Tim: would this meet your needs, or would you prefer totally seperate functions for each level of comparison? I wanted to make it easy to use, but also allow flexibility in case others want to compare different things.
This commit was SVN r1541.
Put a permanent fix into the include files to avoid the circular reference. You only need to include the "base.h" file now and it will pickup the ns.h file for you. Anyone working on components for the name server, however, should be warned that including the ns.h directly by itself (instead of going through base.h) will yield several unresolved errors.
Ignore the replica component - it is bogus at this time.
This commit was SVN r1537.
b) Initialize the Queue DMA based descriptors
c) Reorganize the initialization code.
Todo:
a) Frag and request descriptors and get put/get functions going
b) Add rdma support for elan4, better to have global memory
This commit was SVN r1535.
--disable-mca-dso to build MCA components statically as part of
libmpi, or --enable-mca-dso=LIST to only build some components as DSOs
(see ./configure --help for more details).
This commit was SVN r1521.
- accepts --with-ptl-gm=DIR to specify where the GM headers and
libraries are
- accepts --with-ptl-gm-libdir=DIR if libgm is in a different place
than gm.h
- defines OMPI_MCA_PTL_GM_API_VERSION with the correct GM symbolic
constant suitable for use with gm_init()
- defines OMPI_MCA_PTL_GM_HAVE_RDMA_PUT indicating whether to use
gm_put() or gm_directed_send_with_callback() (a change in the GM API
starting with gm 2.0)
- defines OMPI_MCA_PTL_GM_HAVE_RDMA_GET indicating whether gm_get()
exists or not
This commit was SVN r1519.
inline function to simplify things, since it needs to be invoked in
3 different back-end coll API functions
- Implement MPI_EXSCAN in the basic coll module
This commit was SVN r1499.
a substantial revamp of the session directory system, and the ompi_system_info structure, in accordance
with the startup/shutdown design released last week. If you are using either the session directories
or the ompi_system_info structure, I would recommend avoiding them for the next couple of days as these
files are undergoing considerable change. The documentation will be updated as well during this process.
This commit was SVN r1495.
real commit of the collectives. MPI_SCAN and MPI_EXSCAN are still not
implemented, but lots of other things are in the critical path and
holding up other people, so it's ok to commit without them:
- better checks for sizes in configure, and add defaults for fortran
sizes if we don't have a fortran compiler
- fix some logic that was accidentally broken for size checks for the
file type offset_t
- add some C equivalent types for fortran's complex and double complex
(for use in internal reduction/op functions)
- additionals and slight reorganization of ompi_mpi_init()
ompi_mpi_finalize()
- fully implement all top-level MPI collective calls, including all
param checking for both intra- and inter-communicators (woof)
- change the communicator_t type for stuff that we need in coll, and
update all references throughout the code base to match
- all kinds of updates to the coll framework base
- next cut of the basic coll module -- has all intracommunicator
collectives implemented except scan and exscan (see note above).
All intercommunicator functions return ERR_NOT_IMPLEMENTED.
- MPI_Op is a fixed implementation -- not component-ized yet. So
there are generic C loops for all implementations.
This commit was SVN r1491.
