Optimize Simd handling for g++-12

alpaka-group · Nov 3, 2022 · 165d8c3 · 165d8c3
1 parent c19e566
commit 165d8c3
Showing 1 changed file with 15 additions and 9 deletions.
diff --git a/include/llama/Simd.hpp b/include/llama/Simd.hpp
@@ -206,19 +206,23 @@ namespace llama
                     //}
                     else if constexpr(mapping::isAoS<Mapping>)
                     {
-                        constexpr static auto stride
+                        constexpr static auto srcStride
                             = flatSizeOf<typename Mapping::Flattener::FlatRecordDim, Mapping::alignAndPad>;
-                        auto* base = reinterpret_cast<const std::byte*>(&srcRef(rc));
+                        const auto* srcBaseAddr = reinterpret_cast<const std::byte*>(&srcRef(rc));
+                        ElementSimd elemSimd; // g++-12 really needs the intermediate elemSimd and memcpy
                         for(auto i = 0; i < Traits::lanes; i++)
-                            reinterpret_cast<FieldType*>(&dstSimd(rc))[i]
-                                = *reinterpret_cast<const FieldType*>(base + i * stride);
+                            reinterpret_cast<FieldType*>(&elemSimd)[i]
+                                = *reinterpret_cast<const FieldType*>(srcBaseAddr + i * srcStride);
+                        std::memcpy(&dstSimd(rc), &elemSimd, sizeof(elemSimd));
                     }
                     else
                     {
                         auto b = ArrayIndexIterator{srcRef.view.mapping().extents(), srcRef.arrayIndex()};
+                        ElementSimd elemSimd; // g++-12 really needs the intermediate elemSimd and memcpy
                         for(auto i = 0; i < Traits::lanes; i++)
-                            reinterpret_cast<FieldType*>(&dstSimd(rc))[i]
+                            reinterpret_cast<FieldType*>(&elemSimd)[i]
                                 = srcRef.view(*b++)(cat(typename T::BoundRecordCoord{}, rc)); // scalar loads
+                        std::memcpy(&dstSimd(rc), &elemSimd, sizeof(elemSimd));
                     }
                 });
         }
@@ -266,19 +270,21 @@ namespace llama
                     {
                         constexpr static auto stride
                             = flatSizeOf<typename Mapping::Flattener::FlatRecordDim, Mapping::alignAndPad>;
-                        auto* base = reinterpret_cast<std::byte*>(&dstRef(rc));
+                        auto* dstBaseAddr = reinterpret_cast<std::byte*>(&dstRef(rc));
+                        const ElementSimd elemSimd = srcSimd(rc);
                         for(auto i = 0; i < Traits::lanes; i++)
-                            *reinterpret_cast<FieldType*>(base + i * stride)
-                                = reinterpret_cast<const FieldType*>(&srcSimd(rc))[i];
+                            *reinterpret_cast<FieldType*>(dstBaseAddr + i * stride)
+                                = reinterpret_cast<const FieldType*>(&elemSimd)[i];
                     }
                     else
                     {
                         // TODO(bgruber): how does this generalize conceptually to 2D and higher dimensions? in which
                         // direction should we collect SIMD values?
+                        const ElementSimd elemSimd = srcSimd(rc);
                         auto b = ArrayIndexIterator{dstRef.view.mapping().extents(), dstRef.arrayIndex()};
                         for(auto i = 0; i < Traits::lanes; i++)
                             dstRef.view (*b++)(cat(typename T::BoundRecordCoord{}, rc))
-                                = reinterpret_cast<const FieldType*>(&srcSimd(rc))[i]; // scalar store
+                                = reinterpret_cast<const FieldType*>(&elemSimd)[i]; // scalar store
                     }
                 });
         }