Merge branch '2023.06-software.eessi.io' into 2023.06-sw-R-bundle-CRA…

…N-2023.12-foss-2023a

easystacks/software.eessi.io/2023.06/eessi-2023.06-eb-4.9.0-2023a.yml

@@ -8,4 +8,7 @@ easyconfigs:
       # see https://github.com/easybuilders/easybuild-easyconfigs/pull/19592    
       options:
         from-pr: 19592
-  - R-bundle-CRAN-2023.12-foss-2023a.eb
+  - Rivet-3.1.9-gompi-2023a-HepMC3-3.2.6.eb:
+      options:
+        from-pr: 19679
+  - R-bundle-CRAN-2023.12-foss-2023a.eb

easystacks/software.eessi.io/2023.06/eessi-2023.06-eb-4.9.0-2023b.yml

@@ -5,3 +5,6 @@ easyconfigs:
   - netCDF-4.9.2-gompi-2023b.eb:
       options:
         from-pr: 19534
+  - matplotlib-3.8.2-gfbf-2023b.eb:
+      options:
+        from-pr: 19552

eb_hooks.py

@@ -185,20 +185,21 @@ def parse_hook_fontconfig_add_fonts(ec, eprefix):
 
 
 def parse_hook_openblas_relax_lapack_tests_num_errors(ec, eprefix):
-    """Relax number of failing numerical LAPACK tests for aarch64/neoverse_v1 CPU target."""
+    """Relax number of failing numerical LAPACK tests for aarch64/neoverse_v1 CPU target for OpenBLAS < 0.3.23"""
     cpu_target = get_eessi_envvar('EESSI_SOFTWARE_SUBDIR')
     if ec.name == 'OpenBLAS':
-        # relax maximum number of failed numerical LAPACK tests for aarch64/neoverse_v1 CPU target
-        # since the default setting of 150 that works well on other aarch64 targets and x86_64 is a bit too strict
-        # See https://github.com/EESSI/software-layer/issues/314
-        cfg_option = 'max_failing_lapack_tests_num_errors'
-        if cpu_target == CPU_TARGET_NEOVERSE_V1:
-            orig_value = ec[cfg_option]
-            ec[cfg_option] = 400
-            print_msg("Maximum number of failing LAPACK tests with numerical errors for %s relaxed to %s (was %s)",
-                      ec.name, ec[cfg_option], orig_value)
-        else:
-            print_msg("Not changing option %s for %s on non-AARCH64", cfg_option, ec.name)
+        if LooseVersion(ec.version) < LooseVersion('0.3.23'):
+            # relax maximum number of failed numerical LAPACK tests for aarch64/neoverse_v1 CPU target
+            # since the default setting of 150 that works well on other aarch64 targets and x86_64 is a bit too strict
+            # See https://github.com/EESSI/software-layer/issues/314
+            cfg_option = 'max_failing_lapack_tests_num_errors'
+            if cpu_target == CPU_TARGET_NEOVERSE_V1:
+                orig_value = ec[cfg_option]
+                ec[cfg_option] = 400
+                print_msg("Maximum number of failing LAPACK tests with numerical errors for %s relaxed to %s (was %s)",
+                          ec.name, ec[cfg_option], orig_value)
+            else:
+                print_msg("Not changing option %s for %s on non-AARCH64", cfg_option, ec.name)
     else:
         raise EasyBuildError("OpenBLAS-specific hook triggered for non-OpenBLAS easyconfig?!")
 

eessi-2023.06-known-issues.yml

@@ -1,7 +1,28 @@
 - aarch64/neoverse_v1:
+  - ESPResSo-4.2.1-foss-2023a:
+      - issue: https://github.com/EESSI/software-layer/issues/363
+      - info: "ESPResSo tests failing due to timeouts"
+  - FFTW.MPI-3.3.10-gompi-2023a:
+      - issue: https://github.com/EESSI/software-layer/issues/325
+      - info: "Flaky FFTW tests, random failures"
+  - FFTW.MPI-3.3.10-gompi-2023b:
+      - issue: https://github.com/EESSI/software-layer/issues/325
+      - info: "Flaky FFTW tests, random failures"
   - netCDF-4.9.2-gompi-2023a.eb:
       - issue: https://github.com/EESSI/software-layer/issues/425
       - info: "netCDF intermittent test failures"
   - netCDF-4.9.2-gompi-2023b.eb:
       - issue: https://github.com/EESSI/software-layer/issues/425
       - info: "netCDF intermittent test failures"
+  - OpenBLAS-0.3.21-GCC-12.2.0:
+    - issue: https://github.com/EESSI/software-layer/issues/314
+    - info: "Increased number of numerical errors in OpenBLAS test suite (344 vs max. 150 on x86_64/*)"
+  - SciPy-bundle-2023.02-gfbf-2022b:
+    - issue: https://github.com/EESSI/software-layer/issues/318
+    - info: "numpy built with -march=armv8.4-a instead of -mcpu=native (no SVE) + 2 failing tests (vs 50005 passed) in scipy test suite"
+  - SciPy-bundle-2023.07-gfbf-2023a:
+    - issue: https://github.com/EESSI/software-layer/issues/318
+    - info: "2 failing tests (vs 54409 passed) in scipy test suite"
+  - SciPy-bundle-2023.11-gfbf-2023b:
+    - issue: https://github.com/EESSI/software-layer/issues/318
+    - info: "2 failing tests (vs 54876 passed) in scipy test suite"

init/eessi_environment_variables

@@ -27,8 +27,17 @@ if [ -d $EESSI_PREFIX ]; then
     # determine subdirectory in software layer
     if [ "$EESSI_USE_ARCHDETECT" == "1" ]; then
       # if archdetect is enabled, use internal code
-      export EESSI_SOFTWARE_SUBDIR=$(${EESSI_INIT_DIR_PATH}/eessi_archdetect.sh cpupath)
-      show_msg "archdetect says ${EESSI_SOFTWARE_SUBDIR}"
+      all_cpupaths=$(${EESSI_INIT_DIR_PATH}/eessi_archdetect.sh -a cpupath)
+      # iterate over colon-separated list verifying if the architecture is present
+      #   under $EESSI_PREFIX/software/$EESSI_OS_TYPE; if so use the architecture as best match
+      IFS=: read -r -a archs <<< "${all_cpupaths}"
+      for arch in "${archs[@]}"; do
+        if [ -d ${EESSI_PREFIX}/software/${EESSI_OS_TYPE}/${arch} ]; then
+          export EESSI_SOFTWARE_SUBDIR=${arch}
+          show_msg "archdetect says ${EESSI_SOFTWARE_SUBDIR}"
+          break
+        fi
+      done
     elif [ "$EESSI_USE_ARCHSPEC" == "1" ]; then
       # note: eessi_software_subdir_for_host.py will pick up value from $EESSI_SOFTWARE_SUBDIR_OVERRIDE if it's defined!
       export EESSI_EPREFIX_PYTHON=$EESSI_EPREFIX/usr/bin/python3