X-Git-Url: https://bilbo.iut-bm.univ-fcomte.fr/and/gitweb/book_gpu.git/blobdiff_plain/ecd2ddac55172a779e7e26d3bd5b1b2cb95033d6..f045ded06189b82188fcba9dd6ca383823e34aaa:/BookGPU/Chapters/chapter4/code/convoSepShV.cu~?ds=sidebyside

diff --git a/BookGPU/Chapters/chapter4/code/convoSepShV.cu~ b/BookGPU/Chapters/chapter4/code/convoSepShV.cu~
index 4437a56..a8bf71e 100644
--- a/BookGPU/Chapters/chapter4/code/convoSepShV.cu~
+++ b/BookGPU/Chapters/chapter4/code/convoSepShV.cu~
@@ -2,9 +2,10 @@ __global__ void kernel_convoSepShx8pV(unsigned char *output, int j_dim, int r)
 {
   int ic, jc, p;
   int k = 2*r+1 ;
-  float outval0=0.0, outval1=0.0, outval2=0.0, outval3=0.0, outval4=0.0, outval5=0.0, outval6=0.0, outval7=0.0 ;
-  int bdimX = blockDim.x<<3 ;
-  int tidX = threadIdx.x<<3 ;
+  float outval0=0.0, outval1=0.0, outval2=0.0, outval3=0.0 ;
+  float outval4=0.0, outval5=0.0, outval6=0.0, outval7=0.0 ;
+  int bdimX = blockDim.x<<3 ; // all packets width
+  int tidX = threadIdx.x<<3 ; // one packet offset
     
   // absolute coordinates of the base point
   int j = (__umul24(blockIdx.x,blockDim.x) + threadIdx.x)<<3 ; 
@@ -44,7 +45,7 @@ __global__ void kernel_convoSepShx8pV(unsigned char *output, int j_dim, int r)
 		outval7 += valMask*roi8p[ baseRoi +7 ] ;
 	  }
 	
-  // 8 pixel par thread --> global mem
+  // 8 pixels per thread --> global mem
   output[ idx++ ] = outval0 ;
   output[ idx++ ] = outval1 ;
   output[ idx++ ] = outval2 ;