fixes in _partition_param_sec function (#5613)

There are few fixes:
- When param.ds_secondary_tensor is not None and the param has not been
updated we don't need to update the param.ds_secondary_tensor.
- In HPU the 2nd tensor partition will always be completed before the
all-gather, so we don't need to add synchronize().

deepspeed/runtime/zero/partition_parameters.py

@@ -1664,6 +1664,8 @@ def _partition_param_sec(self, param, buffer=None, has_been_updated=False):
         ##support for NVME secondary param offload
         #print_rank_0(f"SEC Param id {param.ds_id} status is {param.ds_status}", force=True)
         if param.ds_status is ZeroParamStatus.AVAILABLE:
+            if param.ds_secondary_tensor is not None and not has_been_updated:  ##param already partitioned
+                return
             #check padding
             tensor_size = self._aligned_size(param)
             partition_size = tensor_size // self.dp_world_size
@@ -1702,7 +1704,8 @@ def _partition_param_sec(self, param, buffer=None, has_been_updated=False):
                                              sec_numel).copy_(one_dim_param.narrow(0, secondary_start, sec_numel))
 
             # TODO: This is a temporary fix to avoid the issue that 2nd tensor all-gather happens before 2nd tensor partition is done
-            get_accelerator().current_stream().synchronize()
+            if not get_accelerator().resolves_data_dependency():
+                get_accelerator().current_stream().synchronize()
 
             print_rank_0(f"{param.ds_id} partitioned type {param.dtype} dev {param.device} shape {param.shape}",
                          force=False)