Cursor Vs. Select

buddies,
situation: a processing must take place on every row of a table, and
output results to another table, that can't be done via an insert
into..select query (let's assume that it's not possible for now).
There're 2 solutions I have in mind:
1) open a cursor and cycle through each row (The table can have up to
1M rows)
2) create a clustered index (i.e on an identity column) then have a
loop like:
declare @i int, @rows int,
@col1 varchar(20), @col2 varchar(20),... @coln varchar(20),
@outval1 varchar(20),... -- output values
select @i=1, @rows = max(xid) from tblname -- xid is clustered indexed
while (@i<=@rows)
begin
select @col1 = col1, @col2 = col2,...@coln = coln
from tblname
where xid = i
-- do the processing on the variables
-- then insert results to another table
set @i = @i+1
end
I'd like to know your ideas of which one would be more efficient. Any
other solutions are much appreciated
thanks,
Tamy