From: Aaron Kunze <aaron.kunze@intel.com>
Date: Wed, 9 Apr 2014 15:42:55 +0000 (-0700)
Subject: Fixes global size issue, adds #pragma unroll to loops
X-Git-Tag: accepted/tizen/6.0/unified/20201030.111113~3315^2
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=785acc183473293050c6881fbc148b0231f3cf4f;p=platform%2Fupstream%2Fopencv.git

Fixes global size issue, adds #pragma unroll to loops
---

diff --git a/modules/imgproc/src/color.cpp b/modules/imgproc/src/color.cpp
index f3f8772..16a475a 100644
--- a/modules/imgproc/src/color.cpp
+++ b/modules/imgproc/src/color.cpp
@@ -2690,6 +2690,8 @@ struct mRGBA2RGBA
 
 #ifdef HAVE_OPENCL
 
+#define DIVUP(total, grain) (((total) + (grain) - 1) / (grain))
+
 static bool ocl_cvtColor( InputArray _src, OutputArray _dst, int code, int dcn )
 {
     bool ok = false;
@@ -2711,7 +2713,7 @@ static bool ocl_cvtColor( InputArray _src, OutputArray _dst, int code, int dcn )
     {
         pxPerWIy = 4;
     }
-    globalsize[1] /= pxPerWIy;
+    globalsize[1] = DIVUP(globalsize[1], pxPerWIy);
     opts +=  format("-D PIX_PER_WI_Y=%d ", pxPerWIy);
 
     switch (code)
diff --git a/modules/imgproc/src/opencl/cvtcolor.cl b/modules/imgproc/src/opencl/cvtcolor.cl
index 098d5fb..0034395 100644
--- a/modules/imgproc/src/opencl/cvtcolor.cl
+++ b/modules/imgproc/src/opencl/cvtcolor.cl
@@ -131,6 +131,7 @@ __kernel void RGB2Gray(__global const uchar* srcptr, int srcstep, int srcoffset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -158,6 +159,7 @@ __kernel void Gray2RGB(__global const uchar* srcptr, int srcstep, int srcoffset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -189,6 +191,7 @@ __kernel void RGB2YUV(__global const uchar* srcptr, int srcstep, int srcoffset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -232,6 +235,7 @@ __kernel void YUV2RGB(__global const uchar* srcptr, int srcstep, int srcoffset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -281,6 +285,7 @@ __kernel void YUV2RGB_NV12(__global const uchar* srcptr, int srcstep, int srcoff
 
     if (x < cols / 2)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows / 2 )
@@ -353,6 +358,7 @@ __kernel void RGB2YCrCb(__global const uchar* srcptr, int srcstep, int srcoffset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -396,6 +402,7 @@ __kernel void YCrCb2RGB(__global const uchar* src, int src_step, int src_offset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -443,6 +450,7 @@ __kernel void RGB2XYZ(__global const uchar * srcptr, int src_step, int src_offse
 
     if (dx < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (dy < rows)
@@ -483,6 +491,7 @@ __kernel void XYZ2RGB(__global const uchar * srcptr, int src_step, int src_offse
 
     if (dx < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (dy < rows)
@@ -528,6 +537,7 @@ __kernel void RGB(__global const uchar* srcptr, int src_step, int src_offset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -573,6 +583,7 @@ __kernel void RGB5x52RGB(__global const uchar* src, int src_step, int src_offset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -613,6 +624,7 @@ __kernel void RGB2RGB5x5(__global const uchar* src, int src_step, int src_offset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -646,6 +658,7 @@ __kernel void BGR5x52Gray(__global const uchar* src, int src_step, int src_offse
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -678,6 +691,7 @@ __kernel void Gray2BGR5x5(__global const uchar* src, int src_step, int src_offse
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -719,6 +733,7 @@ __kernel void RGB2HSV(__global const uchar* src, int src_step, int src_offset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -765,6 +780,7 @@ __kernel void HSV2RGB(__global const uchar* src, int src_step, int src_offset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -828,6 +844,7 @@ __kernel void RGB2HSV(__global const uchar* srcptr, int src_step, int src_offset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -880,6 +897,7 @@ __kernel void HSV2RGB(__global const uchar* srcptr, int src_step, int src_offset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -950,6 +968,7 @@ __kernel void RGB2HLS(__global const uchar* src, int src_step, int src_offset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1004,6 +1023,7 @@ __kernel void HLS2RGB(__global const uchar* src, int src_step, int src_offset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1066,6 +1086,7 @@ __kernel void RGB2HLS(__global const uchar* srcptr, int src_step, int src_offset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1123,6 +1144,7 @@ __kernel void HLS2RGB(__global const uchar* srcptr, int src_step, int src_offset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1193,6 +1215,7 @@ __kernel void RGBA2mRGBA(__global const uchar* src, int src_step, int src_offset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1223,6 +1246,7 @@ __kernel void mRGBA2RGBA(__global const uchar* src, int src_step, int src_offset
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1275,6 +1299,7 @@ __kernel void BGR2Lab(__global const uchar * src, int src_step, int src_offset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1322,6 +1347,7 @@ __kernel void BGR2Lab(__global const uchar * srcptr, int src_step, int src_offse
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1430,6 +1456,7 @@ __kernel void Lab2BGR(__global const uchar * src, int src_step, int src_offset,
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)
@@ -1478,6 +1505,7 @@ __kernel void Lab2BGR(__global const uchar * srcptr, int src_step, int src_offse
 
     if (x < cols)
     {
+        #pragma unroll
         for (int cy = 0; cy < PIX_PER_WI_Y; ++cy)
         {
             if (y < rows)