From 2e704d4257f235dd1f3224cd590e4fca4e3aaf96 Mon Sep 17 00:00:00 2001
From: Mehdi Goli <mehdi.goli@codeplay.com>
Date: Thu, 10 Nov 2016 18:45:12 +0000
Subject: Adding Memset; optimising MecopyDeviceToHost by removing double
 copying;

---
 unsupported/test/cxx11_tensor_device_sycl.cpp | 13 +++++++++++++
 1 file changed, 13 insertions(+)

(limited to 'unsupported/test/cxx11_tensor_device_sycl.cpp')

diff --git a/unsupported/test/cxx11_tensor_device_sycl.cpp b/unsupported/test/cxx11_tensor_device_sycl.cpp
index 7f79753c5..820bc88d0 100644
--- a/unsupported/test/cxx11_tensor_device_sycl.cpp
+++ b/unsupported/test/cxx11_tensor_device_sycl.cpp
@@ -19,10 +19,23 @@
 
 #include "main.h"
 #include <unsupported/Eigen/CXX11/Tensor>
+#include<stdint.h>
 
 void test_device_sycl(const Eigen::SyclDevice &sycl_device) {
   std::cout <<"Helo from ComputeCpp: the requested device exists and the device name is : "
     << sycl_device.m_queue.get_device(). template get_info<cl::sycl::info::device::name>() <<std::endl;;
+  int sizeDim1 = 100;
+
+  array<int, 1> tensorRange = {{sizeDim1}};
+  Tensor<int, 1> in(tensorRange);
+  Tensor<int, 1> in1(tensorRange);
+  memset(in1.data(), 1,in1.dimensions().TotalSize()*sizeof(int));
+  int * gpu_in_data  = static_cast<int*>(sycl_device.allocate(in.dimensions().TotalSize()*sizeof(int)));
+  sycl_device.memset(gpu_in_data, 1,in.dimensions().TotalSize()*sizeof(int) );
+  sycl_device.memcpyDeviceToHost(in.data(), gpu_in_data, in.dimensions().TotalSize()*sizeof(int) );
+  for (int i=0; i<in.dimensions().TotalSize(); i++)
+    VERIFY_IS_APPROX(in(i), in1(i));
+  sycl_device.deallocate(gpu_in_data);
 }
 void test_cxx11_tensor_device_sycl() {
   cl::sycl::gpu_selector s;
-- 
cgit v1.2.3