diff --git a/configure.ac b/configure.ac index d5d6045bf56b3585a98369997031d30f9e849742..2420d9accf40e42defb9e3995c587e6d042bc955 100644 --- a/configure.ac +++ b/configure.ac @@ -40,7 +40,7 @@ AX_CHECK_ENABLE_DEBUG AC_USE_SYSTEM_EXTENSIONS AC_PROG_CC AM_PROG_CC_C_O -AC_OPENMP +AX_OPENMP # If debug is selected then we also define SWIFT_DEVELOP_MODE to control # any developer code options. diff --git a/m4/ax_openmp.m4 b/m4/ax_openmp.m4 new file mode 100644 index 0000000000000000000000000000000000000000..5d7469380f16fc6ca4b8d5ebf330aa81e12c0701 --- /dev/null +++ b/m4/ax_openmp.m4 @@ -0,0 +1,123 @@ +# =========================================================================== +# https://www.gnu.org/software/autoconf-archive/ax_openmp.html +# =========================================================================== +# +# SYNOPSIS +# +# AX_OPENMP([ACTION-IF-FOUND[, ACTION-IF-NOT-FOUND]]) +# +# DESCRIPTION +# +# This macro tries to find out how to compile programs that use OpenMP a +# standard API and set of compiler directives for parallel programming +# (see http://www-unix.mcs/) +# +# On success, it sets the OPENMP_CFLAGS/OPENMP_CXXFLAGS/OPENMP_F77FLAGS +# output variable to the flag (e.g. -omp) used both to compile *and* link +# OpenMP programs in the current language. +# +# NOTE: You are assumed to not only compile your program with these flags, +# but also link it with them as well. +# +# If you want to compile everything with OpenMP, you should set: +# +# CFLAGS="$CFLAGS $OPENMP_CFLAGS" +# #OR# CXXFLAGS="$CXXFLAGS $OPENMP_CXXFLAGS" +# #OR# FFLAGS="$FFLAGS $OPENMP_FFLAGS" +# +# (depending on the selected language). +# +# The user can override the default choice by setting the corresponding +# environment variable (e.g. OPENMP_CFLAGS). +# +# ACTION-IF-FOUND is a list of shell commands to run if an OpenMP flag is +# found, and ACTION-IF-NOT-FOUND is a list of commands to run it if it is +# not found. If ACTION-IF-FOUND is not specified, the default action will +# define HAVE_OPENMP. +# +# LICENSE +# +# Copyright (c) 2008 Steven G. Johnson <stevenj@alum.mit.edu> +# Copyright (c) 2015 John W. Peterson <jwpeterson@gmail.com> +# Copyright (c) 2016 Nick R. Papior <nickpapior@gmail.com> +# +# This program is free software: you can redistribute it and/or modify it +# under the terms of the GNU General Public License as published by the +# Free Software Foundation, either version 3 of the License, or (at your +# option) any later version. +# +# This program is distributed in the hope that it will be useful, but +# WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General +# Public License for more details. +# +# You should have received a copy of the GNU General Public License along +# with this program. If not, see <https://www.gnu.org/licenses/>. +# +# As a special exception, the respective Autoconf Macro's copyright owner +# gives unlimited permission to copy, distribute and modify the configure +# scripts that are the output of Autoconf when processing the Macro. You +# need not follow the terms of the GNU General Public License when using +# or distributing such scripts, even though portions of the text of the +# Macro appear in them. The GNU General Public License (GPL) does govern +# all other use of the material that constitutes the Autoconf Macro. +# +# This special exception to the GPL applies to versions of the Autoconf +# Macro released by the Autoconf Archive. When you make and distribute a +# modified version of the Autoconf Macro, you may extend this special +# exception to the GPL to apply to your modified version as well. + +#serial 13 + +AC_DEFUN([AX_OPENMP], [ +AC_PREREQ([2.69]) dnl for _AC_LANG_PREFIX + +AC_CACHE_CHECK([for OpenMP flag of _AC_LANG compiler], ax_cv_[]_AC_LANG_ABBREV[]_openmp, [save[]_AC_LANG_PREFIX[]FLAGS=$[]_AC_LANG_PREFIX[]FLAGS +ax_cv_[]_AC_LANG_ABBREV[]_openmp=unknown +# Flags to try: -fopenmp (gcc), -mp (SGI & PGI), +# -qopenmp (icc>=15), -openmp (icc), +# -xopenmp (Sun), -omp (Tru64), +# -qsmp=omp (AIX), +# none +ax_openmp_flags="-qopenmp -fopenmp -openmp -mp -xopenmp -omp -qsmp=omp none" +if test "x$OPENMP_[]_AC_LANG_PREFIX[]FLAGS" != x; then + ax_openmp_flags="$OPENMP_[]_AC_LANG_PREFIX[]FLAGS $ax_openmp_flags" +fi +for ax_openmp_flag in $ax_openmp_flags; do + case $ax_openmp_flag in + none) []_AC_LANG_PREFIX[]FLAGS=$save[]_AC_LANG_PREFIX[] ;; + *) []_AC_LANG_PREFIX[]FLAGS="$save[]_AC_LANG_PREFIX[]FLAGS $ax_openmp_flag" ;; + esac + AC_LINK_IFELSE([AC_LANG_SOURCE([[ +@%:@include <omp.h> + +static void +parallel_fill(int * data, int n) +{ + int i; +@%:@pragma omp parallel for + for (i = 0; i < n; ++i) + data[i] = i; +} + +int +main() +{ + int arr[100000]; + omp_set_num_threads(2); + parallel_fill(arr, 100000); + return 0; +} +]])],[ax_cv_[]_AC_LANG_ABBREV[]_openmp=$ax_openmp_flag; break],[]) +done +[]_AC_LANG_PREFIX[]FLAGS=$save[]_AC_LANG_PREFIX[]FLAGS +]) +if test "x$ax_cv_[]_AC_LANG_ABBREV[]_openmp" = "xunknown"; then + m4_default([$2],:) +else + if test "x$ax_cv_[]_AC_LANG_ABBREV[]_openmp" != "xnone"; then + OPENMP_[]_AC_LANG_PREFIX[]FLAGS=$ax_cv_[]_AC_LANG_ABBREV[]_openmp + fi + m4_default([$1], [AC_DEFINE(HAVE_OPENMP,1,[Define if OpenMP is enabled])]) +fi +])dnl AX_OPENMP diff --git a/src/common_io.c b/src/common_io.c index eea1d91fe2375190e15453e04d7075c976afafea..145eba9015802b8a5ecc871c16ddfd02bff37559 100644 --- a/src/common_io.c +++ b/src/common_io.c @@ -287,9 +287,9 @@ void io_read_array_attribute(hid_t grp, const char* name, /* Check if correct number of element */ if (count != number_element) { error( - "Error found a different number of elements than expected (%lli != " - "%lli) in attribute %s", - count, number_element, name); + "Error found a different number of elements than expected (%llu != " + "%llu) in attribute %s", + (unsigned long long)count, (unsigned long long)number_element, name); } /* Read attribute */ @@ -358,9 +358,9 @@ void io_read_array_dataset(hid_t grp, const char* name, enum IO_DATA_TYPE type, /* Check if correct number of element */ if (count != number_element) { error( - "Error found a different number of elements than expected (%lli != " - "%lli) in dataset %s", - count, number_element, name); + "Error found a different number of elements than expected (%llu != " + "%llu) in dataset %s", + (unsigned long long)count, (unsigned long long)number_element, name); } /* Read dataset */ diff --git a/src/distributed_io.c b/src/distributed_io.c index d56cb1cd7b78335b298692500cd432c043e3f6c9..d2466030eae7635eddb2d4ec5830070e3557cd2c 100644 --- a/src/distributed_io.c +++ b/src/distributed_io.c @@ -169,7 +169,8 @@ void write_distributed_array( h_err = H5Pset_chunk(h_prop, rank, chunk_shape); if (h_err < 0) error("Error while setting chunk size (%llu, %llu) for field '%s'.", - chunk_shape[0], chunk_shape[1], props.name); + (unsigned long long)chunk_shape[0], + (unsigned long long)chunk_shape[1], props.name); /* Are we imposing some form of lossy compression filter? */ if (lossy_compression != compression_write_lossless) diff --git a/src/fof_catalogue_io.c b/src/fof_catalogue_io.c index e437492209b65d405fa0331b5c08c9d4cde9b76b..393792ded0ee2c50c22f455a8de8088246ef73ac 100644 --- a/src/fof_catalogue_io.c +++ b/src/fof_catalogue_io.c @@ -208,7 +208,8 @@ void write_fof_hdf5_array( h_err = H5Pset_chunk(h_prop, rank, chunk_shape); if (h_err < 0) error("Error while setting chunk size (%llu, %llu) for field '%s'.", - chunk_shape[0], chunk_shape[1], props.name); + (unsigned long long)chunk_shape[0], + (unsigned long long)chunk_shape[1], props.name); /* Are we imposing some form of lossy compression filter? */ if (lossy_compression != compression_write_lossless) diff --git a/src/line_of_sight.c b/src/line_of_sight.c index e90eea141b23281ac382e5d776a89f48260e7a22..d0ea412558242f14f2ac560fb658ab69962438cc 100644 --- a/src/line_of_sight.c +++ b/src/line_of_sight.c @@ -316,7 +316,8 @@ void write_los_hdf5_dataset(const struct io_props props, const size_t N, h_err = H5Pset_chunk(h_prop, rank, chunk_shape); if (h_err < 0) error("Error while setting chunk size (%llu, %llu) for field '%s'.", - chunk_shape[0], chunk_shape[1], props.name); + (unsigned long long)chunk_shape[0], + (unsigned long long)chunk_shape[1], props.name); /* Impose check-sum to verify data corruption */ h_err = H5Pset_fletcher32(h_prop); diff --git a/src/single_io.c b/src/single_io.c index 0ae8ef42ad5fa4bce27ca6fd51906a00de81305f..3a8e17a938249c04345826db23062c2296222d28 100644 --- a/src/single_io.c +++ b/src/single_io.c @@ -302,7 +302,8 @@ void write_array_single(const struct engine* e, hid_t grp, const char* fileName, h_err = H5Pset_chunk(h_prop, rank, chunk_shape); if (h_err < 0) error("Error while setting chunk size (%llu, %llu) for field '%s'.", - chunk_shape[0], chunk_shape[1], props.name); + (unsigned long long)chunk_shape[0], + (unsigned long long)chunk_shape[1], props.name); } /* Are we imposing some form of lossy compression filter? */ diff --git a/src/space_extras.c b/src/space_extras.c index c06560671cbfdd772c076c6ba2b760cce532b88c..9d8a88952e5a3d2a670ba63f96001dc27eda26c7 100644 --- a/src/space_extras.c +++ b/src/space_extras.c @@ -184,7 +184,9 @@ void space_allocate_extras(struct space *s, int verbose) { size_t local_cell_id = 0; int current_cell = local_cells[local_cell_id]; int count_in_cell = 0; +#ifdef SWIFT_DEBUG_CHECKS size_t count_extra_gparts = 0; +#endif for (size_t i = 0; i < nr_actual_gparts + expected_num_extra_gparts; ++i) { #ifdef SWIFT_DEBUG_CHECKS @@ -199,7 +201,9 @@ void space_allocate_extras(struct space *s, int verbose) { s->gparts[i].x[1] = cells[current_cell].loc[1] + half_cell_width[1]; s->gparts[i].x[2] = cells[current_cell].loc[2] + half_cell_width[2]; ++count_in_cell; +#ifdef SWIFT_DEBUG_CHECKS count_extra_gparts++; +#endif } /* Once we have reached the number of extra gpart per cell, we move to the @@ -276,7 +280,9 @@ void space_allocate_extras(struct space *s, int verbose) { size_t local_cell_id = 0; int current_cell = local_cells[local_cell_id]; int count_in_cell = 0; +#ifdef SWIFT_DEBUG_CHECKS size_t count_extra_parts = 0; +#endif for (size_t i = 0; i < nr_actual_parts + expected_num_extra_parts; ++i) { #ifdef SWIFT_DEBUG_CHECKS @@ -291,7 +297,9 @@ void space_allocate_extras(struct space *s, int verbose) { s->parts[i].x[1] = cells[current_cell].loc[1] + half_cell_width[1]; s->parts[i].x[2] = cells[current_cell].loc[2] + half_cell_width[2]; ++count_in_cell; +#ifdef SWIFT_DEBUG_CHECKS count_extra_parts++; +#endif } /* Once we have reached the number of extra part per cell, we move to the @@ -357,7 +365,9 @@ void space_allocate_extras(struct space *s, int verbose) { size_t local_cell_id = 0; int current_cell = local_cells[local_cell_id]; int count_in_cell = 0; +#ifdef SWIFT_DEBUG_CHECKS size_t count_extra_sinks = 0; +#endif for (size_t i = 0; i < nr_actual_sinks + expected_num_extra_sinks; ++i) { #ifdef SWIFT_DEBUG_CHECKS @@ -372,7 +382,9 @@ void space_allocate_extras(struct space *s, int verbose) { s->sinks[i].x[1] = cells[current_cell].loc[1] + half_cell_width[1]; s->sinks[i].x[2] = cells[current_cell].loc[2] + half_cell_width[2]; ++count_in_cell; +#ifdef SWIFT_DEBUG_CHECKS count_extra_sinks++; +#endif } /* Once we have reached the number of extra sink per cell, we move to the @@ -439,7 +451,9 @@ void space_allocate_extras(struct space *s, int verbose) { size_t local_cell_id = 0; int current_cell = local_cells[local_cell_id]; int count_in_cell = 0; +#ifdef SWIFT_DEBUG_CHECKS size_t count_extra_sparts = 0; +#endif for (size_t i = 0; i < nr_actual_sparts + expected_num_extra_sparts; ++i) { #ifdef SWIFT_DEBUG_CHECKS @@ -454,7 +468,9 @@ void space_allocate_extras(struct space *s, int verbose) { s->sparts[i].x[1] = cells[current_cell].loc[1] + half_cell_width[1]; s->sparts[i].x[2] = cells[current_cell].loc[2] + half_cell_width[2]; ++count_in_cell; +#ifdef SWIFT_DEBUG_CHECKS count_extra_sparts++; +#endif } /* Once we have reached the number of extra spart per cell, we move to the @@ -521,7 +537,9 @@ void space_allocate_extras(struct space *s, int verbose) { size_t local_cell_id = 0; int current_cell = local_cells[local_cell_id]; int count_in_cell = 0; +#ifdef SWIFT_DEBUG_CHECKS size_t count_extra_bparts = 0; +#endif for (size_t i = 0; i < nr_actual_bparts + expected_num_extra_bparts; ++i) { #ifdef SWIFT_DEBUG_CHECKS @@ -536,7 +554,9 @@ void space_allocate_extras(struct space *s, int verbose) { s->bparts[i].x[1] = cells[current_cell].loc[1] + half_cell_width[1]; s->bparts[i].x[2] = cells[current_cell].loc[2] + half_cell_width[2]; ++count_in_cell; +#ifdef SWIFT_DEBUG_CHECKS count_extra_bparts++; +#endif } /* Once we have reached the number of extra bpart per cell, we move to the diff --git a/src/task.c b/src/task.c index c290bbb11cc4f4f0483b6eb44b578964577c8eb7..333f648769db702651f6438d2a8a3cf9c34de5b2 100644 --- a/src/task.c +++ b/src/task.c @@ -1670,7 +1670,6 @@ void task_dump_active(struct engine *e) { fprintf(file_thread, "%i 0 none none -1 0 %lld %lld %lld %lld %lld 0 %lld\n", engine_rank, (long long int)e->tic_step, (long long int)e->toc_step, e->updates, e->g_updates, e->s_updates, cpufreq); - int count = 0; for (int l = 0; l < e->sched.nr_tasks; l++) { struct task *t = &e->sched.tasks[l]; @@ -1691,7 +1690,6 @@ void task_dump_active(struct engine *e) { (t->ci != NULL) ? t->ci->grav.count : 0, (t->cj != NULL) ? t->cj->grav.count : 0, t->flags); } - count++; } fclose(file_thread); diff --git a/src/threadpool.c b/src/threadpool.c index 632ac917a349af34e24f7b720f7ed0d65727de3a..5fddd59ae833bf0a8de8ac3024cfa7baad987090 100644 --- a/src/threadpool.c +++ b/src/threadpool.c @@ -424,7 +424,7 @@ void threadpool_clean(struct threadpool *tp) { /** * @brief return the threadpool id of the current thread. */ -int threadpool_gettid() { +int threadpool_gettid(void) { int *tid = (int *)pthread_getspecific(threadpool_tid); return *tid; } diff --git a/src/tools.c b/src/tools.c index d771016591371d64b6adc1bc85169021ef262fef..022c3fb3b64aa432332e3e137e2ae7a220b91d3b 100644 --- a/src/tools.c +++ b/src/tools.c @@ -1030,7 +1030,7 @@ int compare_particles(struct part *a, struct part *b, double threshold) { * * @result memory use in Kb. */ -long get_maxrss() { +long get_maxrss(void) { struct rusage usage; getrusage(RUSAGE_SELF, &usage); return usage.ru_maxrss;