diff options
author | jhendersonHDF <jhenderson@hdfgroup.org> | 2023-12-07 18:13:21 (GMT) |
---|---|---|
committer | GitHub <noreply@github.com> | 2023-12-07 18:13:21 (GMT) |
commit | fb84c696a476c39db2c35e7d86b818138e089358 (patch) | |
tree | 56c0fec578e2e0d2eaee455cf3ae133dca541989 /testpar/t_mpi.c | |
parent | dc8b455ac00e899b53127de0d270230f4d4e40a1 (diff) | |
download | hdf5-fb84c696a476c39db2c35e7d86b818138e089358.zip hdf5-fb84c696a476c39db2c35e7d86b818138e089358.tar.gz hdf5-fb84c696a476c39db2c35e7d86b818138e089358.tar.bz2 |
Sync 1.14 branch with develop (#3887)
* Preserve MPI-I/O file hints when fapl is closed (#3755)
* Fix for issue #3025: Save the MPI info in the file struct so H5Fget_access_plist() can retrieve it from there.
* Add compression tests for subfiling (#3769)
* Fix typo in comment (#3775)
* Fixed a file handle leak in the core VFD (#3779)
When opening a file with the core VFD and a file image, if the file
already exists, the file check would leak the POSIX file handle.
Fixes GitHub issue #635
* Fix a format string warning in the C++ examples (#3776)
* Cancel running GitHub workflows on push to same PR (#3772)
* Cancel running GitHub workflows on push to same PR
* Remove github.sha from workflow concurrency groups
* Print some messages in parallel tests on MPI rank 0 only (#3785)
Avoids overly verbose output from all processes emitting progress, etc. info.
* Avoid attempted use of NULL pointer in parallel compression code (#3786)
The parallel compression test code tests for the case where all MPI ranks have no selection in a dataset when writing to it. Add an early exit to the code to avoid attempting to use a NULL pointer due to there being no work to do.
* Don't install h5tools_test_utils test program on system (#3793)
* Add Doxygen to H5FDsplitter.h (#3794)
* H5FD_CURR_SPLITTER_VFD_CONFIG_VERSION
* H5FD_SPLITTER_PATH_MAX
* H5FD_SPLITTER_MAGIC
* H5FD_splitter_vfd_config_t
* H5Pset_fapl_splitter()
* H5Pget_fapl_splitter()
* Update Doxygen initializers & identifiers in VFDs (#3795)
* Add Doxygen for all H5FD_<VFD> initializers
* Add Doxygen for all H5FD_<VFD>_VALUE values
* Mark H5FD_<vfd>_init() calls private in Doxygen
* Fix memory corruption in 'MPI I/O FAPL preserve' test (#3806)
* Fix usage of h5_clean_files in t_pflush2.c (#3807)
* Fix parallel driver check in h5_fixname_real (#3808)
* Fix a couple usages of MPI_Info_get (#3809)
* Remove H5system.c warning on Windows oneAPI. (#3812)
* Add processing of NVHPC flags in linux-gnulibc1 file (#3804)
* Disable testing as tests are failing the same as in CMake
* Use the current toolchain for examples as default (#3810)
* Fix misc. warnings from GCC when compiling with -fsanitize=undefined (#3787)
* Set NVHPC maximum optimization level to -O1 for now (#3800)
* Set NVHPC maximum optimization level to -O1 for now
Compiling HDF5 with NVHPC 23.5 - 23.9 results in test failures in
4 different test files that need to be resolved. Since those tests
pass with an optimization level of -O1 (and -O0) and it is currently
unclear whether the test failures are due to issues in HDF5 or issues
in the 'nvc' compiler, set the maximum optimization level for NVHPC
to -O1 until the test failures are resolved.
* Disable nvhpc Java testing in CMake and amend known issues
* Re-enable testing of Autotools nvhpc
* Update some doxygen links to local refs (#3814)
* Rework MPI Info FAPL preserve PR to use VFD 'ctl' operations (#3782)
* Removed the use of C wrappers from H5P APIs. (#3824)
* fix seg fault on frontier/cray
* fix seg fault on frontier/cray
* fix seg fault on frontier/cray
* removed the use of h5pclose_c
* removed the use of h5pclose_c
* Fortran Wrappers H5VLnative_addr_to_token_f and H5VLnative_token_to_address_f (#3801)
* Added H5VLnative_addr_to_token_f and H5VLnative_token_to_address_f
* Added H5VLnative_addr_to_token_f and H5VLnative_token_to_address_f tests
* Create test for H5Pget_dxpl_mpio (#3825)
* Create test and add to testphdf5
* Renamed h5fuse.sh to h5fuse (#3834)
* provide an alternative to mapfile for older bash
* Disable FP exceptions in H5T init code (#3837)
The H5T floating-point datatype initialization code can raise exceptions when handling signaling NaNs. This change disables FE_INVALID exceptions during initialization.
Also removes the -ieee=full change for NAG Fortran as that shouldn't be necessary anymore.
Fixes #3831
* Add intel oneapi windows build to CI CMake (#3836)
* Remove printf format warning on Windows oneAPI. (#3838)
* Correct ENV variables (#3841)
* Remove Autotools sed hack (#3848)
configure.ac contains a sed line that cleans up incorrect library
flags which was added to paper over some bugs in earlier versions
of the Autotools. These issues are not a problem with the current
versions of the Autootols.
The sed line causes problems on MacOS, so it has been removed.
Fixes #3843
* Make filter unregister callbacks safe for VOL connectors (#3629)
* Make filter callbacks use top-level API functions
When using VOL connectors, H5I_iterate may not provide
valid object pointers to its callback. This change keeps
existing functionality in H5Zunregister() without using
potentially unsafe pointers.
* Filter callbacks use internal API
* Skip MPI work on non-native VOL
* Add extra space in comments for consistency (#3852)
* Add extra space in comments for consistency
* uncomment tfloatsattrs test
* Update Actions badges to link to relevant workflow (#3850)
* Add CMake long double cross-compile defaults (#3683)
HDF5 performs a couple of checks at build time to see if long double
values can be converted correctly (IBM's Power architecture uses a
special format for long doubles). These checks were performed using
TRY_RUN, which is a problem when cross-compiling.
These checks now use default values appropriate for most non-Power
systems when cross-compiling. The cache values can be pre-set if
necessary, which will preempt both the TRY_RUN and the default.
Affected values:
H5_LDOUBLE_TO_LONG_SPECIAL (default no)
H5_LONG_TO_LDOUBLE_SPECIAL (default no)
H5_LDOUBLE_TO_LLONG_ACCURATE (default yes)
H5_LLONG_TO_LDOUBLE_CORRECT (default yes)
H5_DISABLE_SOME_LDOUBLE_CONV (default no)
Fixes GitHub #3585
* Updates for building and testing VOL connectors
* Fix issue with HDF5_VOL_ALLOW_EXTERNAL CMake variable
* Initialize parallel testing with MPI_THREAD_MULTIPLE when testing API
* Add CMake variable to allow specifying a VOL connector's package name
* Remove call to MPI_Init in serial API tests
While previously necessary, it now interferes with VOL connectors that
may need to be initialized with MPI_THREAD_MULTIPLE
* Fixes for CI and presets (#3853)
* Change dest for doxygen (#3856)
* Implement selection vector I/O with collective chunk filling (#3826)
* Changes for ECP-344: Implement selection vector I/O with collective chunk filling.
Also fix a bug in H5FD__mpio_write_vector() to account for fixed size optimization
when computing max address.
* Fixes based on PR review comments:
For H5Dchunk.c: fix H5MM_xfree()
For H5FDmpio.c:
1) Revert the fix to H5FD__mpio_write_vector()
2) Apply the patch from Neil on the proper length of s_sizes reported by H5FD__mpio_vector_build_types()
* Put back the logic of dividing up the work among all the mpi ranks similar to the
original H5D__chunk_collective_fill() routine.
* Add a test to verify the fix for the illegal reference problem in H5FD__mpio_write_vector().
* Do not publish compression headers or docs (#3865)
* Fix typo: look -> loop (#3866)
* Moved the README to markdown and expanded its overview of the files, file generation, and other Fortran wrapper development practices as mentioned in the HDF5 architectural document. I added a new figure and included the SVG file and the original xfig file it was generated from. (#3862)
* Add HDF5_DISABLE_TESTS_REGEX option to skip tests (#3859)
* Fix typo in error message for `MPI_Type_dup`. (#3867)
* Complete the `if command line option` sentence. (#3868)
* Fix h5dump segmentation fault when --vfd-value option is used (#3873)
* Updated URL in funding.yml (#3882)
Using new shortened URL, might look better.
* Remove unused variable from unmerged changes
* Update src/H5Tinit_float.c
Diffstat (limited to 'testpar/t_mpi.c')
-rw-r--r-- | testpar/t_mpi.c | 46 |
1 files changed, 21 insertions, 25 deletions
diff --git a/testpar/t_mpi.c b/testpar/t_mpi.c index eff39d0..0f1e27b 100644 --- a/testpar/t_mpi.c +++ b/testpar/t_mpi.c @@ -53,14 +53,14 @@ test_mpio_overlap_writes(char *filename) MPI_Offset mpi_off; MPI_Status mpi_stat; - if (VERBOSE_MED) - printf("MPIO independent overlapping writes test on file %s\n", filename); - nerrs = 0; /* set up MPI parameters */ MPI_Comm_size(MPI_COMM_WORLD, &mpi_size); MPI_Comm_rank(MPI_COMM_WORLD, &mpi_rank); + if (VERBOSE_MED && MAINPROCESS) + printf("MPIO independent overlapping writes test on file %s\n", filename); + /* Need at least 2 processes */ if (mpi_size < 2) { if (MAINPROCESS) @@ -211,7 +211,7 @@ test_mpio_gb_file(char *filename) MPI_Comm_size(MPI_COMM_WORLD, &mpi_size); MPI_Comm_rank(MPI_COMM_WORLD, &mpi_rank); - if (VERBOSE_MED) + if (VERBOSE_MED && MAINPROCESS) printf("MPI_Offset range test\n"); /* figure out the signness and sizeof MPI_Offset */ @@ -274,12 +274,13 @@ test_mpio_gb_file(char *filename) /* * Verify if we can write to a file of multiple GB sizes. */ - if (VERBOSE_MED) + if (VERBOSE_MED && MAINPROCESS) printf("MPIO GB file test %s\n", filename); if (sizeof_mpi_offset <= 4) { - printf("Skipped GB file range test " - "because MPI_Offset cannot support it\n"); + if (MAINPROCESS) + printf("Skipped GB file range test " + "because MPI_Offset cannot support it\n"); } else { buf = (char *)malloc(MB); @@ -294,7 +295,8 @@ test_mpio_gb_file(char *filename) mrc = MPI_File_open(MPI_COMM_WORLD, filename, MPI_MODE_CREATE | MPI_MODE_RDWR, info, &fh); VRFY((mrc == MPI_SUCCESS), "MPI_FILE_OPEN"); - printf("MPIO GB file write test %s\n", filename); + if (MAINPROCESS) + printf("MPIO GB file write test %s\n", filename); /* instead of writing every bytes of the file, we will just write * some data around the 2 and 4 GB boundaries. That should cover @@ -333,7 +335,8 @@ test_mpio_gb_file(char *filename) */ /* open it again to verify the data written */ /* but only if there was no write errors */ - printf("MPIO GB file read test %s\n", filename); + if (MAINPROCESS) + printf("MPIO GB file read test %s\n", filename); if (errors_sum(writerrs) > 0) { printf("proc %d: Skip read test due to previous write errors\n", mpi_rank); goto finish; @@ -377,7 +380,8 @@ test_mpio_gb_file(char *filename) mrc = MPI_Barrier(MPI_COMM_WORLD); VRFY((mrc == MPI_SUCCESS), "Sync before leaving test"); - printf("Test if MPI_File_get_size works correctly with %s\n", filename); + if (MAINPROCESS) + printf("Test if MPI_File_get_size works correctly with %s\n", filename); mrc = MPI_File_open(MPI_COMM_WORLD, filename, MPI_MODE_RDONLY, info, &fh); VRFY((mrc == MPI_SUCCESS), ""); @@ -432,7 +436,6 @@ finish: static int test_mpio_1wMr(char *filename, int special_request) { - char hostname[128]; int mpi_size, mpi_rank; MPI_File fh; char mpi_err_str[MPI_MAX_ERROR_STRING]; @@ -456,19 +459,8 @@ test_mpio_1wMr(char *filename, int special_request) } /* show the hostname so that we can tell where the processes are running */ - if (VERBOSE_DEF) { -#ifdef H5_HAVE_GETHOSTNAME - if (gethostname(hostname, sizeof(hostname)) < 0) { - printf("gethostname failed\n"); - hostname[0] = '\0'; - } -#else - printf("gethostname unavailable\n"); - hostname[0] = '\0'; -#endif - PRINTID; - printf("hostname=%s\n", hostname); - } + if (VERBOSE_DEF) + h5_show_hostname(); /* Delete any old file in order to start anew. */ /* Must delete because MPI_File_open does not have a Truncate mode. */ @@ -1005,6 +997,10 @@ test_mpio_special_collective(char *filename) static int parse_options(int argc, char **argv) { + int mpi_rank; + + MPI_Comm_rank(MPI_COMM_WORLD, &mpi_rank); + while (--argc) { if (**(++argv) != '-') { break; @@ -1053,7 +1049,7 @@ parse_options(int argc, char **argv) return (1); } H5Pclose(plist); - if (VERBOSE_MED) { + if (VERBOSE_MED && MAINPROCESS) { printf("Test filenames are:\n"); for (i = 0; i < n; i++) printf(" %s\n", filenames[i]); |