Adds a GitHub CI step for OpenMP and some formatting changes

intel · May 22, 2024 · 5c81fb1 · 5c81fb1
1 parent 71a9cfa
commit 5c81fb1
Show file tree

Hide file tree

Showing 2 changed files with 80 additions and 28 deletions.
diff --git a/.github/workflows/c-cpp.yml b/.github/workflows/c-cpp.yml
@@ -135,6 +135,38 @@ jobs:
     - name: Run test suite on SPR
       run: sde -spr -- ./builddir/testexe
 
+  SKX-SKL-openmp:
+
+    runs-on: intel-ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
+
+    - name: Install dependencies
+      run: |
+        sudo apt update
+        sudo apt -y install g++-10 libgtest-dev meson curl git
+
+    - name: Install Intel SDE
+      run: |
+        curl -o /tmp/sde.tar.xz https://downloadmirror.intel.com/784319/sde-external-9.24.0-2023-07-13-lin.tar.xz
+        mkdir /tmp/sde && tar -xvf /tmp/sde.tar.xz -C /tmp/sde/
+        sudo mv /tmp/sde/* /opt/sde && sudo ln -s /opt/sde/sde64 /usr/bin/sde
+
+    - name: Build
+      env:
+        CXX: g++-10
+      run: |
+        make clean
+        meson setup -Dbuild_tests=true -Duse_openmp=true --warnlevel 2 --werror --buildtype release builddir
+        cd builddir
+        ninja
+
+    - name: Run test suite on SKX and SKL
+      run: |
+        sde -skx -- ./builddir/testexe
+        sde -skl -- ./builddir/testexe
+
   SPR-gcc13-special-cases:
 
     runs-on: intel-ubuntu-latest

diff --git a/src/xss-common-keyvaluesort.hpp b/src/xss-common-keyvaluesort.hpp
@@ -393,29 +393,46 @@ X86_SIMD_SORT_INLINE void kvsort_(type1_t *keys,
     arrsize_t pivot_index = kvpartition_unrolled<vtype1, vtype2, 4>(
             keys, indexes, left, right + 1, pivot, &smallest, &biggest);
 
-
 #if defined(XSS_USE_OPENMP) && defined(_OPENMP)
     if (pivot != smallest) {
-        bool parallelLeft = (pivot_index - left) > task_threshold;
-        if (parallelLeft){
-            #pragma omp task if(parallelLeft)
-            kvsort_<vtype1, vtype2>(
-                keys, indexes, left, pivot_index - 1, max_iters - 1, task_threshold);
-        }else{
-            kvsort_<vtype1, vtype2>(
-                keys, indexes, left, pivot_index - 1, max_iters - 1, task_threshold);
+        bool parallel_left = (pivot_index - left) > task_threshold;
+        if (parallel_left) {
+#pragma omp task
+            kvsort_<vtype1, vtype2>(keys,
+                                    indexes,
+                                    left,
+                                    pivot_index - 1,
+                                    max_iters - 1,
+                                    task_threshold);
+        }
+        else {
+            kvsort_<vtype1, vtype2>(keys,
+                                    indexes,
+                                    left,
+                                    pivot_index - 1,
+                                    max_iters - 1,
+                                    task_threshold);
         }
     }
     if (pivot != biggest) {
-        bool parallelRight = (right - pivot_index) > task_threshold;
-
-        if (parallelRight){
-            #pragma omp task if(parallelRight)
-            kvsort_<vtype1, vtype2>(
-                keys, indexes, pivot_index, right, max_iters - 1, task_threshold);
-        }else{
-            kvsort_<vtype1, vtype2>(
-                keys, indexes, pivot_index, right, max_iters - 1, task_threshold); 
+        bool parallel_right = (right - pivot_index) > task_threshold;
+
+        if (parallel_right) {
+#pragma omp task
+            kvsort_<vtype1, vtype2>(keys,
+                                    indexes,
+                                    pivot_index,
+                                    right,
+                                    max_iters - 1,
+                                    task_threshold);
+        }
+        else {
+            kvsort_<vtype1, vtype2>(keys,
+                                    indexes,
+                                    pivot_index,
+                                    right,
+                                    max_iters - 1,
+                                    task_threshold);
         }
     }
 #else
@@ -518,19 +535,22 @@ X86_SIMD_SORT_INLINE void xss_qsort_kv(
         }
 
 #if defined(XSS_USE_OPENMP) && defined(_OPENMP)
-        bool useParallel = arrsize > 10000;
-        arrsize_t taskThreshold = std::max((arrsize_t) 10000, arrsize / 100);
-        if (useParallel){
-            #pragma omp parallel
-            #pragma omp single
-            kvsort_<keytype, valtype>(keys, indexes, 0, arrsize - 1, maxiters, taskThreshold);
-        }else{
-            kvsort_<keytype, valtype>(keys, indexes, 0, arrsize - 1, maxiters, taskThreshold);
+        bool use_parallel = arrsize > 10000;
+        arrsize_t task_threshold = std::max((arrsize_t)10000, arrsize / 100);
+        if (use_parallel) {
+#pragma omp parallel
+#pragma omp single
+            kvsort_<keytype, valtype>(
+                    keys, indexes, 0, arrsize - 1, maxiters, task_threshold);
+        }
+        else {
+            kvsort_<keytype, valtype>(
+                    keys, indexes, 0, arrsize - 1, maxiters, task_threshold);
         }
 #else
         kvsort_<keytype, valtype>(keys, indexes, 0, arrsize - 1, maxiters, 0);
-#endif    
-        
+#endif
+
         replace_inf_with_nan(keys, arrsize, nan_count);
 
         if (descending) {