Erlang logo
User's Guide

Efficiency Guide
User's Guide
Version 7.1

Expand All
Contract All


6 Functions

6.1  Pattern Matching

Pattern matching in function head as well as in case and receive clauses are optimized by the compiler. With a few exceptions, there is nothing to gain by rearranging clauses.

One exception is pattern matching of binaries. The compiler does not rearrange clauses that match binaries. Placing the clause that matches against the empty binary last is usually slightly faster than placing it first.

The following is a rather unnatural example to show another exception:


atom_map1(one) -> 1;
atom_map1(two) -> 2;
atom_map1(three) -> 3;
atom_map1(Int) when is_integer(Int) -> Int;
atom_map1(four) -> 4;
atom_map1(five) -> 5;
atom_map1(six) -> 6.

The problem is the clause with the variable Int. As a variable can match anything, including the atoms four, five, and six, which the following clauses also match, the compiler must generate suboptimal code that executes as follows:

  • First, the input value is compared to one, two, and three (using a single instruction that does a binary search; thus, quite efficient even if there are many values) to select which one of the first three clauses to execute (if any).
  • >If none of the first three clauses match, the fourth clause match as a variable always matches.
  • If the guard test is_integer(Int) succeeds, the fourth clause is executed.
  • If the guard test fails, the input value is compared to four, five, and six, and the appropriate clause is selected. (There is a function_clause exception if none of the values matched.)

Rewriting to either:


atom_map2(one) -> 1;
atom_map2(two) -> 2;
atom_map2(three) -> 3;
atom_map2(four) -> 4;
atom_map2(five) -> 5;
atom_map2(six) -> 6;
atom_map2(Int) when is_integer(Int) -> Int.



atom_map3(Int) when is_integer(Int) -> Int;
atom_map3(one) -> 1;
atom_map3(two) -> 2;
atom_map3(three) -> 3;
atom_map3(four) -> 4;
atom_map3(five) -> 5;
atom_map3(six) -> 6.

gives slightly more efficient matching code.

Another example:


map_pairs1(_Map, [], Ys) ->
map_pairs1(_Map, Xs, [] ) ->
map_pairs1(Map, [X|Xs], [Y|Ys]) ->
    [Map(X, Y)|map_pairs1(Map, Xs, Ys)].

The first argument is not a problem. It is variable, but it is a variable in all clauses. The problem is the variable in the second argument, Xs, in the middle clause. Because the variable can match anything, the compiler is not allowed to rearrange the clauses, but must generate code that matches them in the order written.

If the function is rewritten as follows, the compiler is free to rearrange the clauses:


map_pairs2(_Map, [], Ys) ->
map_pairs2(_Map, [_|_]=Xs, [] ) ->
map_pairs2(Map, [X|Xs], [Y|Ys]) ->
    [Map(X, Y)|map_pairs2(Map, Xs, Ys)].

The compiler will generate code similar to this:

DO NOT (already done by the compiler)

explicit_map_pairs(Map, Xs0, Ys0) ->
    case Xs0 of
	[X|Xs] ->
	    case Ys0 of
		[Y|Ys] ->
		    [Map(X, Y)|explicit_map_pairs(Map, Xs, Ys)];
		[] ->
	[] ->

This is slightly faster for probably the most common case that the input lists are not empty or very short. (Another advantage is that Dialyzer can deduce a better type for the Xs variable.)

6.2  Function Calls

This is an intentionally rough guide to the relative costs of different calls. It is based on benchmark figures run on Solaris/Sparc:

  • Calls to local or external functions (foo(), m:foo()) are the fastest calls.
  • Calling or applying a fun (Fun(), apply(Fun, [])) is about three times as expensive as calling a local function.
  • Applying an exported function (Mod:Name(), apply(Mod, Name, [])) is about twice as expensive as calling a fun or about six times as expensive as calling a local function.

Notes and Implementation Details

Calling and applying a fun does not involve any hash-table lookup. A fun contains an (indirect) pointer to the function that implements the fun.


Tuples are not fun(s). A "tuple fun", {Module,Function}, is not a fun. The cost for calling a "tuple fun" is similar to that of apply/3 or worse. Using "tuple funs" is strongly discouraged, as they might not be supported in a future Erlang/OTP release, and because there exists a superior alternative from R10B, namely the fun Module:Function/Arity syntax.

apply/3 must look up the code for the function to execute in a hash table. It is therefore always slower than a direct call or a fun call.

It no longer matters (from a performance point of view) whether you write:

Module:Function(Arg1, Arg2)


apply(Module, Function, [Arg1,Arg2])

The compiler internally rewrites the latter code into the former.

The following code is slightly slower because the shape of the list of arguments is unknown at compile time.

apply(Module, Function, Arguments)

6.3  Memory Usage in Recursion

When writing recursive functions, it is preferable to make them tail-recursive so that they can execute in constant memory space:


list_length(List) ->
    list_length(List, 0).

list_length([], AccLen) -> 
    AccLen; % Base case

list_length([_|Tail], AccLen) ->
    list_length(Tail, AccLen + 1). % Tail-recursive


list_length([]) ->
    0. % Base case
list_length([_ | Tail]) ->
    list_length(Tail) + 1. % Not tail-recursive