Add an upper bound to the channel a that connects par_granges to the processors. This prevents running up a huge backlog of positions to print in the even to very low coverage areas for base-depth.
On low coverage datasets, more space per thread is needed. On high coverage, less since more time is spent on calculations allowing printing to keep up.
The selected defaults were optimized for 100X data to be very near the same time for an unbounded channel.
Add an upper bound to the channel a that connects par_granges to the processors. This prevents running up a huge backlog of positions to print in the even to very low coverage areas for base-depth.
On low coverage datasets, more space per thread is needed. On high coverage, less since more time is spent on calculations allowing printing to keep up.
The selected defaults were optimized for 100X data to be very near the same time for an unbounded channel.