fix redundant macro in hl_device_functions.cuh

paddle/cuda/include/hl_device_functions.cuh

@@ -16,8 +16,6 @@ limitations under the License. */
 #ifndef HL_DEVICE_FUNCTIONS_CUH_
 #define HL_DEVICE_FUNCTIONS_CUH_
 
-#if defined(__CUDA_ARCH__) && __CUDA_ARCH__ < 600
-
 namespace hppl {
 
 static __inline__ __device__ double atomicAdd(double* address, double val) {
@@ -42,6 +40,4 @@ static __inline__ __device__ double atomicAdd(double* address, double val) {
 using hppl::atomicAdd;
 #endif
 
-#endif
-
 #endif /* HL_DEVICE_FUNCTIONS_CUH_ */