Fixing Warning shfl

src/util/cuda/moderngpu/intrinsics.hxx

@@ -167,7 +167,7 @@ MGPU_DEVICE type_t shfl_down(type_t x, int offset, int width = warp_size) {
   u.t = x;
 
   iterate<num_words>([&](int i) {
-    u.x[i] = __shfl_down(u.x[i], offset, width);
+    u.x[i] = __shfl_down_sync(0xFFFFFFFF,u.x[i], offset, width);
   });
   return u.t;
 }
@@ -176,7 +176,7 @@ template<typename type_t, typename op_t>
 MGPU_DEVICE type_t shfl_up_op(type_t x, int offset, op_t op, 
   int width = warp_size) {
 
-  type_t y = shfl_up(x, offset, width);
+  type_t y = shfl_up_sync(0xFFFFFFFF,x, offset, width);
   int lane = (width - 1) & threadIdx.x;
   if(lane >= offset) x = op(x, y);
   return x;
@@ -186,7 +186,7 @@ template<typename type_t, typename op_t>
 MGPU_DEVICE type_t shfl_down_op(type_t x, int offset, op_t op, 
   int width = warp_size) {
 
-  type_t y = shfl_down(x, offset, width);
+  type_t y = shfl_down_sync(0xFFFFFFFF,x, offset, width);
   int lane = (width - 1) & threadIdx.x;
   if(lane < width - offset) x = op(x, y);
   return x;
@@ -200,7 +200,7 @@ MGPU_DEVICE inline c_type shfl_##dir##_op(c_type x, int offset, \
   asm( \
     "{.reg ."#ptx_type" r0;" \
     ".reg .pred p;" \
-    "shfl."#dir".b32 r0|p, %1, %2, %3;" \
+    "shfl.sync."#dir".b32 r0|p, %1, %2, %3,0xFFFFFFFF;" \
     "@p "#ptx_op"."#ptx_type" r0, r0, %4;" \
     "mov."#ptx_type" %0, r0; }" \
     : "="#r(result) : #r(x), "r"(offset), "r"(mask), #r(x)); \

src/util/object_si_d.hpp

@@ -41,16 +41,15 @@ struct object_si_d_e
 	 * \param dst destination object
 	 *
 	 */
-	object_si_d_e(const v_src & src, v_dst & dst)
+	__device__ __host__ object_si_d_e(const v_src & src, v_dst & dst)
 	:src(src),dst(dst)
 	{
 	};
 
 	//! It call the functor for each member
     template<typename T>
-    void operator()(T& t)
+    __device__ __host__ void operator()(T& t)
     {
-//    	typedef typename boost::mpl::at<typename v_dst::type,typename boost::mpl::int_<T::value>>::type dtype;
     	typedef decltype(src.template get<boost::mpl::at<v_prp,boost::mpl::int_<T::value>>::type::value>()) stype;
 
     	// In case of layout switch use this
@@ -167,13 +166,13 @@ struct object_si_d<v_src,v_dst,OBJ_NORMAL,prp...>
 template<typename v_src, typename v_dst, int... prp>
 struct object_si_d<v_src,v_dst,OBJ_ENCAP,prp...>
 {
-	inline object_si_d(const v_src && vs, v_dst && vd)
+	__device__ __host__ inline object_si_d(const v_src && vs, v_dst && vd)
 	{
 		object_si_d_e<v_src,v_dst,prp...> obj(vs,vd);
 		boost::mpl::for_each_ref< boost::mpl::range_c<int,0,v_dst::max_prop> >(obj);
 	}
 
-	inline object_si_d(const v_src & vs, v_dst & vd)
+	__device__ __host__ inline object_si_d(const v_src & vs, v_dst & vd)
 	{
 		object_si_d_e<v_src,v_dst,prp...> obj(vs,vd);
 		boost::mpl::for_each_ref< boost::mpl::range_c<int,0,v_dst::max_prop> >(obj);
@@ -183,11 +182,11 @@ struct object_si_d<v_src,v_dst,OBJ_ENCAP,prp...>
 template<typename v_src, typename v_dst>
 struct object_si_d<v_src,v_dst,OBJ_ENCAP>
 {
-	inline object_si_d(const v_src && vs, v_dst && vd)
+	__device__ __host__  inline object_si_d(const v_src && vs, v_dst && vd)
 	{
 	}
 
-	inline object_si_d(const v_src & vs, v_dst & vd)
+	__device__ __host__  inline object_si_d(const v_src & vs, v_dst & vd)
 	{
 	}
 };