Noam Ross recently shared a very useful guide to speeding up your R code.
- Get a bigger computer (for example, renting an instance on the Amazon cloud for a few cents an hour)
- Use parallel programming techniques
- Using the R byte-compiler
- Profiling and benchmarking your code
- Using high-performance packages (like xts, for time series)
- And lastly, rewriting your code to use more efficient constructs
One other tip that can have some great performance benefits is linking R to parallel BLAS libraries (Revolution R does this by default). For more details on how to speed up your R code read Noam's excellent guide, linked below.