Merge branch 'NEON' into 'master_pre_stage'

Neon

See merge request !18

Doxyfile.in

@@ -893,62 +893,61 @@ EXCLUDE                = @top_srcdir@/src/GPU/check_for_gpu.F90 \
 			@top_srcdir@/src/elpa2/elpa2_trans_ev_band_to_full_template.F90 \
 			@top_srcdir@/src/elpa2/elpa2_trans_ev_tridi_to_band_template.F90 \
 			@top_srcdir@/src/elpa2/kernels/simple_template.F90 \
-			@top_srcdir@/src/elpa2/kernels/complex_avx-avx2_1hv_template.c \
-			@top_srcdir@/src/elpa2/kernels/complex_avx-avx2_2hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/real_template.F90 \
 			@top_srcdir@/src/elpa2/kernels/complex_sse_1hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_6hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_6hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/complex_sse_1hv_double_precision.c \
-			@top_srcdir@/src/elpa2/kernels/real_sse_6hv_template.c \
+			@top_srcdir@/src/elpa2/kernels/real_128bit_256bit_512bit_BLOCK_template.c \
 			@top_srcdir@/src/elpa2/kernels/complex_template.F90 \
 			@top_srcdir@/src/elpa2/kernels/complex_avx-avx2_2hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx512_2hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx512_2hv_single_precision.c \
-			@top_srcdir@/src/elpa2/kernels/real_sse_4hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/complex_avx-avx2_2hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/complex_avx512_1hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_sse_6hv_single_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_sparc64_6hv_single_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_neon_arch64_6hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx512_4hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/asm_x86_64_double_precision.s \
 			@top_srcdir@/src/elpa2/kernels/real_avx512_4hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_sse_6hv_double_precision.c \
-			@top_srcdir@/src/elpa2/kernels/real_sse_2hv_template.c \
+			@top_srcdir@/src/elpa2/kernels/real_sparc64_6hv_double_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_neon_arch64_6hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/complex_avx512_1hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/asm_x86_64_single_precision.s \
-			@top_srcdir@/src/elpa2/kernels/real_avx512_6hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_2hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/complex_simple.F90 \
 			@top_srcdir@/src/elpa2/kernels/complex_avx-avx2_1hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real.F90 \
 			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_2hv_double_precision.c \
-			@top_srcdir@/src/elpa2/kernels/real_avx512_4hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/complex_avx-avx2_1hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/complex_avx512_2hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/mod_single_hh_trafo_real.F90 \
-			@top_srcdir@/src/elpa2/kernels/complex_sse_1hv_template.c \
+			@top_srcdir@/src/elpa2/kernels/complex_128bit_256bit_512bit_BLOCK_template.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_4hv_double_precision.c \
-			@top_srcdir@/src/elpa2/kernels/real_avx512_2hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/complex_avx512_2hv_double_precision.c \
-			@top_srcdir@/src/elpa2/kernels/complex_sse_2hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_4hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_bgq.f90 \
 			@top_srcdir@/src/elpa2/kernels/real_sse_2hv_single_precision.c \
-			@top_srcdir@/src/elpa2/kernels/complex_avx512_2hv_template.c \
+			@top_srcdir@/src/elpa2/kernels/real_sparc64_2hv_single_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_neon_arch64_2hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_simple.F90 \
 			@top_srcdir@/src/elpa2/kernels/complex.F90 \
-			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_2hv_template.c \
-			@top_srcdir@/src/elpa2/kernels/complex_avx512_1hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/real_sse_2hv_double_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_sparc64_2hv_double_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_neon_arch64_2hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/complex_sse_2hv_double_precision.c \
-			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_4hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/real_bgp.f90 \
 			@top_srcdir@/src/elpa2/kernels/real_sse_4hv_double_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_sparc64_4hv_double_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_neon_arch64_4hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx512_6hv_double_precision.c \
 			@top_srcdir@/src/elpa2/kernels/complex_sse_2hv_single_precision.c \
-			@top_srcdir@/src/elpa2/kernels/real_avx-avx2_6hv_template.c \
 			@top_srcdir@/src/elpa2/kernels/real_avx512_6hv_single_precision.c \
 			@top_srcdir@/src/elpa2/kernels/real_sse_4hv_single_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_sparc64_4hv_single_precision.c \
+			@top_srcdir@/src/elpa2/kernels/real_neon_arch64_4hv_single_precision.c \
 			@top_srcdir@/src/elpa2/elpa2_compute_complex_template.F90 \
 			@top_srcdir@/src/elpa2/elpa2_bandred_template.F90 \
 			@top_srcdir@/src/elpa2/pack_unpack_gpu.F90 \

Makefile.am

@@ -110,6 +110,7 @@ EXTRA_libelpa@SUFFIX@_private_la_DEPENDENCIES = \
   src/elpa2/kernels/complex_template.F90 \
   src/elpa2/kernels/simple_template.F90 \
   src/elpa2/kernels/simple_block4_template.F90 \
+  src/elpa2/kernels/simple_block6_template.F90 \
   src/elpa2/pack_unpack_cpu.F90 \
   src/elpa2/pack_unpack_gpu.F90 \
   src/elpa2/compute_hh_trafo.F90 \
@@ -194,9 +195,10 @@ if WITH_REAL_GENERIC_SIMPLE_BLOCK4_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_simple_block4.F90
 endif
 
-#if WITH_REAL_GENERIC_SIMPLE_BLOCK6_KERNEL
-#  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_simple_block6.F90
-#endif
+if WITH_REAL_GENERIC_SIMPLE_BLOCK6_KERNEL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_simple_block6.F90
+endif
+
 if WITH_REAL_BGP_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_bgp.f90
 endif
@@ -227,6 +229,13 @@ if WITH_REAL_SPARC64_BLOCK2_KERNEL
 #endif
 endif
 
+if WITH_REAL_NEON_ARCH64_BLOCK2_KERNEL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_neon_arch64_2hv_double_precision.c
+if WANT_SINGLE_PRECISION_REAL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_neon_arch64_2hv_single_precision.c
+endif
+endif
+
 if WITH_REAL_VSX_BLOCK2_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_2hv_double_precision.c
 if WANT_SINGLE_PRECISION_REAL
@@ -269,6 +278,13 @@ if WITH_REAL_SPARC64_BLOCK4_KERNEL
 #endif
 endif
 
+if WITH_REAL_NEON_ARCH64_BLOCK4_KERNEL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_neon_arch64_4hv_double_precision.c
+if WANT_SINGLE_PRECISION_REAL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_neon_arch64_4hv_single_precision.c
+endif
+endif
+
 if WITH_REAL_VSX_BLOCK4_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_4hv_double_precision.c
 if WANT_SINGLE_PRECISION_REAL
@@ -311,6 +327,13 @@ if WITH_REAL_SPARC64_BLOCK6_KERNEL
 #endif
 endif
 
+if WITH_REAL_NEON_ARCH64_BLOCK6_KERNEL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_neon_arch64_6hv_double_precision.c
+if WANT_SINGLE_PRECISION_REAL
+  libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_neon_arch64_6hv_single_precision.c
+endif
+endif
+
 if WITH_REAL_VSX_BLOCK6_KERNEL
   libelpa@SUFFIX@_private_la_SOURCES += src/elpa2/kernels/real_vsx_6hv_double_precision.c
 if WANT_SINGLE_PRECISION_REAL
@@ -768,28 +791,15 @@ EXTRA_DIST = \
   src/elpa2/elpa2_trans_ev_band_to_full_template.F90 \
   src/elpa2/elpa2_trans_ev_tridi_to_band_template.F90 \
   src/elpa2/elpa2_tridiag_band_template.F90 \
-  src/elpa2/kernels/complex_avx-avx2_1hv_template.c \
-  src/elpa2/kernels/complex_avx-avx2_2hv_template.c \
-  src/elpa2/kernels/complex_avx512_1hv_template.c \
-  src/elpa2/kernels/complex_avx512_2hv_template.c \
-  src/elpa2/kernels/complex_sse_1hv_template.c \
-  src/elpa2/kernels/complex_sse_2hv_template.c \
+  src/elpa2/kernels/complex_128bit_256bit_512bit_BLOCK_template.c \
   src/elpa2/kernels/complex_template.F90 \
-  src/elpa2/kernels/real_avx-avx2_2hv_template.c \
-  src/elpa2/kernels/real_avx-avx2_4hv_template.c \
-  src/elpa2/kernels/real_avx-avx2_6hv_template.c \
-  src/elpa2/kernels/real_avx512_2hv_template.c \
-  src/elpa2/kernels/real_avx512_4hv_template.c \
-  src/elpa2/kernels/real_avx512_6hv_template.c \
-  src/elpa2/kernels/real_vsx_2hv_template.c \
   src/elpa2/kernels/real_vsx_4hv_template.c \
   src/elpa2/kernels/real_vsx_6hv_template.c \
-  src/elpa2/kernels/real_sse_2hv_template.c \
-  src/elpa2/kernels/real_sse_4hv_template.c \
-  src/elpa2/kernels/real_sse_6hv_template.c \
+  src/elpa2/kernels/real_128bit_256bit_512bit_BLOCK_template.c \
   src/elpa2/kernels/real_template.F90 \
   src/elpa2/kernels/simple_template.F90 \
   src/elpa2/kernels/simple_block4_template.F90 \
+  src/elpa2/kernels/simple_block6_template.F90 \
   src/elpa2/pack_unpack_cpu.F90 \
   src/elpa2/pack_unpack_gpu.F90 \
   src/elpa2/qr/elpa_pdgeqrf_template.F90 \

ci_test_scripts/generate_gitlab_ci_tests.py

@@ -268,6 +268,8 @@ print("  - export BLOCK_SIZE=16")
 print("  - if [ \"$MEDIUM_MATRIX\" = \"yes\" ]; then export MATRIX_SIZE=1500 && export NUMBER_OF_EIGENVECTORS=750; fi")
 print("  - if [ \"$LARGE_MATRIX\" = \"yes\" ]; then export MATRIX_SIZE=5000 && export NUMBER_OF_EIGENVECTORS=500; fi")
 print("  - if [ \"$GPU_BLOCKSIZE\" = \"yes\" ]; then export BLOCK_SIZE=128 ; fi")
+print("  - if [ -z \"$PIPELINE_MPI_TASKS\" ]; then export MPI_TASKS=2; else xport MPI_TASKS=$PIPELINE_MPI_TASKS; fi")
+
 print("  - echo \"This test will run with matrix size na = $MATRIX_SIZE, nev= $NUMBER_OF_EIGENVECTORS, on a blacs grid with blocksize nblk= $BLOCK_SIZE \" ")
 print("  - export SKIP_STEP=0")
 print("  - ./autogen.sh")
@@ -320,7 +322,7 @@ print("    - avx")
 print("  script:")
 print("    - ./ci_test_scripts/run_ci_tests.sh -c \" CFLAGS=\\\"-O3 -mavx\\\" FCFLAGS=\\\"-O3 -axAVX\\\" SCALAPACK_LDFLAGS=\\\"$MKL_INTEL_SCALAPACK_LDFLAGS_NO_MPI_NO_OMP\\\"  \
         SCALAPACK_FCFLAGS=\\\"$MKL_INTEL_SCALAPACK_FCFLAGS_NO_MPI_NO_OMP\\\" --with-mpi=no FC=ifort --enable-shared=no --enable-static=yes --disable-avx2 --enable-optional-argument-in-C-API || { cat config.log; exit 1; } \" -j 8 \
-        -t 2 -m 150 -n 50 -b 16 -s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM ")
+        -t $MPI_TASKS -m 150 -n 50 -b 16 -s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM ")
 print("\n\n")
 
 print("# test distcheck")
@@ -330,10 +332,10 @@ print("    - distcheck")
 print("  script:")
 print("    - ./ci_test_scripts/run_ci_tests.sh -c \" CC=gcc FC=gfortran SCALAPACK_LDFLAGS=\\\"$MKL_GFORTRAN_SCALAPACK_LDFLAGS_NO_MPI_NO_OMP\\\"  \
         SCALAPACK_FCFLAGS=\\\"$MKL_GFORTRAN_SCALAPACK_FCFLAGS_NO_MPI_NO_OMP\\\" --enable-option-checking=fatal --with-mpi=no --disable-sse-assembly \
-        --disable-sse --disable-avx --disable-avx2 || { cat config.log; exit 1; } \" -t 2 -m 150 -n 50 -b 16 -s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM ")
+        --disable-sse --disable-avx --disable-avx2 || { cat config.log; exit 1; } \" -t $MPI_TASKS -m 150 -n 50 -b 16 -s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM ")
 print("    - ./ci_test_scripts/run_distcheck_tests.sh  -c \"  CC=gcc FC=gfortran SCALAPACK_LDFLAGS=\\\\\\\"$MKL_GFORTRAN_SCALAPACK_LDFLAGS_NO_MPI_NO_OMP\\\\\\\"  \
         SCALAPACK_FCFLAGS=\\\\\\\"$MKL_GFORTRAN_SCALAPACK_FCFLAGS_NO_MPI_NO_OMP\\\\\\\" --enable-option-checking=fatal --with-mpi=no --disable-sse-assembly \
-        --disable-sse --disable-avx --disable-avx2 \" -t 2 -m 150 -n 50 -b 16 -S$SLURM ")
+        --disable-sse --disable-avx --disable-avx2 \" -t $MPI_TASKS -m 150 -n 50 -b 16 -S$SLURM ")
 print("\n\n")
 
 print("distcheck-mpi:")
@@ -344,12 +346,12 @@ print("    - ./ci_test_scripts/run_ci_tests.sh -c \" FC=mpiifort FCFLAGS=\\\"-xH
   SCALAPACK_LDFLAGS=\\\"$MKL_INTEL_SCALAPACK_LDFLAGS_MPI_NO_OMP\\\" \
   SCALAPACK_FCFLAGS=\\\"$MKL_INTEL_SCALAPACK_FCFLAGS_MPI_NO_OMP\\\" \
   --enable-option-checking=fatal --with-mpi=yes \
- --disable-sse-assembly --disable-sse --disable-avx --disable-avx2 || { cat config.log; exit 1; } \" -t 2 -m 150 \
+ --disable-sse-assembly --disable-sse --disable-avx --disable-avx2 || { cat config.log; exit 1; } \" -t $MPI_TASKS -m 150 \
  -n 50 -b 16 -s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM ")
 print("    - ./ci_test_scripts/run_distcheck_tests.sh  -c \" FC=mpiifort FCFLAGS=\\\\\\\"-xHost\\\\\\\" \
  CFLAGS=\\\\\\\"-march=native\\\\\\\" SCALAPACK_LDFLAGS=\\\\\\\"$MKL_INTEL_SCALAPACK_LDFLAGS_MPI_NO_OMP\\\\\\\"  \
  SCALAPACK_FCFLAGS=\\\\\\\"$MKL_INTEL_SCALAPACK_FCFLAGS_MPI_NO_OMP\\\\\\\" --enable-option-checking=fatal \
- --with-mpi=yes --disable-sse-assembly --disable-sse --disable-avx --disable-avx2 \" -t 2 -m 150 -n 50 -b 16 -S$SLURM ")
+ --with-mpi=yes --disable-sse-assembly --disable-sse --disable-avx --disable-avx2 \" -t $MPI_TASKS -m 150 -n 50 -b 16 -S$SLURM ")
 print("\n\n")
 
 print("distcheck-no-autotune:")
@@ -360,12 +362,12 @@ print("    - ./ci_test_scripts/run_ci_tests.sh -c \" FC=mpiifort FCFLAGS=\\\"-xH
   SCALAPACK_LDFLAGS=\\\"$MKL_INTEL_SCALAPACK_LDFLAGS_MPI_NO_OMP\\\" \
   SCALAPACK_FCFLAGS=\\\"$MKL_INTEL_SCALAPACK_FCFLAGS_MPI_NO_OMP\\\" \
   --enable-option-checking=fatal --with-mpi=yes \
-  --disable-sse-assembly --disable-sse --disable-avx --disable-avx2 --disable-autotuning || { cat config.log; exit 1; } \" -t 2 -m 150 \
+  --disable-sse-assembly --disable-sse --disable-avx --disable-avx2 --disable-autotuning || { cat config.log; exit 1; } \" -t $MPI_TASKS -m 150 \
   -n 50 -b 16 -s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM ")
 print("    - ./ci_test_scripts/run_distcheck_tests.sh  -c \" FC=mpiifort FCFLAGS=\\\\\\\"-xHost\\\\\\\" \
  CFLAGS=\\\\\\\"-march=native\\\\\\\" SCALAPACK_LDFLAGS=\\\\\\\"$MKL_INTEL_SCALAPACK_LDFLAGS_MPI_NO_OMP\\\\\\\"  \
  SCALAPACK_FCFLAGS=\\\\\\\"$MKL_INTEL_SCALAPACK_FCFLAGS_MPI_NO_OMP\\\\\\\" --enable-option-checking=fatal \
- --with-mpi=yes --disable-sse-assembly --disable-sse --disable-avx --disable-avx2 --disable-autotuning \" -t 2 -m 150 -n 50 -b 16 -S$SLURM ")
+ --with-mpi=yes --disable-sse-assembly --disable-sse --disable-avx --disable-avx2 --disable-autotuning \" -t $MPI_TASKS -m 150 -n 50 -b 16 -S$SLURM ")
 print("\n\n")
 
 
@@ -389,7 +391,7 @@ python_ci_tests = [
     'SCALAPACK_FCFLAGS=\\"$MKL_ANACONDA_INTEL_SCALAPACK_FCFLAGS_MPI_OMP \\" '
     '--enable-option-checking=fatal --with-mpi=yes --enable-openmp '
     '--disable-gpu --enable-avx --enable-python --enable-python-tests || { cat config.log; exit 1; }'
-    '" -j 8 -t 2 -m $MATRIX_SIZE -n $NUMBER_OF_EIGENVECTORS -b $BLOCK_SIZE '
+    '" -j 8 -t $MPI_TASKS -m $MATRIX_SIZE -n $NUMBER_OF_EIGENVECTORS -b $BLOCK_SIZE '
     '-s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM',
     "\n",
     "python-distcheck:",
@@ -403,7 +405,7 @@ python_ci_tests = [
     'SCALAPACK_FCFLAGS=\\\"$MKL_ANACONDA_INTEL_SCALAPACK_FCFLAGS_MPI_OMP\\\" '
     '--enable-option-checking=fatal --with-mpi=yes --enable-openmp '
     '--disable-gpu --enable-avx --enable-python --enable-python-tests || { cat config.log; exit 1; }'
-    '" -j 8 -t 2 -m 150 -n 50 -b 16 '
+    '" -j 8 -t $MPI_TASKS -m 150 -n 50 -b 16 '
     '-s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM',
     "\n",
     '    - ./ci_test_scripts/run_distcheck_tests.sh -c "'
@@ -413,7 +415,7 @@ python_ci_tests = [
     'SCALAPACK_FCFLAGS=\\\\\\\"$MKL_ANACONDA_INTEL_SCALAPACK_FCFLAGS_MPI_OMP \\\\\\\" '
     '--enable-option-checking=fatal --with-mpi=yes --enable-openmp '
     '--disable-gpu --enable-avx --enable-python --enable-python-tests'
-    '" -t 2 -m 150 -n 50 -b 16 -S $SLURM  || { chmod u+rwX -R . ; exit 1 ; }',
+    '" -t $MPI_TASKS -m 150 -n 50 -b 16 -S $SLURM  || { chmod u+rwX -R . ; exit 1 ; }',
     "\n",
 ]
 print("\n".join(python_ci_tests))
@@ -456,7 +458,7 @@ for comp, s, a in product(
                     SCALAPACK_LDFLAGS=\\\"$MKL_INTEL_SCALAPACK_LDFLAGS_MPI_NO_OMP\\\" \
                     SCALAPACK_FCFLAGS=\\\"$MKL_INTEL_SCALAPACK_FCFLAGS_MPI_NO_OMP\\\" \
                     --enable-option-checking=fatal  --disable-avx2 --prefix=$PWD/installdest --disable-avx2 || { cat config.log; exit 1; } \" \
-                    -t 2 -m 150 -n 50 -b 16 -S $SLURM -p test_project_"+stage[s]+api[a]+" -e "+projectBinary+" \
+                    -t $MPI_TASKS -m 150 -n 50 -b 16 -S $SLURM -p test_project_"+stage[s]+api[a]+" -e "+projectBinary+" \
                     -C \" FC=mpiifort PKG_CONFIG_PATH=../../installdest/lib/pkgconfig  \
                      --enable-option-checking=fatal || { cat config.log; exit 1; } \" ")
         if (a == "legacy_api"):
@@ -464,7 +466,7 @@ for comp, s, a in product(
                     SCALAPACK_LDFLAGS=\\\"$MKL_INTEL_SCALAPACK_LDFLAGS_MPI_NO_OMP\\\" \
                     SCALAPACK_FCFLAGS=\\\"$MKL_INTEL_SCALAPACK_FCFLAGS_MPI_NO_OMP\\\" \
                     --enable-option-checking=fatal  --enable-legacy-interface --disable-avx2 --prefix=$PWD/installdest --disable-avx2 || { cat config.log; exit 1; } \" \
-                    -t 2 -m 150 -n 50 -b 16 -S $SLURM -p test_project_"+stage[s]+api[a]+" -e "+projectBinary+" \
+                    -t $MPI_TASKS -m 150 -n 50 -b 16 -S $SLURM -p test_project_"+stage[s]+api[a]+" -e "+projectBinary+" \
                     -C \" FC=mpiifort PKG_CONFIG_PATH=../../installdest/lib/pkgconfig  \
                      --enable-option-checking=fatal || { cat config.log; exit 1; } \" ")
 
@@ -474,7 +476,7 @@ for comp, s, a in product(
                     SCALAPACK_LDFLAGS=\\\"$MKL_GFORTRAN_SCALAPACK_LDFLAGS_MPI_NO_OMP\\\" \
                     SCALAPACK_FCFLAGS=\\\"$MKL_GFORTRAN_SCALAPACK_FCFLAGS_MPI_NO_OMP\\\" \
                     --enable-option-checking=fatal  --disable-avx2 --prefix=$PWD/installdest --disable-avx2 || { cat config.log; exit 1; } \" \
-                    -t 2 -m 150 -n 50 -b 16 -S $SLURM -p test_project_"+stage[s]+api[a]+" -e "+projectBinary+" \
+                    -t $MPI_TASKS -m 150 -n 50 -b 16 -S $SLURM -p test_project_"+stage[s]+api[a]+" -e "+projectBinary+" \
                     -C \" FC=mpif90 PKG_CONFIG_PATH=../../installdest/lib/pkgconfig \
                      --enable-option-checking=fatal || { cat config.log; exit 1; } \" ")
         if (a == "legacy_api"):
@@ -482,7 +484,7 @@ for comp, s, a in product(
                     SCALAPACK_LDFLAGS=\\\"$MKL_GFORTRAN_SCALAPACK_LDFLAGS_MPI_NO_OMP\\\" \
                     SCALAPACK_FCFLAGS=\\\"$MKL_GFORTRAN_SCALAPACK_FCFLAGS_MPI_NO_OMP\\\" \
                     --enable-option-checking=fatal  --enable-legacy-interface  --disable-avx2 --prefix=$PWD/installdest --disable-avx2 || { cat config.log; exit 1; } \" \
-                    -t 2 -m 150 -n 50 -b 16 -S $SLURM -p test_project_"+stage[s]+api[a]+" -e "+projectBinary+" \
+                    -t $MPI_TASKS -m 150 -n 50 -b 16 -S $SLURM -p test_project_"+stage[s]+api[a]+" -e "+projectBinary+" \
                     -C \" FC=mpif90 PKG_CONFIG_PATH=../../installdest/lib/pkgconfig \
                      --enable-option-checking=fatal || { cat config.log; exit 1; } \" ")
     print("\n\n")
@@ -559,7 +561,8 @@ matrix_size = {
         "small"   : "150",
 }
 
-MPI_TASKS=2
+#MPI_TASKS=2
+
 #                             sorted(coverage.keys()),     
 for cc, fc, m, o, p, a, b, g, instr, addr, na in product(
                              sorted(c_compiler.keys()),
@@ -699,9 +702,9 @@ for cc, fc, m, o, p, a, b, g, instr, addr, na in product(
 
     print("# " + cc + "-" + fc + "-" + m + "-" + o + "-" + p + "-" + a + "-" + b + "-" +g + "-" + cov + "-" + instr + "-" + addr)
     print(cc + "-" + fc + "-" + m + "-" + o + "-" + p + "-" +a + "-" +b + "-" +g + "-" + cov + "-" + instr + "-" + addr + "-jobs:")
-    if (MasterOnly):
-        print("  only:")
-        print("    - /.*master.*/")
+    #if (MasterOnly):
+    #    print("  only:")
+    #    print("    - /.*master.*/")
     if (instr == "power8"):
         print("  allow_failure: true")
     print("  tags:")
@@ -741,7 +744,7 @@ for cc, fc, m, o, p, a, b, g, instr, addr, na in product(
                 + libs + " " + ldflags + " " + " "+ scalapackldflags +" " + scalapackfcflags \
                 + " --enable-option-checking=fatal" + " " + mpi_configure_flag + " " + openmp[o] \
 + " " + precision[p] + " " + assumed_size[a] + " " + band_to_full_blocking[b] \
-+ " " +gpu[g] + INSTRUCTION_OPTIONS + "\" -j 8 -t " + str(MPI_TASKS) + " -m $MATRIX_SIZE -n $NUMBER_OF_EIGENVECTORS -b $BLOCK_SIZE -s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM")
++ " " +gpu[g] + INSTRUCTION_OPTIONS + "\" -j 8 -t $MPI_TASKS -m $MATRIX_SIZE -n $NUMBER_OF_EIGENVECTORS -b $BLOCK_SIZE -s $SKIP_STEP -i $INTERACTIVE_RUN -S $SLURM")
 
     if ( instr == "avx2" or instr == "avx512" or instr == "knl" or g == "with-gpu"):
         print("    - export REQUESTED_MEMORY="+memory)    
@@ -767,7 +770,7 @@ for cc, fc, m, o, p, a, b, g, instr, addr, na in product(
                 + libs + " " + ldflags + " " + " "+ scalapackldflags +" " + scalapackfcflags \
                 + " --enable-option-checking=fatal --enable-scalapack-tests" + " " + mpi_configure_flag + " " + openmp[o] \
                 + " " + precision[p] + " " + assumed_size[a] + " " + band_to_full_blocking[b] \
-                + " " +gpu[g] + INSTRUCTION_OPTIONS + "\" -j 8 -t " + str(MPI_TASKS) + " -m $MATRIX_SIZE -n $NUMBER_OF_EIGENVECTORS -b $BLOCK_SIZE -s $SKIP_STEP -q \"srun\" -S $SLURM")
+                + " " +gpu[g] + INSTRUCTION_OPTIONS + "\" -j 8 -t $MPI_TASKS -m $MATRIX_SIZE -n $NUMBER_OF_EIGENVECTORS -b $BLOCK_SIZE -s $SKIP_STEP -q \"srun\" -S $SLURM")
             
 
         else:
@@ -775,7 +778,7 @@ for cc, fc, m, o, p, a, b, g, instr, addr, na in product(
                 + libs + " " + ldflags + " " + " "+ scalapackldflags +" " + scalapackfcflags \
                 + " --enable-option-checking=fatal" + " " + mpi_configure_flag + " " + openmp[o] \
                 + " " + precision[p] + " " + assumed_size[a] + " " + band_to_full_blocking[b] \
-                + " " +gpu[g] + INSTRUCTION_OPTIONS + "\" -j 8 -t " + str(MPI_TASKS) + " -m $MATRIX_SIZE -n $NUMBER_OF_EIGENVECTORS -b $BLOCK_SIZE -s $SKIP_STEP -q \"srun\" -i $INTERACTIVE_RUN -S $SLURM")
+                + " " +gpu[g] + INSTRUCTION_OPTIONS + "\" -j 8 -t $MPI_TASKS -m $MATRIX_SIZE -n $NUMBER_OF_EIGENVECTORS -b $BLOCK_SIZE -s $SKIP_STEP -q \"srun\" -i $INTERACTIVE_RUN -S $SLURM")
 
     # do the test
 
@@ -787,12 +790,12 @@ for cc, fc, m, o, p, a, b, g, instr, addr, na in product(
                 openmp_threads=" 1 "
         else:
             openmp_threads=" 1 "
-        for na in sorted(matrix_size.keys(),reverse=True):
-            cores = set_number_of_cores(MPI_TASKS, o)
-            #print("    - echo \" srun  --ntasks=1 --cpus-per-task="+str(cores)+" $SRUN_COMMANDLINE_RUN\" ")
-            print("    -  echo \"na= $MATRIX_SIZE, nev= $NUMBER_OF_EIGENVECTORS nblock= $BLOCK_SIZE\" ")
-            #print("    - srun --ntasks-per-core=1 --ntasks=1 --cpus-per-task="+str(cores)+" $SRUN_COMMANDLINE_RUN \
-            #                             /scratch/elpa/bin/run_elpa.sh "+str(MPI_TASKS) + openmp_threads +" \" TEST_FLAGS=\\\" $MATRIX_SIZE $NUMBER_OF_EIGENVECTORS $BLOCK_SIZE " +"\\\"  || { cat test-suite.log; exit 1; }\"")
+        #for na in sorted(matrix_size.keys(),reverse=True):
+        #    cores = set_number_of_cores(MPI_TASKS, o)
+        #    #print("    - echo \" srun  --ntasks=1 --cpus-per-task="+str(cores)+" $SRUN_COMMANDLINE_RUN\" ")
+        #    print("    -  echo \"na= $MATRIX_SIZE, nev= $NUMBER_OF_EIGENVECTORS nblock= $BLOCK_SIZE\" ")
+        #    #print("    - srun --ntasks-per-core=1 --ntasks=1 --cpus-per-task="+str(cores)+" $SRUN_COMMANDLINE_RUN \
+        #    #                             /scratch/elpa/bin/run_elpa.sh "+str(MPI_TASKS) + openmp_threads +" \" TEST_FLAGS=\\\" $MATRIX_SIZE $NUMBER_OF_EIGENVECTORS $BLOCK_SIZE " +"\\\"  || { cat test-suite.log; exit 1; }\"")
 
         if (cov == "coverage"):
             print("    - ./ci_test_scripts/ci_coverage_collect")

ci_test_scripts/run_ci_tests.sh

@@ -125,7 +125,7 @@ then
     CLUSTER="draco"
   fi
 
-  echo "Running on $CLUSTER with runner $CI_RUNNER_DESCRIPTION with tag $CI_RUNNER_TAGS"
+  echo "Running on $CLUSTER with runner $CI_RUNNER_DESCRIPTION with tag $CI_RUNNER_TAGS on $mpiTasks tasks"
 
   # GPU runners
   if [ "$CI_RUNNER_TAGS" == "gpu" ]
@@ -144,16 +144,17 @@ then
     cat ./run_${CLUSTER}_1node_2GPU.sh
     echo " "
     echo "Submitting to SLURM"
-    sbatch -W ./run_${CLUSTER}_1node_2GPU.sh
-    exitCode=$?
+    if sbatch -W ./run_${CLUSTER}_1node_2GPU.sh; then
+      exitCode=$?
+    else
+      exitCode=$?
+      echo "Submission exited with exitCode $exitCode"
+    fi
 
-    echo " "
-    echo "Exit Code of sbatch: $exitCode"
-    echo " "
-    if (( $exitCode > 0 ))
-    then
+    #if (( $exitCode > 0 ))
+    #then
       cat ./ELPA_CI_2gpu.err.*
-    fi
+    #fi
     
   fi
 
@@ -174,24 +175,31 @@ then
     cat ./run_${CLUSTER}_1node.sh
     echo " "
     echo "Submitting to SLURM"
-    sbatch -W ./run_${CLUSTER}_1node.sh
-    exitCode=$?
+    if sbatch -W ./run_${CLUSTER}_1node.sh; then
+      exitCode=$?
+    else
+      exitCode=$?
+      echo "Submission excited with exitCode $exitCode"
+    fi
 
     echo " "
     echo "Exit Code of sbatch: $exitCode"
     echo " "
     cat ./ELPA_CI.out.*
-    if [ $exitCode -ne 0 ]
-    then
+    #if [ $exitCode -ne 0 ]
+    #then
       cat ./ELPA_CI.err.*
-    fi
+    #fi
 
   fi
 
-  if [ $exitCode -ne 0 ]
+  #if [ $exitCode -ne 0 ]
+  #then
+  if [ -f ./test-suite.log ]
   then
     cat ./test-suite.log
   fi
+  #fi
 
   exit $exitCode
 

ci_test_scripts/run_distcheck_tests.sh

@@ -102,7 +102,7 @@ then
     CLUSTER="draco"
   fi
 
-  echo "Running on $CLUSTER with runner $CI_RUNNER_DESCRIPTION with tag $CI_RUNNER_TAGS"
+  echo "Running on $CLUSTER with runner $CI_RUNNER_DESCRIPTION with tag $CI_RUNNER_TAGS on $mpiTasks tasks"
 
   #distcheck
   if [[ "$CI_RUNNER_TAGS" =~ "distcheck" ]]
@@ -126,17 +126,21 @@ then
     echo "Exit Code of sbatch: $exitCode"
     echo " "
     cat ./ELPA_CI.out.*
-    if [ $exitCode -ne 0 ]
-    then
+    #if [ $exitCode -ne 0 ]
+    #then
       cat ./ELPA_CI.err.*
+    #fi
+    if [ -f ./test-suite.log ]
+    then
+      cat ./test-suite.log
     fi
 
   fi
 
-  if [ $exitCode -ne 0 ]
-  then
+  #if [ $exitCode -ne 0 ]
+  #then
     cat ./test-suite.log
-  fi
+  #fi
 
   exit $exitCode
 

ci_test_scripts/run_project_tests.sh

@@ -120,7 +120,7 @@ then
     CLUSTER="draco"
   fi
 
-  echo "Running on $CLUSTER with runner $CI_RUNNER_DESCRIPTION with tag $CI_RUNNER_TAGS"
+  echo "Running on $CLUSTER with runner $CI_RUNNER_DESCRIPTION with tag $CI_RUNNER_TAGS on $mpiTasks tasks"
 
   #project_test
   if [[ "$CI_RUNNER_TAGS" =~ "project_test" ]]
@@ -129,18 +129,18 @@ then
     echo "mkdir -p build" >> ./run_${CLUSTER}_1node.sh
     echo "pushd build"  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running autogen " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running autogen " >> ./run_${CLUSTER}_1node.sh
     echo "../autogen.sh"  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running configure " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running configure " >> ./run_${CLUSTER}_1node.sh
     echo "../configure " "$configureArgs" >> ./run_${CLUSTER}_1node.sh
     echo " " >> ./run_${CLUSTER}_1node.sh
     echo "export TASKS=$mpiTasks" >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running make " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running make " >> ./run_${CLUSTER}_1node.sh
     echo "make -j 8"  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running make install" >> ./run_${CLUSTER}_1node.sh
+    echo "#Running make install" >> ./run_${CLUSTER}_1node.sh
     echo "make install" >> ./run_${CLUSTER}_1node.sh
     echo "popd" >> ./run_${CLUSTER}_1node.sh
     echo "mkdir -p $projectName/build" >> ./run_${CLUSTER}_1node.sh
@@ -149,19 +149,19 @@ then
     echo " "  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo " Testting project "  >> ./run_${CLUSTER}_1node.sh
+    echo " #Testting project "  >> ./run_${CLUSTER}_1node.sh
 
     echo " "  >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
 
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running autogen " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running autogen " >> ./run_${CLUSTER}_1node.sh
     echo "../autogen.sh" >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running configure " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running configure " >> ./run_${CLUSTER}_1node.sh
     echo "../configure " "$projectConfigureArgs " >> ./run_${CLUSTER}_1node.sh
     echo " "  >> ./run_${CLUSTER}_1node.sh
-    echo "Running make " >> ./run_${CLUSTER}_1node.sh
+    echo "#Running make " >> ./run_${CLUSTER}_1node.sh
     echo "make -j 8" >> ./run_${CLUSTER}_1node.sh
     echo "export LD_LIBRARY_PATH=$MKL_HOME/lib/intel64:\$LD_LIBRARY_PATH" >> ./run_${CLUSTER}_1node.sh
     echo "./$projectExecutable" >> ./run_${CLUSTER}_1node.sh
@@ -184,18 +184,17 @@ then
     echo "Exit Code of sbatch: $exitCode"
     echo " "
     cat ./ELPA_CI.out.*
-    if [ $exitCode -ne 0 ]
-    then
+    #if [ $exitCode -ne 0 ]
+    #then
       cat ./ELPA_CI.err.*
+    #fi
+    if [ -f ./test-suite.log ]
+    then
+      cat ./test-suite.log
     fi
 
   fi
 
-  if [ $exitCode -ne 0 ]
-  then
-    cat ./test-suite.log
-  fi
-
   exit $exitCode
 
 fi

configure.ac

@@ -50,13 +50,15 @@ if test x$_cv_gnu_make_command = x ; then
         AC_MSG_ERROR([Need GNU Make])
 fi
 
+enable_legacy=no
+
 AC_MSG_CHECKING(whether legacy interface should be provided)
 AC_ARG_ENABLE([legacy-interface],
               AS_HELP_STRING([--enable-legacy-interface],
                              [build legacy API, default no]),
               [
 	       if test x"$enableval" = x"yes"; then
-	         enable_legayc=yes
+	         enable_legacy=yes
 	       else
 	         enable_legacy=no
 	       fi
@@ -227,9 +229,9 @@ fi
 
 dnl check which MPI binray invokes a MPI job
 if test x"$with_mpi" = x"yes"; then
-  AC_CHECK_PROGS([MPI_BINARY], [mpiexec.hydra mpiexec mpirun poe runjob srun], [no])
+  AC_CHECK_PROGS([MPI_BINARY], [mpiexec.hydra mpiexec mpirun poe runjob srun aprun], [no])
   if test x"$MPI_BINARY" = x"no"; then
-    AC_MSG_ERROR([Could not find either of the MPI binaries: mpiexec.hydra, mpiexec, mpirun, poe, runjob, srun])
+    AC_MSG_ERROR([Could not find either of the MPI binaries: mpiexec.hydra, mpiexec, mpirun, poe, runjob, srun, aprun])
   fi
 fi
 
@@ -613,6 +615,7 @@ m4_define(elpa_m4_generic_kernels, [
         real_generic
         real_generic_simple
         real_generic_simple_block4
+        real_generic_simple_block6
         complex_generic
         complex_generic_simple
 ])
@@ -636,6 +639,12 @@ m4_define(elpa_m4_sparc64_kernels, [
         real_sparc64_block6
 ])
 
+m4_define(elpa_m4_neon_arch64_kernels, [
+        real_neon_arch64_block2
+        real_neon_arch64_block4
+        real_neon_arch64_block6
+])
+
 m4_define(elpa_m4_vsx_kernels, [
         real_vsx_block2
         real_vsx_block4
@@ -681,7 +690,7 @@ m4_define(elpa_m4_gpu_kernels, [
         complex_gpu
 ])
 
-m4_define(elpa_m4_kernel_types, [generic sparc64 vsx sse sse_assembly avx avx2 avx512 bgp bgq gpu])
+m4_define(elpa_m4_kernel_types, [generic sparc64 neon_arch64 vsx sse sse_assembly avx avx2 avx512 bgp bgq gpu])
 
 m4_define(elpa_m4_all_kernels,
           m4_foreach_w([elpa_m4_type],
@@ -715,6 +724,7 @@ AC_DEFUN([ELPA_SELECT_KERNELS], [
 dnl Modify list of kernels with configure arguments
 ELPA_SELECT_KERNELS([generic],[enable])
 ELPA_SELECT_KERNELS([sparc64],[disable])
+ELPA_SELECT_KERNELS([neon_arch64],[disable])
 ELPA_SELECT_KERNELS([vsx],[disable])
 ELPA_SELECT_KERNELS([sse],[enable])
 ELPA_SELECT_KERNELS([sse_assembly],[enable])
@@ -730,7 +740,7 @@ m4_foreach_w([elpa_m4_kind],[real complex],[
 ])
 
 if test x"${enable_bgp}" = x"yes" -o x"$enable_bgq" = x"yes"; then
-        m4_foreach_w([elpa_m4_kernel], elpa_m4_sparc64_kernels elpa_m4_vsx_kernels elpa_m4_sse_kernels elpa_m4_avx_kernels elpa_m4_avx2_kernels elpa_m4_avx512_kernels, [
+        m4_foreach_w([elpa_m4_kernel], elpa_m4_sparc64_kernels elpa_m4_neon_arch64_kernels elpa_m4_vsx_kernels elpa_m4_sse_kernels elpa_m4_avx_kernels elpa_m4_avx2_kernels elpa_m4_avx512_kernels, [
                 if x"$use_[]elpa_m4_kernel[]" = x"yes" ; then
                         echo "Disabling elpa_m4_kernel due to BGP/BGQ option"
                 fi
@@ -790,7 +800,7 @@ AC_DEFUN([ELPA_KERNEL_DEPENDS],[
                   ])
           fi