forked from pmodels/mpich
-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Mschaara/adio merge main #62
Merged
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Coll: Add CVAR to select composition for a few blocking collectives
We have been manually setting MPI error class index in src/mpi/errhan/baseerrnames.txt, and it has been out-of-sync with the values defined in mpi.h. This commit does following: * Remove the index from baseerrnames.txt and directly load the defined value from mpi.h instead. * Add UNKNOWN entry to map the missing entries in baseerrnames.txt. * When autogen.sh fails to extract error messages, do not generate dummy defmsg.h, fail instead. The dummy header will break the error class message translation anyway.
Add some indentation to the generated src/mpi/errhan/defmsg.h.
Re-initializing %KnownErrRoutines for every line of the source code is waste of time. Just load once can make the script run much faster -- from 3 seconds to a fraction of a second.
Use strict prevents misuse of global variables in Perl scripts.
maint: fix and cleanup maint/extracterrmsgs Approved-by: Ken Raffenetti
Sometime user may want to disable f08 even when their compiler passes the PAC_FC_2008_SUPPORT test. Add explicit --disable-f08 option for that.
The test suite is now configured independently and no longer need these variables.
Now we only use a single $FC compiler, there isn't a need to check compatibility between F77 and FC.
By default we will build all Fortran interfaces. However, sometime user may desire disable certain interface. Use separate options so user can easily do so. Currently the f90 interface depends on f77 interface. But it should be possible to only build f77, or only build f08, or build f77 + f08. Remove and replace the variable enable_fc from configure since it is ambiguous whether it refers to the FC compiler or the interface features.
config/fortran: clean up Fortran binding config options Approved-by: Ken Raffenetti
Adding --enable-g to the configure line was ignored. Make it is a synonym for --enable-g=all.
Only consider the nodes used in a launch as part of the total core count. See pmodels#5835.
hydra: Fix spawn when specifying launch hosts Approved-by: Hui Zhou <[email protected]>
…llgatherv mpi/coll: add TSP-based scatter/ring-based-allgather algorithm for Ibcast
submodule: Update UCX to v1.12.0 Approved-by: Hui Zhou <[email protected]>
Left over from previous commits.
Remove the AC_MSG_ERROR on needing the Fortran mismatch flag and add the flag to the mpif77 wrapper. This will not affect users of modern Fortran MPI applications who are likely to use `mpifort` or `mpif90` and uses the mpi module interface. Users who use `mpif77` is likely to use `mpif.h` interface and will require the ignore-mismatch compiler flags to compile their MPI applications. By adding the extra flags to mpif77 wrapper only, we hope to strike a balance of convenience and impact.
fortran: remove fortran mismatched arg check Approved-by: Ken Raffenetti
When testsuite autogen is run outside mpich, supply an empty mtest_mpix.h so the testing can proceed. The header files is for mpich specific extension and will fail with non-mpich implementation anyway. The empty stub will allow strict testing to proceed. The regular mpich testing is not affected since mpich's top level autogen will always supply this file.
test: supply empty mtest_mpix.h Approved-by: Ken Raffenetti
Routine to select only gentran-based algorithm for iallreduce.
Routine to select only gentran-based algorithm for ibcast.
Routine to select only gentran-based algorithm for ibarrier.
release.pl: add creation of libpmi tarball Approved-by: Ken Raffenetti
Add a signaure to the first packet when proxy first connect to server. This guards against random connection and port scans.
hydra: enhance proxy initial handshake Approved-by: Ken Raffenetti
It should allocate a buffer size of true_lb + true_extent rather than just true_extent.
romio: fix MPIU_external32_buffer_setup Approved-by: Ken Raffenetti
Author: Taru Doodi <[email protected]>
coll: allreduce ring
Add the same --with-thread-package options as in MPL for consistent thread support. This fixes compile error when MPL is configured with --with-thread-package=none
pmi: add configure option to disable threads Approved-by: Ken Raffenetti
Signed-off-by: Mohamad Chaarawi <[email protected]>
Signed-off-by: Mohamad Chaarawi <[email protected]>
Signed-off-by: Mohamad Chaarawi <[email protected]>
Co-authored-by: Surabhi Jain <[email protected]>
Multi-leaders based composition: It has `num_leaders` per node, which reduce the data within sub-node_comm. It is followed by intra_node reduce and inter_node allreduce on the piece of data the leader is responsible for. A shared memory buffer is allocated per leader. If size of message exceeds this shm buffer, the message is chunked. Constraints: For a comm, all nodes should have same number of ranks per node, op should be commutative. Co-authored-by: Surabhi Jain <[email protected]>
Co-authored-by: Surabhi Jain <[email protected]>
coll: Add multileader allreduce composition
Signed-off-by: Mohamad Chaarawi <[email protected]>
Signed-off-by: Mohamad Chaarawi <[email protected]>
The header file simple_pmiutil.h was in src/pmi/simple and it is now being refactored. It is still needed by gforker and remshell. This commit restore the file into src/pm/util where it belongs. This fixes the build for gforker and remshell.
pm/util: restore simple_pmiutil.h Approved-by: Ken Raffenetti
test on an MPI request should not block when using the daos driver as the event API has a mode to make progress without waiting. Fix the issue in the poll callback to use the NOWAIT option on event test. Signed-off-by: Mohamad Chaarawi <[email protected]>
adio/daos: make daos_event_wait non-blocking Approved-by: Rob Latham
This reverts commit 819f697.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Pull Request Description
Author Checklist
Particularly focus on why, not what. Reference background, issues, test failures, xfail entries, etc.
Commits are self-contained and do not do two things at once.
Commit message is of the form:
module: short description
Commit message explains what's in the commit.
Whitespace checker. Warnings test. Additional tests via comments.
For non-Argonne authors, check contribution agreement.
If necessary, request an explicit comment from your companies PR approval manager.