Merge branch 'develop' of github.com:lammps/lammps into alphataubio-kokkos-fixes

cmake/CMakeLists.txt

@@ -141,7 +141,7 @@ endif()
 
 # silence nvcc warnings
 if((PKG_KOKKOS) AND (Kokkos_ENABLE_CUDA) AND NOT (CMAKE_CXX_COMPILER_ID STREQUAL "Clang"))
-  set(CMAKE_TUNE_DEFAULT "${CMAKE_TUNE_DEFAULT} -Xcudafe --diag_suppress=unrecognized_pragma")
+  set(CMAKE_TUNE_DEFAULT "${CMAKE_TUNE_DEFAULT} -Xcudafe --diag_suppress=unrecognized_pragma -Xcudafe --diag_suppress=128")
 endif()
 
 # we require C++11 without extensions. Kokkos requires at least C++17 (currently)

cmake/Modules/Packages/H5MD.cmake

@@ -3,7 +3,7 @@ enable_language(C)
 # we don't use the parallel i/o interface.
 set(HDF5_PREFER_PARALLEL FALSE)
 
-find_package(HDF5 REQUIRED)
+find_package(HDF5 COMPONENTS C REQUIRED)
 
 # parallel HDF5 will import incompatible MPI headers with a serial build
 if((NOT BUILD_MPI) AND HDF5_IS_PARALLEL)

doc/src/compute_composition_atom.rst

@@ -88,6 +88,10 @@ too frequently.
 
 ----------
 
+.. include:: accel_styles.rst
+
+----------
+
 Output info
 """""""""""
 

doc/src/compute_pressure.rst

@@ -125,10 +125,6 @@ where thermo_temp is the ID of a similarly defined compute of style
 
 ----------
 
-.. include:: accel_styles.rst
-
-----------
-
 Output info
 """""""""""
 

doc/src/compute_temp_deform.rst

@@ -128,6 +128,12 @@ See the :doc:`Howto thermostat <Howto_thermostat>` page for a
 discussion of different ways to compute temperature and perform
 thermostatting.
 
+----------
+
+.. include:: accel_styles.rst
+
+----------
+
 Output info
 """""""""""
 

doc/src/compute_temp_partial.rst

@@ -82,12 +82,6 @@ See the :doc:`Howto thermostat <Howto_thermostat>` page for a
 discussion of different ways to compute temperature and perform
 thermostatting.
 
-----------
-
-.. include:: accel_styles.rst
-
-----------
-
 Output info
 """""""""""
 

doc/src/fix_addforce.rst

@@ -115,10 +115,6 @@ correctly, the minimization will not converge properly.
 
 ----------
 
-.. include:: accel_styles.rst
-
-----------
-
 Restart, fix_modify, output, run start/stop, minimize info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/fix_aveforce.rst

@@ -71,10 +71,6 @@ to it.
 
 ----------
 
-.. include:: accel_styles.rst
-
-----------
-
 Restart, fix_modify, output, run start/stop, minimize info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/fix_nph_body.rst

@@ -79,8 +79,6 @@ It also means that changing attributes of *thermo_temp* or
 
 ----------
 
-.. include:: accel_styles.rst
-
 Restart, fix_modify, output, run start/stop, minimize info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/fix_npt_body.rst

@@ -103,8 +103,6 @@ remaining thermal degrees of freedom, and the bias is added back in.
 
 ----------
 
-.. include:: accel_styles.rst
-
 Restart, fix_modify, output, run start/stop, minimize info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/fix_nvt_body.rst

@@ -85,8 +85,6 @@ remaining thermal degrees of freedom, and the bias is added back in.
 
 ----------
 
-.. include:: accel_styles.rst
-
 Restart, fix_modify, output, run start/stop, minimize info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/fix_wall_flow.rst

@@ -123,6 +123,12 @@ also be potentially mitigated by using more multiple walls.
   conservative as possible (every timestep if needed).  Those are the
   default settings.
 
+----------
+
+.. include:: accel_styles.rst
+
+----------
+
 Restart, fix_modify, output, run start/stop, minimize info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/min_style.rst

@@ -3,6 +3,8 @@
 min_style cg command
 ====================
 
+Accelerator Variant: *cg/kk*
+
 min_style hftn command
 ======================
 

doc/src/pair_coul_diel.rst

@@ -71,6 +71,10 @@ The global cutoff (:math:`r_c`) specified in the pair_style command is used.
 
 ----------
 
+.. include:: accel_styles.rst
+
+----------
+
 Mixing, shift, table, tail correction, restart, rRESPA info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/pair_coul_slater.rst

@@ -77,6 +77,10 @@ The global decay length of the charge (:math:`\lambda`) specified in the pair_st
 
 ----------
 
+.. include:: accel_styles.rst
+
+----------
+
 Mixing, shift, table, tail correction, restart, rRESPA info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/pair_granular.rst

@@ -741,10 +741,6 @@ atom types.
 
 ----------
 
-.. include:: accel_styles.rst
-
-----------
-
 Mixing, shift, table, tail correction, restart, rRESPA info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/pair_mesodpd.rst

@@ -287,6 +287,10 @@ concentration profiles of the two chemical species as
 
 ----------
 
+.. include:: accel_styles.rst
+
+----------
+
 Mixing, shift, table, tail correction, restart, rRESPA info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/pair_mie.rst

@@ -62,6 +62,10 @@ cutoff specified in the pair_style command is used.
 
 ----------
 
+.. include:: accel_styles.rst
+
+----------
+
 Mixing, shift, table, tail correction, restart, rRESPA info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/pair_sph_heatconduction.rst

@@ -39,6 +39,10 @@ above.
 
 ----------
 
+.. include:: accel_styles.rst
+
+----------
+
 Mixing, shift, table, tail correction, restart, rRESPA info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/pair_sph_lj.rst

@@ -43,6 +43,10 @@ above.
 
 ----------
 
+.. include:: accel_styles.rst
+
+----------
+
 Mixing, shift, table, tail correction, restart, rRESPA info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/src/pair_sph_taitwater.rst

@@ -52,6 +52,10 @@ above.
 
 ----------
 
+.. include:: accel_styles.rst
+
+----------
+
 Mixing, shift, table, tail correction, restart, rRESPA info
 """""""""""""""""""""""""""""""""""""""""""""""""""""""""""
 

doc/utils/check-styles.py

@@ -61,16 +61,50 @@ region = {}
 total = 0
 
 index_pattern = re.compile(r"^.. index:: (compute|fix|pair_style|angle_style|bond_style|dihedral_style|improper_style|kspace_style|dump)\s+([a-zA-Z0-9/_]+)$")
+accel_pattern = re.compile(r"^.. include::\s+accel_styles.rst$")
 style_pattern = re.compile(r"(.+)Style\((.+),(.+)\)")
+cmd_pattern = re.compile(r"^.. index:: ([a-zA-Z0-9/_]+)$")
 upper = re.compile("[A-Z]+")
-gpu = re.compile("(.+)/gpu$")
-intel = re.compile("(.+)/intel$")
-kokkos = re.compile("(.+)/kk$")
-kokkos_skip = re.compile("(.+)/kk/(host|device)$")
-omp = re.compile("(.+)/omp$")
-opt = re.compile("(.+)/opt$")
+gpu = re.compile("(.+)/gpu\\s*$")
+intel = re.compile("(.+)/intel\\s*$")
+kokkos = re.compile("(.+)/kk$\\s*")
+kokkos_skip = re.compile("(.+)/kk/(host|device)\\s*$")
+omp = re.compile("(.+)/omp\\s*$")
+opt = re.compile("(.+)/opt\\s*$")
 removed = re.compile("(.*)Deprecated$")
 
+accel_pattern = re.compile(r"^.. include::\s+accel_styles.rst$")
+
+def require_accel_include(path):
+    found = False
+    needs = False
+    # handle exceptions
+    if path == "src/min_style.rst" : needs = True
+    if path == "src/atom_style.rst" : needs = True
+    # check file
+    with open(path, 'r') as reader:
+        for line in reader:
+            m = accel_pattern.match(line)
+            if m: found = True
+            m = index_pattern.match(line)
+            if m:
+                if gpu.match(line): needs = True
+                if omp.match(line): needs = True
+                if kokkos.match(line): needs = True
+                if intel.match(line): needs = True
+                if opt.match(line): needs = True
+            m = cmd_pattern.match(line)
+            if m:
+                if gpu.match(line): needs = True
+                if omp.match(line): needs = True
+                if kokkos.match(line): needs = True
+                if intel.match(line): needs = True
+                if opt.match(line): needs = True
+    if needs and not found:
+        print("Missing '.. include:: accel_style.rst' in file ", path)
+    if not needs and found:
+        print("Have '.. include:: accel_style.rst' without accelerated styles in file ", path)
+
 def load_index_entries_in_file(path):
     entries = []
     with open(path, 'r') as reader:
@@ -82,6 +116,11 @@ def load_index_entries_in_file(path):
                 entries.append((command_type, style))
     return entries
 
+def check_accel_includes():
+    rst_files = glob(os.path.join(doc_dir, '*.rst'))
+    for f in rst_files:
+        require_accel_include(f)
+
 def load_index_entries():
     index = {'compute': set(), 'fix': set(), 'pair_style': set(), 'angle_style': set(),
              'bond_style': set(), 'dihedral_style': set(), 'improper_style': set(),
@@ -249,6 +288,8 @@ Total number of styles (including suffixes): %d""" \
        len(fix), len(improper), len(integrate), len(kspace), \
        len(minimize), len(pair), len(reader), len(region), total))
 
+check_accel_includes()
+
 index = load_index_entries()
 
 total_index = 0

lib/h5md/Makefile.h5cc

@@ -4,7 +4,8 @@ CC=h5cc
 
 # -DH5_NO_DEPRECATED_SYMBOLS is required here to ensure we are using
 # the v1.8 API when HDF5 is configured to default to using the v1.6 API.
-CFLAGS=-D_DEFAULT_SOURCE -O2 -DH5_NO_DEPRECATED_SYMBOLS  -Wall -fPIC
+#CFLAGS=-D_DEFAULT_SOURCE -O2 -DH5_NO_DEPRECATED_SYMBOLS  -Wall -fPIC
+CFLAGS=-D_DEFAULT_SOURCE -O2  -Wall -fPIC
 HDF5_PATH=/usr
 INC=-I include
 AR=ar

src/MAKE/MACHINES/Makefile.perlmutter_kokkos

@@ -9,7 +9,7 @@ SHELL = /bin/sh
 KOKKOS_ABSOLUTE_PATH = $(shell cd $(KOKKOS_PATH); pwd)
 
 CC =		$(KOKKOS_ABSOLUTE_PATH)/bin/nvcc_wrapper
-CCFLAGS =	-g -O3 -DNDEBUG -Xcudafe --diag_suppress=unrecognized_pragma
+CCFLAGS =	-g -O3 -DNDEBUG -Xcudafe --diag_suppress=unrecognized_pragma -Xcudafe --diag_suppress=128
 SHFLAGS =	-fPIC
 DEPFLAGS =	-M
 

src/MAKE/MACHINES/Makefile.summit_kokkos

@@ -9,7 +9,7 @@ SHELL = /bin/sh
 KOKKOS_ABSOLUTE_PATH = $(shell cd $(KOKKOS_PATH); pwd)
 
 CC =		$(KOKKOS_ABSOLUTE_PATH)/bin/nvcc_wrapper
-CCFLAGS =	-g -O3 -DNDEBUG -Xcudafe --diag_suppress=unrecognized_pragma
+CCFLAGS =	-g -O3 -DNDEBUG -Xcudafe --diag_suppress=unrecognized_pragma -Xcudafe --diag_suppress=128
 SHFLAGS =	-fPIC
 DEPFLAGS =	-M
 

src/MAKE/OPTIONS/Makefile.kokkos_cuda_mpi

@@ -10,7 +10,7 @@ KOKKOS_ABSOLUTE_PATH = $(shell cd $(KOKKOS_PATH); pwd)
 export MPICH_CXX = $(KOKKOS_ABSOLUTE_PATH)/bin/nvcc_wrapper
 export OMPI_CXX = $(KOKKOS_ABSOLUTE_PATH)/bin/nvcc_wrapper
 CC =		mpicxx
-CCFLAGS =	-g -O3 -DNDEBUG -Xcudafe --diag_suppress=unrecognized_pragma
+CCFLAGS =	-g -O3 -DNDEBUG -Xcudafe --diag_suppress=unrecognized_pragma -Xcudafe --diag_suppress=128
 SHFLAGS =	-fPIC
 # uncomment when compiling with Intel 21.5 or older
 FMTFLAGS = 	# -std=c++11