When I run my app queries in BigQuery console, they take 3-4 seconds to return nearly a million rows. The same queries in my streamlit python app (local, community cloud, GCP) all take about 180 seconds to complete including writing the results into a dataframe. Upon instrumentation it shows most of the time is with rowiterator. Is there a different way to do this? Environments have plenty of RAM and CPU.

├─ 163.313 RowIterator._items_iter  google/api_core/page_iterator.py:206
│     [62 frames hidden]  google, requests, urllib3, http, sock...
│        88.930 _SSLSocket.read  <built-in>
├─ 8.598 QueryJob.__iter__  google/cloud/bigquery/job/query.py:2107
│     [27 frames hidden]  google, requests, urllib3, http, sock...
├─ 3.988 Row.__getitem__  google/cloud/bigquery/table.py:1501
│     [2 frames hidden]  google
├─ 3.569 Client.query  google/cloud/bigquery/client.py:3362
│     [25 frames hidden]  google, requests, urllib3, http, sock...
└─ 2.815 [self]  users.py

        bq_client = st.session_state.bq_client
        sql_query = f"""
                    SELECT *
                        FROM `dataexploration-193817.user_data.all_users_progress`
                    WHERE
                        first_open BETWEEN PARSE_DATE('%Y/%m/%d','{start_date}') AND CURRENT_DATE() 
                    """
        rows_raw = bq_client.query(sql_query)
        rows = [dict(row) for row in rows_raw]
        df_user_list = pd.DataFrame(rows)