Erlang Profiling Techniques
Utilize Erlang's built-in profiling tools to identify performance bottlenecks and optimize your application code.
Erlang Profiling Techniques is a free Erlang OTP: Distributed & Fault-Tolerant Systems Programming lesson on CoddyKit — lesson 1 of 4. You can read the complete lesson below for free — then practise it hands-on in the browser with a built-in code editor and a 24/7 AI tutor. It is part of the Erlang OTP: Distributed & Fault-Tolerant Systems Programming learning path, one of 4 lessons in the course, and your progress syncs across the web and the CoddyKit app.
Why Profile Erlang Code?
Ever wonder why your Erlang application feels a bit sluggish? That's where profiling comes in!
Profiling is like giving your code an X-ray. It helps you:
- Spot performance bottlenecks.
- Understand how functions spend their time.
- Optimize resource usage (CPU, memory).
In Erlang, profiling is crucial for building efficient, high-performance systems.
Meet `fprof`: CPU & Memory
One of Erlang's most powerful built-in profiling tools is fprof. It's designed to give you detailed insights into how your program uses system resources.
fprof can profile:
- CPU usage: Which functions are consuming the most processing time?
- Memory usage: How much memory each function allocates.
It's great for deep dives into specific parts of your code.
Profiling CPU with `fprof`
Let's see fprof in action! This example creates a CPU-intensive calculation and then uses fprof to measure where the time is spent.
Run this code and observe the output, which will be the fprof report.
-module(cpu_profiler).
-export([run/0, long_calculation/1]).
% A function designed to consume CPU time
long_calculation(N) ->
lists:foldl(fun(I, Acc) -> math:sqrt(I) + Acc end, 0.0, lists:seq(1, N)).
run() ->
io:format("~n--- Starting fprof CPU profiling ---~n"),
fprof:start([cpu]), % Start profiling for CPU
_Result = long_calculation(100000), % Call the function to profile
fprof:stop(), % Stop collecting data
fprof:profile(), % Process collected data
io:format("~n--- fprof CPU Report ---~n"),
fprof:analyse({dest, user}), % Print the analysis to the console
io:format("--- End fprof CPU Report ---~n"),
ok.Deciphering `fprof` Reports
The fprof report can look a bit intimidating at first! Here are the key columns to focus on:
acc(Accumulated): Total time spent in a function, including time in functions it calls.self(Self time): Time spent directly in this function, excluding time in functions it calls. This helps pinpoint the exact bottleneck.calls: How many times the function was called.
Look for functions with high self times to find areas for optimization.
`fprof` for Memory Usage
While we focused on CPU, fprof can also help with memory usage. By calling fprof:start([memory]), you can track memory allocation per function.
Memory profiling helps identify "memory leaks" or functions that allocate excessively large data structures, which can be critical for long-running systems.
The report structure is similar, but focuses on bytes allocated rather than CPU cycles.
`eprof`: Time-Based Profiling
Another useful tool is eprof. While fprof is very detailed, eprof provides a simpler, more high-level overview of execution times.
eprof is excellent for quickly identifying which functions take the longest to run, without the deep call-graph analysis of fprof.
It's often used for a quick check before diving into more detailed profiling.
Running an `eprof` Test
Let's use eprof to measure the execution time of a list manipulation operation. This gives a clear picture of how long the function itself takes.
Run this example and check the output for execution statistics.
-module(time_profiler).
-export([run/0, quick_operation/1]).
% A function that processes a list
quick_operation(N) ->
lists:map(fun(I) -> I * 2 end, lists:seq(1, N)).
run() ->
io:format("~n--- Starting eprof execution time profiling ---~n"),
eprof:start(), % Start eprof
_Result = quick_operation(50000), % Call the function
eprof:stop(), % Stop collecting data
io:format("~n--- eprof Report ---~n"),
eprof:log({dest, user}), % Print the log to console
eprof:stop_profiling(), % Clean up eprof
io:format("--- End eprof Report ---~n"),
ok.Interpreting `eprof` Statistics
eprof output is simpler than fprof. It typically shows you:
{: The function being reported., , } {calls, N}: How many times this function was called.{total, Time}: The total execution time for all calls to this function.{average, Time}: The average execution time per call.
This helps you quickly see which functions are cumulatively taking the most time.
Basic Tracing with `dbg`
While primarily a debugging tool, dbg can also be used for basic tracing to see function calls in real-time. It's less about performance metrics and more about understanding flow.
This example shows how to set a trace on a function and then call it. The trace output typically appears directly in the shell as the code runs.
-module(dbg_example).
-export([run/0, simple_func/1]).
simple_func(X) ->
io:format("Inside simple_func with ~p~n", [X]),
X * 2.
run() ->
io:format("~n--- Starting dbg trace for simple_func ---~n"),
dbg:tracer(), % Start the tracer process
dbg:p(all, call), % Trace all process calls (optional, but good for context)
dbg:tp(dbg_example, simple_func, []), % Set trace pattern on simple_func/1
io:format("Calling simple_func(5)...~n"),
_Result = simple_func(5),
io:format("Calling simple_func(10)...~n"),
_Result2 = simple_func(10),
dbg:stop_clear(), % Stop tracing and clear patterns
io:format("--- dbg trace finished ---~n"),
ok.Profiling Tool Check
You've learned about Erlang's powerful profiling tools. Now, let's test your understanding!
Consider a scenario where you suspect a specific function is causing high CPU load due to complex calculations within its own body, rather than in functions it calls.
Profiling Power-Up!
Great job! You've taken your first steps into Erlang profiling.
We covered:
fprof: For detailed CPU and memory profiling, usingselftime to pinpoint bottlenecks.eprof: For a high-level overview of function execution times (totalandaverage).dbg: A brief look at its use for basic function call tracing to understand program flow.
These tools are your allies in building efficient and robust Erlang applications. Keep exploring them!
Frequently asked questions
Is the “Erlang Profiling Techniques” lesson free?
Yes — the full text of “Erlang Profiling Techniques” is free to read here on the web, and the Erlang OTP: Distributed & Fault-Tolerant Systems Programming course includes 4 lessons in total. To practise it interactively (a built-in code editor and a 24/7 AI tutor) and unlock the rest of the Erlang OTP: Distributed & Fault-Tolerant Systems Programming course, upgrade to CoddyKit PRO.
What will I learn in “Erlang Profiling Techniques”?
Utilize Erlang's built-in profiling tools to identify performance bottlenecks and optimize your application code. You practise Erlang OTP: Distributed & Fault-Tolerant Systems Programming with hands-on code you run directly in the browser, and a 24/7 AI tutor answers your questions as you work through the lesson.
Do I need any experience to start Erlang OTP: Distributed & Fault-Tolerant Systems Programming?
No prior experience is required. Erlang OTP: Distributed & Fault-Tolerant Systems Programming on CoddyKit is structured for beginners through advanced learners; this is — lesson 1 of 4, so you can start here or from the beginning and move at your own pace.
How long does the “Erlang Profiling Techniques” lesson take?
Most CoddyKit lessons take about 5–10 minutes. Each one is bite-sized and interactive, so you make steady progress and pick up exactly where you left off across the web and the app.
Can I write and run code in this Erlang OTP: Distributed & Fault-Tolerant Systems Programming lesson?
Yes. Every Erlang OTP: Distributed & Fault-Tolerant Systems Programming lesson includes a built-in code editor, so you write and run real code right in your browser and get instant AI feedback — no local setup required.