Question

I’m getting into parallel programming and I’m studying mapreduce and other distributed algorithms. Is it best just to learn mapreduce or is there a more general algorithm that will serve me better?

Answer 1

It depends what you intend to use the algorithm(s) for.

MapReduce is a generalised and very useful programming model. (Google bases many of it’s internal indexing processes on it). Learning it certainly won’t do you any harm.

MapReduce is a programming model and
an associated implementation for
processing and generating large data
sets. Users specify a map function
that processes a key/value pair to
generate a set of intermediate
key/value pairs, and a reduce function
that merges all intermediate values
associated with the same intermediate
key. Many real world tasks are
expressible in this model, as shown in
the paper.

The most important parallel processing concept to learn is quite simple: synchronisation is what you need to minimise if you want to attain effective speedup.

Strive for:

• Large granularity of work chunks
• Keep size work chunks similiar in size
• Minimise the number of synchronisation steps