Mongorestore is very slow on sharded cluster timeseries collection

Hi all,
I am trying to restore the regular collection dump to sharded timeseries collection. It is in sharded cluster.
mongorestore is very slow and it is taking around 45 minutes to restore 20 million documents in sharded time series collection.
I am also using --noIndexRestore --maintainInsertionOrder as insertion order is important to timeseries collections and indexes are already setup on time series collection before restore.

What are the ways the improve the time taken by mongorestore ?

Thanks in advance.