Merge pull request opencv#3083 from akarsakov:ocl_transpose_check_locmem

shubhra · Aug 13, 2014 · 3cd7941 · 3cd7941
2 parents 01bd496 + 3deea1c
commit 3cd7941
Showing 1 changed file with 7 additions and 0 deletions.
diff --git a/modules/core/src/matrix.cpp b/modules/core/src/matrix.cpp
@@ -3002,6 +3002,13 @@ static bool ocl_transpose( InputArray _src, OutputArray _dst )
         CV_Assert(dst.cols == dst.rows);
         kernelName += "_inplace";
     }
+    else
+    {
+        // check required local memory size
+        size_t required_local_memory = (size_t) TILE_DIM*(TILE_DIM+1)*CV_ELEM_SIZE(type);
+        if (required_local_memory > ocl::Device::getDefault().localMemSize())
+            return false;
+    }
 
     ocl::Kernel k(kernelName.c_str(), ocl::core::transpose_oclsrc,
                   format("-D T=%s -D T1=%s -D cn=%d -D TILE_DIM=%d -D BLOCK_ROWS=%d -D rowsPerWI=%d",