Add index to HPU devices (deepspeedai#7497)

deepcharm · web-flow · commit 047a7599d246 · 2025-08-19T00:30:56.000Z
The [PR deepspeedai#7266](deepspeedai#7266) enforces the devices having explicit device indices (i.e. 'hpu:0', 'cuda:0', etc). This PR aligns HPU devices to the requirement. Signed-off-by: Max Kovalenko <mkovalenko@habana.ai>
diff --git a/accelerator/hpu_accelerator.py b/accelerator/hpu_accelerator.py
@@ -54,8 +54,9 @@ def handles_memory_backpressure(self):
         return True
 
     def device_name(self, device_index=None):
-        # ignoring device_index.
-        return 'hpu'
+        if device_index is None:
+            return self._name
+        return '{}:{}'.format(self._name, device_index)
 
     def device(self, device_index=None):
         return torch.device(self.device_name(device_index))
