Device property memoryBusWidth implementation.

+ Device property memoryBusWidth is added to hipDeviceProp_t struct.
+ Device attribute hipDeviceAttributeMemoryBusWidth is added to hipDeviceAttribute_t struct.
+ Tests update.

include/hip_runtime_api.h

@@ -81,7 +81,8 @@ typedef struct hipDeviceProp_t {
     int maxThreadsDim[3];                       ///< Max number of threads in each dimension (XYZ) of a block.
     int maxGridSize[3];                         ///< Max grid dimensions (XYZ).
     int clockRate;                              ///< Max clock frequency of the multiProcessors in khz.
-    int memoryClockRate;                        ///< Max memory clock frequency in khz.
+    int memoryClockRate;                        ///< Max global memory clock frequency in khz.
+    int memoryBusWidth;                         ///< Global memory bus width in bits.
     size_t totalConstMem;                       ///< Size of shared memory region (in bytes).
     int major;                                  ///< Major compute capability.  On HCC, this is an approximation and features may differ from CUDA CC.  See the arch feature flags for portable ways to query feature caps.
     int minor;                                  ///< Minor compute capability.  On HCC, this is an approximation and features may differ from CUDA CC.  See the arch feature flags for portable ways to query feature caps.
@@ -145,6 +146,7 @@ typedef enum hipDeviceAttribute_t {
     hipDeviceAttributeMaxRegistersPerBlock,                 ///< Maximum number of 32-bit registers available to a thread block. This number is shared by all thread blocks simultaneously resident on a multiprocessor.
     hipDeviceAttributeClockRate,                            ///< Peak clock frequency in kilohertz.
     hipDeviceAttributeMemoryClockRate,                      ///< Peak memory clock frequency in kilohertz.
+    hipDeviceAttributeMemoryBusWidth,                       ///< Global memory bus width in bits.
     hipDeviceAttributeMultiprocessorCount,                  ///< Number of multiprocessors on the device.
     hipDeviceAttributeComputeMode,                          ///< Compute mode that device is currently in.
     hipDeviceAttributeL2CacheSize,                          ///< Size of L2 cache in bytes. 0 if the device doesn't have L2 cache.

include/nvcc_detail/hip_runtime_api.h

@@ -243,7 +243,9 @@ inline static hipError_t hipDeviceGetAttribute(int* pi, hipDeviceAttribute_t att
     case hipDeviceAttributeClockRate:
         cdattr = cudaDevAttrClockRate; break;
     case hipDeviceAttributeMemoryClockRate:
-        cdattr = cudaDevAttrMemoryClockRate:; break;
+        cdattr = cudaDevAttrMemoryClockRate; break;
+    case hipDeviceAttributeMemoryBusWidth:
+        cdattr = cudaDevAttrGlobalMemoryBusWidth; break;
     case hipDeviceAttributeMultiprocessorCount:
         cdattr = cudaDevAttrMultiProcessorCount; break;
     case hipDeviceAttributeComputeMode:

samples/1_Utils/hipInfo/hipInfo.cpp

@@ -80,6 +80,7 @@ void printDeviceProp (int deviceId)
     cout << setw(w1) << "pciDeviceID: " << props.pciDeviceID << endl;
     cout << setw(w1) << "multiProcessorCount: " << props.multiProcessorCount << endl;
     cout << setw(w1) << "maxThreadsPerMultiProcessor: " << props.maxThreadsPerMultiProcessor << endl;
+    cout << setw(w1) << "memoryBusWidth: " << props.memoryBusWidth << endl;
     cout << setw(w1) << "clockRate: " << (float)props.clockRate / 1000.0 << " Mhz" << endl;
     cout << setw(w1) << "memoryClockRate: " << (float)props.memoryClockRate / 1000.0 << " Mhz" << endl;
     cout << setw(w1) << "clockInstructionRate: " << (float)props.clockInstructionRate / 1000.0 << " Mhz" << endl;

src/hip_hcc.cpp

@@ -300,7 +300,7 @@ hipError_t ihipDevice_t::getProperties(hipDeviceProp_t* prop)
 
 
     // Get Max clock frequency
-    err = hsa_agent_get_info(_hsa_agent, (hsa_agent_info_t)HSA_AMD_AGENT_INFO_MAX_CLOCK_FREQUENCY,&prop->clockRate);
+    err = hsa_agent_get_info(_hsa_agent, (hsa_agent_info_t)HSA_AMD_AGENT_INFO_MAX_CLOCK_FREQUENCY, &prop->clockRate);
     prop->clockRate *= 1000.0;   // convert Mhz to Khz.
     DeviceErrorCheck(err);
 
@@ -356,12 +356,11 @@ hipError_t ihipDevice_t::getProperties(hipDeviceProp_t* prop)
 */
 
     // Get memory properties
-
-    err = hsa_agent_iterate_regions(_hsa_agent,get_region_info,prop);
+    err = hsa_agent_iterate_regions(_hsa_agent, get_region_info, prop);
     DeviceErrorCheck(err);
 
     // Get the size of the region we are using for Accelerator Memory allocations:
-    hsa_region_t *am_region = static_cast<hsa_region_t*> (_acc.get_hsa_am_region());
+    hsa_region_t *am_region = static_cast<hsa_region_t*>(_acc.get_hsa_am_region());
     err = hsa_region_get_info(*am_region, HSA_REGION_INFO_SIZE, &prop->totalGlobalMem);
     DeviceErrorCheck(err);
     // maxSharedMemoryPerMultiProcessor should be as the same as group memory size.
@@ -370,7 +369,11 @@ hipError_t ihipDevice_t::getProperties(hipDeviceProp_t* prop)
 
     // Get Max memory clock frequency
     err = hsa_region_get_info(*am_region, (hsa_region_info_t)HSA_AMD_REGION_INFO_MAX_CLOCK_FREQUENCY, &prop->memoryClockRate);
+    DeviceErrorCheck(err);
     prop->memoryClockRate *= 1000.0;   // convert Mhz to Khz.
+
+    // Get global memory bus width in bits
+    err = hsa_region_get_info(*am_region, (hsa_region_info_t)HSA_AMD_REGION_INFO_BUS_WIDTH, &prop->memoryBusWidth);
     DeviceErrorCheck(err);
 
     // Set feature flags - these are all mandatory for HIP on HCC path:
@@ -845,6 +848,8 @@ hipError_t hipDeviceGetAttribute(int* pi, hipDeviceAttribute_t attr, int device)
             *pi = prop->clockRate; break;
         case hipDeviceAttributeMemoryClockRate:
             *pi = prop->memoryClockRate; break;
+        case hipDeviceAttributeMemoryBusWidth:
+            *pi = prop->memoryBusWidth; break;
         case hipDeviceAttributeMultiprocessorCount:
             *pi = prop->multiProcessorCount; break;
         case hipDeviceAttributeComputeMode:

tests/src/hipGetDeviceAttribute.cpp

@@ -68,6 +68,7 @@ int main(int argc, char *argv[])
     CHECK(test_hipDeviceGetAttribute(deviceId, hipDeviceAttributeMaxRegistersPerBlock, props.regsPerBlock));
     CHECK(test_hipDeviceGetAttribute(deviceId, hipDeviceAttributeClockRate, props.clockRate));
     CHECK(test_hipDeviceGetAttribute(deviceId, hipDeviceAttributeMemoryClockRate, props.memoryClockRate));
+    CHECK(test_hipDeviceGetAttribute(deviceId, hipDeviceAttributeMemoryBusWidth, props.memoryBusWidth));
     CHECK(test_hipDeviceGetAttribute(deviceId, hipDeviceAttributeMultiprocessorCount, props.multiProcessorCount));
     CHECK(test_hipDeviceGetAttribute(deviceId, hipDeviceAttributeComputeMode, props.computeMode));
     CHECK(test_hipDeviceGetAttribute(deviceId, hipDeviceAttributeL2CacheSize, props.l2CacheSize));