Dynamic Dispatching for split3

opencv · Jan 25, 2022 · 5e89b9a · 5e89b9a
1 parent 266835c
commit 5e89b9a
Show file tree

Hide file tree

Showing 4 changed files with 43 additions and 16 deletions.
diff --git a/modules/gapi/src/backends/fluid/gfluidcore.cpp b/modules/gapi/src/backends/fluid/gfluidcore.cpp
@@ -2509,26 +2509,18 @@ GAPI_FLUID_KERNEL(GFluidSplit3, cv::gapi::core::GSplit3, false)
 
     static void run(const View &src, Buffer &dst1, Buffer &dst2, Buffer &dst3)
     {
-        const auto *in   =  src.InLine<uchar>(0);
-              auto *out1 = dst1.OutLine<uchar>();
-              auto *out2 = dst2.OutLine<uchar>();
-              auto *out3 = dst3.OutLine<uchar>();
+        const auto *in = src.InLine<uchar>(0);
+        auto *out1     = dst1.OutLine<uchar>();
+        auto *out2     = dst2.OutLine<uchar>();
+        auto *out3     = dst3.OutLine<uchar>();
 
         GAPI_Assert(3 == src.meta().chan);
         int width = src.length();
+        int w = 0;
 
-        int w = 0; // cycle counter
-
-    #if CV_SIMD128
-        for (; w <= width-16; w+=16)
-        {
-            v_uint8x16 a, b, c;
-            v_load_deinterleave(&in[3*w], a, b, c);
-            v_store(&out1[w], a);
-            v_store(&out2[w], b);
-            v_store(&out3[w], c);
-        }
-    #endif
+#if CV_SIMD
+        w = split3_simd(in, out1, out2, out3, width);
+#endif
 
         for (; w < width; w++)
         {

diff --git a/modules/gapi/src/backends/fluid/gfluidcore_func.dispatch.cpp b/modules/gapi/src/backends/fluid/gfluidcore_func.dispatch.cpp
@@ -207,6 +207,13 @@ ABSDIFFC_SIMD(float)
 
 #undef ABSDIFFC_SIMD
 
+int split3_simd(const uchar in[], uchar out1[], uchar out2[],
+                uchar out3[], const int width)
+{
+    CV_CPU_DISPATCH(split3_simd, (in, out1, out2, out3, width),
+                    CV_CPU_DISPATCH_MODES_ALL);
+}
+
 } // namespace fluid
 } // namespace gapi
 } // namespace cv

diff --git a/modules/gapi/src/backends/fluid/gfluidcore_func.hpp b/modules/gapi/src/backends/fluid/gfluidcore_func.hpp
@@ -163,6 +163,9 @@ ABSDIFFC_SIMD(float)
 
 #undef ABSDIFFC_SIMD
 
+int split3_simd(const uchar in[], uchar out1[], uchar out2[],
+                uchar out3[], const int width);
+
 }  // namespace fluid
 }  // namespace gapi
 }  // namespace cv

diff --git a/modules/gapi/src/backends/fluid/gfluidcore_func.simd.hpp b/modules/gapi/src/backends/fluid/gfluidcore_func.simd.hpp
@@ -184,6 +184,9 @@ ABSDIFFC_SIMD(float)
 
 #undef ABSDIFFC_SIMD
 
+int split3_simd(const uchar in[], uchar out1[], uchar out2[],
+                uchar out3[], const int width);
+
 #ifndef CV_CPU_OPTIMIZATION_DECLARATIONS_ONLY
 
 struct scale_tag {};
@@ -1568,6 +1571,28 @@ ABSDIFFC_SIMD(float)
 
 #undef ABSDIFFC_SIMD
 
+//-------------------------
+//
+// Fluid kernels: Split3
+//
+//-------------------------
+
+int split3_simd(const uchar in[], uchar out1[], uchar out2[],
+                uchar out3[], const int width)
+{
+    constexpr int nlanes = v_uint8::nlanes;
+    int x = 0;
+    for (; x <= width - nlanes; x += nlanes)
+    {
+        v_uint8 a, b, c;
+        v_load_deinterleave(&in[3 * x], a, b, c);
+        vx_store(&out1[x], a);
+        vx_store(&out2[x], b);
+        vx_store(&out3[x], c);
+    }
+    return x;
+}
+
 #endif  // CV_CPU_OPTIMIZATION_DECLARATIONS_ONLY
 
 CV_CPU_OPTIMIZATION_NAMESPACE_END