What is the most efficient way to process a large query with a map/reduce script? I have a query that returns 100,000-150,000 rows. That consolidates down to 10,000-12,000 rows. Then some calculations run which result in usually less than 100 updates that need to happen.

The piece I would like to improve is processing the initial 100-150,000 rows. Here are the methods I've tried...

Try 1: Loop through the results in the input phase and create an array.
Result: I didn't keep this code but it hit a script execution limit looping through the results. Not really surprising.

Try 2: Just return a search in the input phase so the map phase and process the results.