[UR] Add a UR device info enum for checking native bfloat16 conversions (#17372)

Signed-off-by: jinge90 <ge.jin@intel.com>

Author: jinge90

unified-runtime/include/ur_api.h

@@ -460,6 +460,8 @@ etors:
       desc: "[int32_t][optional-query] return min power limit in milliwatts."
     - name: MAX_POWER_LIMIT
       desc: "[int32_t][optional-query] return max power limit in milliwatts."
+    - name: BFLOAT16_CONVERSIONS_NATIVE
+      desc: "[$x_bool_t] support for native bfloat16 conversions"
 --- #--------------------------------------------------------------------------
 type: function
 desc: "Retrieves various information about device"

unified-runtime/include/ur_print.hpp

@@ -254,6 +254,8 @@ UR_APIEXPORT ur_result_t UR_APICALL urDeviceGetInfo(ur_device_handle_t hDevice,
         UR_MEMORY_SCOPE_CAPABILITY_FLAG_WORK_GROUP;
     return ReturnValue(Capabilities);
   }
+  case UR_DEVICE_INFO_BFLOAT16_CONVERSIONS_NATIVE:
+    return ReturnValue(false);
   case UR_DEVICE_INFO_SUB_GROUP_SIZES_INTEL: {
     // NVIDIA devices only support one sub-group size (the warp size)
     int WarpSize = 0;

unified-runtime/scripts/core/device.yml

@@ -1023,6 +1023,8 @@ UR_APIEXPORT ur_result_t UR_APICALL urDeviceGetInfo(ur_device_handle_t hDevice,
 #else
     return ReturnValue(ur_bool_t{false});
 #endif
+  case UR_DEVICE_INFO_BFLOAT16_CONVERSIONS_NATIVE:
+    return ReturnValue(false);
   case UR_DEVICE_INFO_ASYNC_BARRIER:
     return ReturnValue(false);
   case UR_DEVICE_INFO_IL_VERSION:

unified-runtime/source/adapters/cuda/device.cpp

@@ -1038,6 +1038,14 @@ ur_result_t urDeviceGetInfo(
       return ze2urResult(errc);
     return ReturnValue(UrRootDev);
   }
+  case UR_DEVICE_INFO_BFLOAT16_CONVERSIONS_NATIVE: {
+    bool Bfloat16ConversionSupport =
+        (Device->Platform->zeDriverExtensionMap.count(
+            ZE_BFLOAT16_CONVERSIONS_EXT_NAME)) ||
+        ((Device->ZeDeviceProperties->deviceId & 0xfff) == 0x201 ||
+         (Device->ZeDeviceProperties->deviceId & 0xff0) == 0xbd0);
+    return ReturnValue(Bfloat16ConversionSupport);
+  }
   case UR_DEVICE_INFO_COMMAND_BUFFER_SUPPORT_EXP:
     return ReturnValue(true);
   case UR_DEVICE_INFO_COMMAND_BUFFER_UPDATE_CAPABILITIES_EXP: {

unified-runtime/source/adapters/hip/device.cpp

@@ -344,6 +344,8 @@ UR_APIEXPORT ur_result_t UR_APICALL urDeviceGetInfo(ur_device_handle_t hDevice,
     return ReturnValue(bool{0});
   case UR_DEVICE_INFO_ATOMIC_64:
     return ReturnValue(bool{1});
+  case UR_DEVICE_INFO_BFLOAT16_CONVERSIONS_NATIVE:
+    return ReturnValue(bool{0});
   case UR_DEVICE_INFO_MEM_CHANNEL_SUPPORT:
     return ReturnValue(bool{0});
   case UR_DEVICE_INFO_IMAGE_SRGB:

unified-runtime/source/adapters/level_zero/device.cpp

@@ -1514,6 +1514,13 @@ UR_APIEXPORT ur_result_t UR_APICALL urDeviceGetInfo(ur_device_handle_t hDevice,
     return ReturnValue(UR_EXP_DEVICE_2D_BLOCK_ARRAY_CAPABILITY_FLAG_LOAD |
                        UR_EXP_DEVICE_2D_BLOCK_ARRAY_CAPABILITY_FLAG_STORE);
   }
+  case UR_DEVICE_INFO_BFLOAT16_CONVERSIONS_NATIVE: {
+    bool Supported = false;
+    UR_RETURN_ON_FAILURE(cl_adapter::checkDeviceExtensions(
+        cl_adapter::cast<cl_device_id>(hDevice),
+        {"cl_intel_bfloat16_conversions"}, Supported));
+    return ReturnValue(Supported);
+  }
   case UR_DEVICE_INFO_COMMAND_BUFFER_SUPPORT_EXP: {
     cl_device_id Dev = cl_adapter::cast<cl_device_id>(hDevice);
     size_t ExtSize = 0;

unified-runtime/source/adapters/native_cpu/device.cpp

@@ -2561,6 +2561,23 @@ TEST_P(urDeviceGetInfoTest, SuccessUseNativeAssert) {
                              property_value);
 }
 
+TEST_P(urDeviceGetInfoTest, SuccessBfloat16ConversionsNative) {
+  size_t property_size = 0;
+  const ur_device_info_t property_name =
+      UR_DEVICE_INFO_BFLOAT16_CONVERSIONS_NATIVE;
+
+  ASSERT_SUCCESS_OR_OPTIONAL_QUERY(
+      urDeviceGetInfo(device, property_name, 0, nullptr, &property_size),
+      property_name);
+  ASSERT_EQ(property_size, sizeof(ur_bool_t));
+
+  uint32_t property_value = 0;
+  ASSERT_QUERY_RETURNS_VALUE(urDeviceGetInfo(device, property_name,
+                                             property_size, &property_value,
+                                             nullptr),
+                             property_value);
+}
+
 TEST_P(urDeviceGetInfoTest, SuccessThrottleReasons) {
   // TODO: enable when driver/library version mismatch is fixed in CI.
   // See https://github.com/intel/llvm/issues/17614