:00
dt_create_indexed.c Have a consistent behavior. Independing on the MPI type that will get created if the 2006-06-06 18:22:36 +00:00
dt_create_struct.c Small optimization. 2006-06-06 18:23:06 +00:00
dt_create_vector.c Minor cleanups and add the same consistent behavior as the one described on the 2006-06-06 18:24:09 +00:00
dt_create.c Fix the datatype bug noticed by Rainer. Under some circumstances (and only for 2006-05-27 06:21:27 +00:00
dt_destroy.c Make sure all datatype files have the correct updated copyright information. 2006-02-12 21:03:01 +00:00
dt_external32.c MPI standard state that MPI_LONG_LONG and MPI_LONG_LONG_INT are synonyms. Thanks to 2006-04-24 21:24:10 +00:00
dt_get_count.c * Rename two files that had conflicting basenames with files in mpi/c/. This 2006-04-22 20:03:45 +00:00
dt_match_size.c Correctly mark the Fortran data-types as being Fortran (not C and change it later 2006-06-05 20:44:17 +00:00
dt_module.c Add another flag to mark the data that are really contiguous. Really here means that 2006-06-08 21:27:50 +00:00
dt_optimize.c Make sure all datatype files have the correct updated copyright information. 2006-02-12 21:03:01 +00:00
dt_sndrcv.c Make sure all datatype files have the correct updated copyright information. 2006-02-12 21:03:01 +00:00
fake_stack.c Make sure all datatype files have the correct updated copyright information. 2006-02-12 21:03:01 +00:00
Makefile.am * Rename two files that had conflicting basenames with files in mpi/c/. This 2006-04-22 20:03:45 +00:00
position.c Get rid of the storage in the convertor. It wasn't working as expected in all 2006-06-08 23:35:07 +00:00