Jayank Mayukh

All PostsGitHubLinkedIn

How to write long loops in javascript that do not block the UI

June 12, 2021

Some time ago at work, I was given the task of fixing long UI freezes happening in our application. While fixing the issue I learned few things that I think a lot of javascript developers may find useful. In this article, I will try to explain them with some simple examples.

Javascript is Single Threaded

Note: Javascript can have limited multi-threaded functionality via workers.

While this is not something many people do not know about, it is easy to forget about the implications of this fact. Let us try to understand what it means to be single-threaded. For this conversation we may note the following two points:

  1. A single thread can do one task at a time.
  2. Javascript runs on the same thread that controls the UI.

Now imagine that you have a CPU-intensive task in your code that is going to take some time, let’s say 200 milliseconds. When you run this code, since a single thread can do only one task at a time and your CPU intensive task is running on the same thread which controls the UI, the UI freezes for 200 milliseconds. 200 millisecond may not sound like a lot of time, a delay of 200 milliseconds in UI is noticeably annoying as a user, and things can get ugly pretty quickly if this CPU intensive task is run frequently.

Writing a UI Freezing code

Let us create a dummy problem which we will solve in this article. Create a directory and create two files called index.html and ui-blocking.js in the directory. Paste the following code in index.html.

<!DOCTYPE html>
    <title>UI Blocking</title>
    <script type="text/javascript" src="./ui-blocking.js"></script>
            color: blue;
            color: red;
    <button onclick="cpuIntensiveTask()">
        Button with cpuIntensiveTask
        Button without cpuIntensiveTask

Paste this code in ui-blocking.js.

// try not to exceed 10 billion
const CYCLES = 1000000000;

// dummy cpu intensive task
function cpuIntensiveTask() {
    for (let i = 0; i < CYCLES; i++) {
        // some useful code

Now serve the directory on your localhost using your favourite server or you can just head over to this JS Fiddle.

Try clicking both the buttons multiple times. We can already see some difference in UI response. You can try changing the value of CYCLES. If you open your browser’s console, you will see the time it took to execute cpuIntensiveTask.

Using setTimeout Based Asynchronous Loops

Often the code blocking the UI is in form of a loop, where individual iterations do not take much time but collectively they take enough time to make things worse. Such kinds of issues can be fixed using Asynchronous Loops. This is how an async version of cpuIntensiveTask might look like:

Note: This code snippet uses async/await syntax which was introduced in ES7.

const CYCLES = 1000000000;

async function asyncCpuIntensiveTask(iter = 0) {
    if (iter !== CYCLES){
        setTimeout(() => {
            // some useful code
            asyncCpuIntensiveTask(iter + 1);
        }, 0);

Update your code to use asyncCpuIntensiveTask function instead of cpuIntensiveTask or open this JS Fiddle. This time you will notice that both the buttons, one with an asyncCpuIntensiveTask as well as the one without any callback, are equally fast in showing the UI updates.


setTimeout is a special function, it lets us schedule a callback function at least after a certain amount of time has passed. The ‘at least’ part is important, the browser does not just start executing the callback function as soon as the specified amount of time has passed. There may be some other task running on the thread and the browser cannot just stop that task and run our callback. So it waits for the call stack to become empty, that is, it waits for the currently running task to finish. Once the call stack is empty, the callback that we passed to setTimeout is executed. Also, the browser does not just keep track of setTimeout callbacks, there are callbacks to be executed from input events such as clicks, keydown, etc. too. These callbacks from the input event get a higher priority than callbacks from setTimeout.

In asyncCpuIntensiveTask we break each iteration of our loop into its own setTimeout. Every time the call stack becomes empty, the next iteration of our loop is executed. If the user makes some input during the execution of an iteration of our loop, the callback for this input event is put into the queue of callbacks to be executed. When the call stack becomes empty, the callbacks from the input event are executed first before our loop’s next iteration, since callback from input has higher priority than a setTimeout. This way, the maximum time user needs to wait for the response to their action is reduced to the time taken by an individual iteration of our loop.

Making it better

While the code shown above works fine, we would want to have a generic function so that we can use it anywhere. Also, we would want to know when this loop finishes. This can be accomplished in a variety of ways. One way to do it as follows:

async function asyncFor(iters, callback) {
    for (let i = 0; i < iters; i++) {
        const promise = new Promise((resolve, _reject) => {
            setTimeout(() => {
            }, 0);
        await promise;

The principle remains the same, but we wrap our setTimeout in a promise and await it before we move on to the next iteration. Since asyncFor returns a Promise, we can await (or use then) it to schedule some task to run after it. Open this JS Fiddle to see this code in action. You can also see how it compares to a regular for loop. Regular for loop will complete much faster, but no UI updates would be seen while its executing whereas, AsyncFor takes a much longer time to finish, but all the UI updates are visible.


Using setTimeout to break long loops can often fix problems of sluggish UI caused by long executing loops. A drawback of this method is that it takes much longer to execute the loop completely since a lot of other code is executed between each iteration. Another drawback is that sometimes it can complicate the implementation with confusing async loops. An alternative to this method is to use web workers, I will write about it in a future post.

If you found this article interesting, you might want to read the following too:

A lot of what I have written here was learned from these articles.