Fix ACE when # rows is not divisible by CPU count.
Created by: QuLogic
The division rounds down, so the last thread does not include the last few rows. For example, 3296 rows (from test_ace
) divided by 10 CPUs is 329.6 rows each, rounded down to 329, *10 CPU = 3290, and 6 rows are left out.
Fixes #2 (closed) (at least as originally reported).