From df697f65209713976715944b735c67a6a1012c2e Mon Sep 17 00:00:00 2001
From: Yan Wang <yan.wang@linux.intel.com>
Date: Thu, 22 Jan 2015 16:36:16 +0800
Subject: [PATCH] Optimize runLBPClassifierStumpSimple by built-in mad24.

Signed-off-by: Yan Wang <yan.wang@linux.intel.com>
---
 modules/objdetect/src/opencl/cascadedetect.cl | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/modules/objdetect/src/opencl/cascadedetect.cl b/modules/objdetect/src/opencl/cascadedetect.cl
index ce676426f5..465fa0c65d 100644
--- a/modules/objdetect/src/opencl/cascadedetect.cl
+++ b/modules/objdetect/src/opencl/cascadedetect.cl
@@ -397,8 +397,8 @@ __kernel void runLBPClassifierStumpSimple(
 
         for( tileIdx = groupIdx; tileIdx < totalTiles; tileIdx += ngroups )
         {
-            int iy = ((tileIdx / ntiles.x)*local_size_y + ly)*ystep;
-            int ix = ((tileIdx % ntiles.x)*local_size_x + lx)*ystep;
+            int iy = mad24((tileIdx / ntiles.x), local_size_y, ly) * ystep;
+            int ix = mad24((tileIdx % ntiles.x), local_size_x, lx) * ystep;
 
             if( ix < worksize.x && iy < worksize.y )
             {