What is the best/fastest method to create a table (Oracle 11gR2 dB)?

Question

Assuming there are no statistics for the source tables - tables are every day dropped and recreated
My tables have a few million rows
I try to create a table populated with data as fast as it is possible via
1. createtable_name as SELECT * from emp;
2. create table_name
Parallel degree 4
  as SELECT * from emp;
I 1 case I got timing about 34 Sec
In case 2 the table was created in 15 Sec
Is in oracle other possibilities to create a table much faster that with this CTAS?
Or it will be faster when I will create a table via create table_name
  column1datatype[NULL | NOT NULL],
  column2datatype[NULL | NOT NULL],
  column_ndatatype[NULL | NOT NULL]
or maybe should I use this one method
INSERT /*+ APPEND */ INTO empSELECT * FROM all_objects;
INSERT /*+ APPEND_VALUES */ INTO emp SELECT * FROM all_objects;
or it is better to simple create a table
and than use FORALL BULK INTO COLLECTION statement
whit combination of
INSERT /*+ APPEND */ INTO emp;

Answer 1

Assuming there are no statistics for the source tables - tables are every day dropped and recreated
My tables have a few million rows
The 'fastest' way is to NOT drop and recreate the tables every day.  Unless the table changes structure every day just create the tables ONE TIME. The each day you can truncate them with the REUSE STORAGE clause.
Then populate the tables using DIRECT-PATH loading such as by using the APPEND hint. That can also be done in PARALLEL if desired.
FORALL and BULK COLLECTION would ONLY be used if the data is ALREADY in collections as part of complex ETL data cleansing/conversion. If the data already exists in external files or other tables there is generally no need to use PL/SQL or collections.
What PROBLEM are you trying to solve?
You need to give us much more info if you really want help. For ETL the task of 'loading a table fast' is generally WAY DOWN the list of concerns.
1. where is the data now?
2. How much data is there?
3. what cleansing/conversion needs to be done on the data?
4. what are the requirements for restart/recovery needed?
5. what are the requirements for detecting and reporting on data issues?