Merge pull request #3637 from wangyan42164:LBP_mad24_opt
This commit is contained in:
commit
bdebf4cea7
@ -397,8 +397,8 @@ __kernel void runLBPClassifierStumpSimple(
|
|||||||
|
|
||||||
for( tileIdx = groupIdx; tileIdx < totalTiles; tileIdx += ngroups )
|
for( tileIdx = groupIdx; tileIdx < totalTiles; tileIdx += ngroups )
|
||||||
{
|
{
|
||||||
int iy = ((tileIdx / ntiles.x)*local_size_y + ly)*ystep;
|
int iy = mad24((tileIdx / ntiles.x), local_size_y, ly) * ystep;
|
||||||
int ix = ((tileIdx % ntiles.x)*local_size_x + lx)*ystep;
|
int ix = mad24((tileIdx % ntiles.x), local_size_x, lx) * ystep;
|
||||||
|
|
||||||
if( ix < worksize.x && iy < worksize.y )
|
if( ix < worksize.x && iy < worksize.y )
|
||||||
{
|
{
|
||||||
|
Loading…
x
Reference in New Issue
Block a user