Update square sample for recent HIP ease-of-use improvements

[ROCm/hip-tests commit: 421a50e830]

projects/hip-tests/samples/0_Intro/square/Makefile

@@ -1,20 +1,17 @@
 HIP_PATH?= $(wildcard /opt/rocm/hip)
 HIPCC=$(HIP_PATH)/bin/hipcc
 
-all: square.hip.out
+all: square.out
 
+# Step
+square.cpp:
+	$(HIP_PATH)/bin/hipify-perl square.cu > square.cpp
+
+square.out: square.cpp
+	$(HIPCC) $(CXXFLAGS) square.cpp -o $@
+
+# make step to make native cuda app, if on cuda machine
 square.cuda.out : square.cu
 	nvcc square.cu -o $@
-
-#hipify square.cu > square.cpp
-# Then review & finish port in square.cpp
-#
-
-square.hip.out: square.hipref.cpp
-	$(HIPCC) $(CXXFLAGS) square.hipref.cpp -o $@
-
-
-
-
 clean:
 	rm -f *.o *.out	

projects/hip-tests/samples/0_Intro/square/README.md

@@ -1,16 +1,13 @@
 # Square.md
 
 Simple test which shows how to use hipify to port CUDA code to HIP.  
-See related [blog](http://gpuopen.com/hip-to-be-squared-an-introductory-hip-tutorial) that explains the example.
+See related [blog](http://gpuopen.com/hip-to-be-squared-an-introductory-hip-tutorial) that explains the example. 
+Now it is even simpler and requires no manual modification to the hipified source code - just hipify and compile:
 
 1. Add hip/bin path to the PATH  :
     <code>export PATH=$PATH:[MYHIP]/bin</code>
 
-2. Do <code>$ hipify square.cu > square.cpp </code>
-
-3. Manually edit square.cpp to add hipLaunchParms lp to kernel parms:
-    <code>vector_square(hipLaunchParm lp, T *C_d, const T *A_d, size_t N)</code>
-
-    (see square.hipref.cpp for the correct output after running hipify and the above manual step)
-
-4. make
+2. <code>$ make </code>
+   Make runs these steps.  This can be performed on either CUDA or AMD platform:
+   <code>hipify-perl square.cu > square.cpp </code>    # convert cuda code to hip code
+   <code>hipcc square.cpp</code>                       # compile into executable

projects/hip-tests/samples/0_Intro/square/square.cu

@@ -38,7 +38,7 @@ THE SOFTWARE.
  */
 template <typename T>
 __global__ void
-vector_square(T *C_d, const T *A_d, size_t N)
+vector_square(T *C_d, T *A_d, size_t N)
 {
     size_t offset = (blockIdx.x * blockDim.x + threadIdx.x);
     size_t stride = blockDim.x * gridDim.x ;