diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
new file mode 100644
index 0000000000000000000000000000000000000000..b8bf29826722aa1c6b650f30d7ab55af06940b29
--- /dev/null
+++ b/.gitlab-ci.yml
@@ -0,0 +1,31 @@
+---
+variables:
+# work around https://gitlab.com/gitlab-org/gitlab-ci-multi-runner/issues/2148
+# value should be "recursive" without workaround
+  GIT_SUBMODULE_STRATEGY: none
+
+before_script:
+- echo "Running make concurrently with $(nproc) cores, where possibly"
+# work around https://gitlab.com/gitlab-org/gitlab-ci-multi-runner/issues/2148
+- git submodule sync --recursive
+- git submodule update --init --recursive
+# end workaround
+# make sure we can apply the patches
+- git submodule foreach --recursive git reset --hard HEAD
+- patches/apply_patches.sh
+# detect number of processors, but make sure never to run with -j"", as the
+# build host would likely run out of memory
+- ": ${DUNECI_PARALLEL:=$(nproc || echo 1)}"
+- export DUNECI_PARALLEL
+
+git--gcc:
+  image: registry.dune-project.org/joe/dune-perftool-ci/perftool-base:git
+  script: duneci-standard-test
+
+git--clang:
+  image: registry.dune-project.org/joe/dune-perftool-ci/perftool-base:git
+  script: duneci-standard-test --opts=/duneci/opts.clang
+
+git--pacxx:
+  image: registry.dune-project.org/joe/dune-perftool-ci/pacxx-perftool-base:git
+  script: duneci-standard-test --opts=/duneci/opts.pacxx
diff --git a/dune/perftool/common/opcounter.hh b/dune/perftool/common/opcounter.hh
index dbdb595cd39f65a98f43ffdabecd7ab31a83a2c0..2762695e8ad5f11acb5f68d83ed70c0457cda55e 100644
--- a/dune/perftool/common/opcounter.hh
+++ b/dune/perftool/common/opcounter.hh
@@ -116,7 +116,12 @@ namespace oc {
 
     F _v;
 
-    struct Counters {
+    // The alignment is necessary to work around
+    // https://bugs.llvm.org/show_bug.cgi?id=34219:
+    // "SLP vectorizer: aligned store to unaligned address"
+    // The bug is present in 3.8.0 (tags/RELEASE_380/final 263969) and in
+    // clang version 6.0.0 (trunk 310993)
+    struct alignas(16) Counters {
 
       size_type addition_count;
       size_type multiplication_count;