Optimization Guide

Performance Optimizations

A “closed loop” is any streamly code that generates a stream using unfold (or conceptually any stream generation combinator) and ends up eliminating it with a fold (conceptually any stream elimination combinator). It is essentially a loop processing multiple elements in a stream sequence, just like a for or while loop in imperative programming.

Closed loops are generated in a modular fashion by stream generation, transformation and elimination combinators in streamly. Combinators transfer data to the next stream pipeline stage using data constructors. These data constructors are eliminated by the compiler using stream fusion optimizations, generating a very efficient loop.

However, stream fusion optimization depends on proper inlining of the combinators involved. The fusion-plugin package mentioned earlier fills gaps for several optimizations that GHC does not perform automatically. It automatically inlines the internal definitions that involve the constructors we want to eliminate. In some cases fusion-plugin may not help and programmer may have to annotate the code manually for complete fusion. In this section we mention some of the cases where programmer annotation may help in stream fusion.

Remember, you need to worry about performance only where it matters, try to optimize the fast path and not everything blindly.

INLINE annotations

It may help to add INLINE annotations on any intermediate functions involved in a closed loop. In some cases you may have to add an inline phase as well as described below.

Usually GHC has three inline phases - the first phase is pahse-2, the second phase is phase-1 and the last one is phase-0.

Early INLINE

Generally, you only have to inline the combinators or functions participating in a loop and not the whole loop itself. But sometimes you may want to inline the whole loop itself inside a larger function. In most cases you can just add an INLINE pragma on the function containing the loop. But you may need some special considerations in some (not common) cases.

In some cases you may have to use INLINE[2] instead of INLINE which means inline the function early in phase-2. This may sometimes be needed on the because the performance of several combinators in streamly depends on getting inlined in phase-2 and if you use a plain INLINE annotation GHC may decide to delay the inlining in some cases. This is not very common but may be needed sometimes. Perhaps GHC can be fixed or we can resolve this using fusion-plugin in future.

Delayed INLINE

When a function is passed to a higher order function e.g. a function passed to concatMap or unfoldMany then we want the function to be inlined after the higher order is inlined so that proper fusion of the higher order function can occur. For such cases we usually add INLINE[1] on the function being passed to instruct GHC not to inline it too early.

Strictness annotations

  • Strictness annotations on data, specially the data used as accumulator in folds and scans, can help in improving performance.
  • Strictness annotations on function arguments can help the compiler unbox constructors in certain cases, improving performance.
  • Sometimes using -XStrict extension can help improve performance, if so you may be missing some strictness annotations. -XStrict can be used as an aid to detect missing annotations, using it blindly may regress performance.

Use tail recursion

Do not use a strict foldr or lazy foldl unless you know what you are doing. Use lazy foldr for lazily transforming the stream and strict foldl for reducing the stream. If you are manually writing recursive code, try to use tail recursion where possible.

Build times and space considerations

Haskell, being a pure functional language, confers special powers on GHC. It allows GHC to do whole program optimization. In a closed loop all the components of the loop are inlined and GHC fuses them together, performs many optimizing transformations and churns out an optimized fused loop code. Let’s call it whole-loop-optimization.

To be able to fuse the loop by whole-loop-optimization all the parts of the loop must be operated on by GHC at the same time to fuse them together. The amount of time and memory required to do so depends on the size of the loop. Huge loops can take a lot of time and memory. We have seen GHC take 4-5 GB of memory when a lot of combinators are used in a single module.

If a module takes too much time and space we can break it into multiple modules moving some non-inlined parts in another module. There is another advantage of breaking large modules, it can take advantage of parallel compilation if they do not depend on each other.