From 6cbeeb8e59cb3107cb46cf8fc6f467de97f7a857 Mon Sep 17 00:00:00 2001
From: Aditya Atluri <Aditya.Atluri@amd.com>
Date: Tue, 7 Mar 2017 11:24:32 -0600
Subject: [PATCH] added new field to hipDeviceProp_t structure gcnArch.

1. It is an integer containing gfx values 701, 801, 802, 803
2. On NV path, it is zero

Change-Id: I2b4c7f48981d0214d8c6b1905d2cc85b16203419


[ROCm/hip commit: f86f3b3b33ce43035ae95ad201e7409e3416259e]
---
 projects/hip/include/hip/hip_runtime_api.h    |  1 +
 .../samples/0_Intro/square/square.hipref.cpp  | 10 +++++----
 projects/hip/src/hip_hcc.cpp                  | 21 +++++++++++++++----
 3 files changed, 24 insertions(+), 8 deletions(-)

diff --git a/projects/hip/include/hip/hip_runtime_api.h b/projects/hip/include/hip/hip_runtime_api.h
index 28d67fc01a..818c0b7c34 100644
--- a/projects/hip/include/hip/hip_runtime_api.h
+++ b/projects/hip/include/hip/hip_runtime_api.h
@@ -106,6 +106,7 @@ typedef struct hipDeviceProp_t {
     size_t maxSharedMemoryPerMultiProcessor;    ///< Maximum Shared Memory Per Multiprocessor.
     int isMultiGpuBoard;                        ///< 1 if device is on a multi-GPU board, 0 if not.
     int canMapHostMemory;                       ///< Check whether HIP can map host memory
+    int gcnArch;                                ///< AMD GCN Arch Value. Eg: 803, 701
  } hipDeviceProp_t;
 
 
diff --git a/projects/hip/samples/0_Intro/square/square.hipref.cpp b/projects/hip/samples/0_Intro/square/square.hipref.cpp
index 0073c1399a..e694bfb8a4 100644
--- a/projects/hip/samples/0_Intro/square/square.hipref.cpp
+++ b/projects/hip/samples/0_Intro/square/square.hipref.cpp
@@ -32,7 +32,7 @@ THE SOFTWARE.
 	}\
 }
 
-/* 
+/*
  * Square each element in the array A and write to array C.
  */
 template <typename T>
@@ -58,16 +58,18 @@ int main(int argc, char *argv[])
     hipDeviceProp_t props;
     CHECK(hipGetDeviceProperties(&props, 0/*deviceID*/));
     printf ("info: running on device %s\n", props.name);
-
+    #ifdef __HIP_PLATFORM_HCC__
+      printf ("info: architecture on AMD GPU device is: %d\n",props.gcnArch);
+    #endif
     printf ("info: allocate host mem (%6.2f MB)\n", 2*Nbytes/1024.0/1024.0);
     A_h = (float*)malloc(Nbytes);
     CHECK(A_h == 0 ? hipErrorMemoryAllocation : hipSuccess );
     C_h = (float*)malloc(Nbytes);
     CHECK(C_h == 0 ? hipErrorMemoryAllocation : hipSuccess );
 	// Fill with Phi + i
-    for (size_t i=0; i<N; i++) 
+    for (size_t i=0; i<N; i++)
     {
-        A_h[i] = 1.618f + i; 
+        A_h[i] = 1.618f + i;
     }
 
     printf ("info: allocate device mem (%6.2f MB)\n", 2*Nbytes/1024.0/1024.0);
diff --git a/projects/hip/src/hip_hcc.cpp b/projects/hip/src/hip_hcc.cpp
index f0a94ea9ca..fec5bb6a8a 100644
--- a/projects/hip/src/hip_hcc.cpp
+++ b/projects/hip/src/hip_hcc.cpp
@@ -758,11 +758,24 @@ hipError_t ihipDevice_t::initProperties(hipDeviceProp_t* prop)
     prop->isMultiGpuBoard = 0 ? gpuAgentsCount < 2 : 1;
 
     // Get agent name
-#if HIP_USE_PRODUCT_NAME
+
     err = hsa_agent_get_info(_hsaAgent, (hsa_agent_info_t)HSA_AMD_AGENT_INFO_PRODUCT_NAME, &(prop->name));
-#else
-    err = hsa_agent_get_info(_hsaAgent, HSA_AGENT_INFO_NAME, &(prop->name));
-#endif
+    char archName[256];
+    err = hsa_agent_get_info(_hsaAgent, HSA_AGENT_INFO_NAME, &archName);
+    
+    if(strcmp(archName,"gfx701")==0){
+      prop->gcnArch = 701;
+    }
+    if(strcmp(archName,"gfx801")==0){
+      prop->gcnArch = 801;
+    }
+    if(strcmp(archName,"gfx802")==0){
+      prop->gcnArch = 802;
+    }
+    if(strcmp(archName,"gfx803")==0){
+      prop->gcnArch = 803;
+    }
+
     DeviceErrorCheck(err);
 
     // Get agent node