The cost of JavaScript in 2019

The cost of JavaScript in 2019

In 2019, the dominant costs of processing scripts are now download and CPU execution time.

One large change to the cost of JavaScript over the last few years has been an improvement in how fast browsers can parse and compile script. In 2019, the dominant costs of processing scripts are now download and CPU execution time.

User interaction can be delayed if the browser’s main thread is busy executing JavaScript, so optimizing bottlenecks with script execution time and network can be impactful.

Actionable high-level guidance

What does this mean for web developers? Parse & compile costs are no longer as slow as we once thought. The three things to focus on for JavaScript bundles are:

  • Improve download time
    Keep your JavaScript bundles small, especially for mobile devices. Small bundles improve download speeds, lower memory usage, and reduce CPU costs.Avoid having just a single large bundle; if a bundle exceeds ~50–100 kB, split it up into separate smaller bundles. (With HTTP/2 multiplexing, multiple request and response messages can be in flight at the same time, reducing the overhead of additional requests.)On mobile you’ll want to ship much less especially because of network speeds, but also to keep plain memory usage low.* Improve execution time
    Avoid Long Tasks that can keep the main thread busy and can push out how soon pages are interactive. Post-download, script execution time is now a dominant cost.* Avoid large inline scripts (as they’re still parsed and compiled on the main thread). A good rule of thumb is: if the script is over 1 kB, avoid inlining it (also because 1 kB is when code caching kicks in for external scripts).
Why does download and execution time matter?

Why is it important to optimize download and execution times? Download times are critical for low-end networks. Despite the growth in 4G (and even 5G) across the world, our effective connection types remain inconsistent with many of us running into speeds that feel like 3G (or worse) when we’re on the go.

JavaScript execution time is important for phones with slow CPUs. Due to differences in CPU, GPU, and thermal throttling, there are huge disparities between the performance of high-end and low-end phones. This matters for the performance of JavaScript, as execution is CPU-bound.

In fact, of the total time a page spends loading in a browser like Chrome, anywhere up to 30% of that time can be spent in JavaScript execution. Below is a page load from a site with a pretty typical workload ( on a high-end desktop machine:

JavaScript processing represents 10–30% of time spent in V8 during page load.

On mobile, it takes 3–4× longer for a median phone (Moto G4) to execute Reddit’s JavaScript compared to a high-end device (Pixel 3), and over 6× as long on a low-end device (the <$100 Alcatel 1X):

The cost of Reddit’s JavaScript across a few different device classes (low-end, average, and high-end)

Note: Reddit has different experiences for desktop and mobile web, and so the MacBook Pro results cannot be compared to the other results.

When you’re trying to optimize JavaScript execution time, keep an eye out for Long Tasks that might be monopolizing the UI thread for long periods of time. These can block critical tasks from executing even if the page looks visually ready. Break these up into smaller tasks. By splitting up your code and prioritizing the order in which it is loaded, you can get pages interactive faster and hopefully have lower input latency.

Long tasks monopolize the main thread. You should break them up.

What has V8 done to improve parse/compile?

Raw JavaScript parsing speed in V8 has increased 2× since Chrome 60. At the same time, raw parse (and compile) cost has become less visible/important due to other optimization work in Chrome that parallelizes it.

V8 has reduced the amount of parsing and compilation work on the main thread by an average of 40% (e.g. 46% on Facebook, 62% on Pinterest) with the highest improvement being 81% (YouTube), by parsing and compiling on a worker thread. This is in addition to the existing off-main-thread streaming parse/compile.

V8 parse times across different versions

We can also visualize the CPU time impact of these changes across different versions of V8 across Chrome releases. In the same amount of time it took Chrome 61 to parse Facebook’s JS, Chrome 75 can now parse both Facebook’s JS AND 6 times Twitter’s JS.

In the time it took Chrome 61 to parse Facebook’s JS, Chrome 75 can now parse both Facebook’s JS and 6 times Twitter’s JS.

Let’s dive into how these changes were unlocked. In short, script resources can be streaming-parsed and-compiled on a worker thread, meaning:

  • V8 can parse+compile JavaScript without blocking the main thread.
  • Streaming starts once the full HTML parser encounters a <script> tag. For parser-blocking scripts, the HTML parser yields, while for async scripts it continues.
  • For most real-world connection speeds, V8 parses faster than download, so V8 is done parsing+compiling a few milliseconds after the last script bytes are downloaded.

The not-so-short explanation is… Much older versions of Chrome would download a script in full before beginning to parse it, which is a straightforward approach but it doesn’t fully utilize the CPU. Between versions 41 and 68, Chrome started parsing async and deferred scripts on a separate thread as soon as the download begins.

Scripts arrive in multiple chunks. V8 starts streaming once it’s seen at least 30 kB.

In Chrome 71, we moved to a task-based setup where the scheduler could parse multiple async/deferred scripts at once. The impact of this change was a ~20% reduction in main thread parse time, yielding an overall ~2% improvement in TTI/FID as measured on real-world websites.

Chrome 71 moved to a task-based setup where the scheduler could parse multiple async/deferred scripts at once.

In Chrome 72, we switched to using streaming as the main way to parse: now also regular synchronous scripts are parsed that way (not inline scripts though). We also stopped canceling task-based parsing if the main thread needs it, since that just unnecessarily duplicates any work already done.

Previous versions of Chrome supported streaming parsing and compilation where the script source data coming in from the network had to make its way to Chrome’s main thread before it would be forwarded to the streamer.

This often resulted in the streaming parser waiting for data that arrived from the network already, but had not yet been forwarded to the streaming task as it was blocked by other work on the main thread (like HTML parsing, layout, or JavaScript execution).

We are now experimenting with starting parsing on preload, and the main-thread-bounce was a blocker for this beforehand.

Leszek Swirski’s BlinkOn presentation goes into more detail:

“Parsing JavaScript in zero* time” as presented by Leszek Swirski at BlinkOn 10.

How do these changes reflect what you see in DevTools?

In addition to the above, there was an issue in DevTools that rendered the entire parser task in a way that hints that it’s using CPU (full block). However, the parser blocks whenever it’s starved for data (that needs to go over the main thread). Since we moved from a single streamer thread to streaming tasks, this became really obvious. Here’s what you’d use to see in Chrome 69:

The DevTools issue that rendered the entire parser task in a way that hints that it’s using CPU (full block)

The “parse script” task is shown to take 1.08 seconds. However, parsing JavaScript isn’t really that slow! Most of that time is spent doing nothing except waiting for data to go over the main thread.

Chrome 76 paints a different picture:

In Chrome 76, parsing is broken up into multiple smaller streaming tasks.

In general, the DevTools performance pane is great for getting a high-level overview of what’s happening on your page. For detailed V8-specific metrics such as JavaScript parse and compile times, we recommend using Chrome Tracing with Runtime Call Stats (RCS). In RCS results, Parse-Background and Compile-Background tell you how much time was spent parsing and compiling JavaScript off the main thread, whereas Parse and Compile captures the main thread metrics.

What is the real-world impact of these changes?

Let’s look at some examples of real-world sites and how script streaming applies.

Main thread vs. worker thread time spent parsing and compiling Reddit’s JS on a MacBook Pro has several 100 kB+ bundles which are wrapped in outer functions causing lots of lazy compilation on the main thread. In the above chart, the main thread time is all that really matters because keeping the main thread busy can delay interactivity. Reddit spends most of its time on the main thread with minimum usage of the Worker/Background thread.

They’d benefit from splitting up some of their larger bundles into smaller ones (e.g 50 kB each) without the wrapping to maximize parallelization — so that each bundle could be streaming-parsed + compiled separately and reduce main thread parse/compile during start-up.

Main thread vs. worker thread time spent parsing and compiling Facebook’s JS on a MacBook Pro

We can also look at a site like Facebook loads ~6MB of compressed JS across ~292 requests, some of it async, some preloaded, and some fetched with a lower priority. A lot of their scripts are very small and granular — this can help with overall parallelization on the Background/Worker thread as these smaller scripts can be streaming-parsed/compiled at the same time.

Note, you’re probably not Facebook and likely don’t have a long-lived app like Facebook or Gmail where this much script may be justifiable on desktop. However, in general, keep your bundles coarse and only load what you need.

Although most JavaScript parsing and compilation work can happen in a streaming fashion on a background thread, some work still has to happen on the main thread. When the main thread is busy, the page can’t respond to user input. Do keep an eye on the impact both downloading and executing code has on your UX.

Note: Currently, not all JavaScript engines and browsers implement script streaming as a loading optimization. We still believe the overall guidance here leads to good user experiences across the board.

The cost of parsing JSON

Because the JSON grammar is much simpler than JavaScript’s grammar, JSON can be parsed more efficiently than JavaScript. This knowledge can be applied to improve start-up performance for web apps that ship large JSON-like configuration object literals (such as inline Redux stores). Instead of inlining the data as a JavaScript object literal, like so:

const data = { foo: 42, bar: 1337 }; // 🐌

…it can be represented in JSON-stringified form, and then JSON-parsed at runtime:

const data = JSON.parse('{"foo":42,"bar":1337}'); // 🚀

As long as the JSON string is only evaluated once, the JSON.parse approach is much faster compared to the JavaScript object literal, especially for cold loads.

There’s an additional risk when using plain object literals for large amounts of data: they could be parsed twice!

  1. The first pass happens when the literal gets preparsed.
  2. The second pass happens when the literal gets lazy-parsed.

The first pass can’t be avoided. Luckily, the second pass can be avoided by placing the object literal at the top-level, or within a PIFE.

What about parse/compile on repeat visits?

V8’s (byte)code-caching optimization can help. When a script is first requested, Chrome downloads it and gives it to V8 to compile. It also stores the file in the browser’s on-disk cache. When the JS file is requested a second time, Chrome takes the file from the browser cache and once again gives it to V8 to compile. This time, however, the compiled code is serialized, and is attached to the cached script file as metadata.

Visualization of how code caching works in V8

The third time, Chrome takes both the file and the file’s metadata from the cache, and hands both to V8. V8 deserializes the metadata and can skip compilation. Code caching kicks in if the first two visits happen within 72 hours. Chrome also has eager code caching if a service worker is used to cache scripts. You can read more about code caching in code caching for web developers.


Download and execution time are the primary bottlenecks for loading scripts in 2019. Aim for a small bundle of synchronous (inline) scripts for your above-the-fold content with one or more deferred scripts for the rest of the page. Break down your large bundles so you focus on only shipping code the user needs when they need it. This maximizes parallelization in V8.

On mobile, you’ll want to ship a lot less script because of network, memory consumption and execution time for slower CPUs. Balance latency with cacheability to maximize the amount of parsing and compilation work that can happen off the main thread.

Mobile App Development Company India | Ecommerce Web Development Company India

Mobile App Development Company India | Ecommerce Web Development Company India

Best Mobile App Development Company India, WebClues Global is one of the leading web and mobile app development company. Our team offers complete IT solutions including Cross-Platform App Development, CMS & E-Commerce, and UI/UX Design.

We are custom eCommerce Development Company working with all types of industry verticals and providing them end-to-end solutions for their eCommerce store development.

Know more about Top E-Commerce Web Development Company

JavaScript developers should you be using Web Workers?

JavaScript developers should you be using Web Workers?

Do you think JavaScript developers should be making more use of Web Workers to shift execution off of the main thread?

Originally published by David Gilbertson at

So, Web Workers. Those wonderful little critters that allow us to execute JavaScript off the main thread.

Also known as “no, you’re thinking of Service Workers”.

Photo by Caleb Jones on Unsplash

Before I get into the meat of the article, please sit for a lesson in how computers work:

Understood? Good.

For the red/green colourblind, let me explain. While a CPU is doing one thing, it can’t be doing another thing, which means you can’t sort a big array while a user scrolls the screen.

This is bad, if you have a big array and users with fingers.

Enter, Web Workers. These split open the atomic concept of a ‘CPU’ and allow us to think in terms of threads. We can use one thread to handle user-facing work like touch events and rendering the UI, and different threads to carry out all other work.

Check that out, the main thread is green the whole way through, ready to receive and respond to the gentle caress of a user.

You’re excited (I can tell), if we only have UI code on the main thread and all other code can go in a worker, things are going to be amazing (said the way Oprah would say it).

But cool your jets for just a moment, because websites are mostly about the UI — it’s why we have screens. And a lot of a user’s interactions with your site will be tapping on the screen, waiting for a response, reading, tapping, looking, reading, and so on.

So we can’t just say “here’s some JS that takes 20ms to run, chuck it on a thread”, we must think about where that execution time exists in the user’s world of tap, read, look, read, tap…

I like to boil this down to one specific question:

Is the user waiting anyway?

Imagine we have created some sort of git-repository-hosting website that shows all sorts of things about a repository. We have a cool feature called ‘issues’. A user can even click an ‘issues’ tab in our website to see a list of all issues relating to the repository. Groundbreaking!

When our users click this issues tab, the site is going to fetch the issue data, process it in some way — perhaps sort, or format dates, or work out which icon to show — then render the UI.

Inside the user’s computer, that’ll look exactly like this.

Look at that processing stage, locking up the main thread even though it has nothing to do with the UI! That’s terrible, in theory.

But think about what the human is actually doing at this point. They’re waiting for the common trio of network/process/render; just sittin’ around with less to do than the Bolivian Navy.

Because we care about our users, we show a loading indicator to let them know we’ve received their request and are working on it — putting the human in a ‘waiting’ state. Let’s add that to the diagram.

Now that we have a human in the picture, we can mix in a Web Worker and think about the impact it will have on their life:


First thing to note is that we’re not doing anything in parallel. We need the data from the network before we process it, and we need to process the data before we can render the UI. The elapsed time doesn’t change.

(BTW, the time involved in moving data to a Web Worker and back is negligible: 1ms per 100 KB is a decent rule of thumb.)

So we can move work off the main thread and have a page that is responsive during that time, but to what end? If our user is sitting there looking at a spinner for 600ms, have we enriched their experience by having a responsive screen for the middle third?


I’ve fudged these diagrams a little bit to make them the gorgeous specimens of graphic design that they are, but they’re not really to scale.

When responding to a user request, you’ll find that the network and DOM-manipulating part of any given task take much, much longer than the pure-JS data processing part.

I saw an article recently making the case that updating a Redux store was a good candidate for Web Workers because it’s not UI work (and non-UI work doesn’t belong on the main thread).

Chucking the data processing over to a worker thread sounds sensible, but the idea struck me as a little, umm, academic.

First, let’s split instances of ‘updating a store’ into two categories:

  1. Updating a store in response to a user interaction, then updating the UI in response to the data change
  2. Not that first one

If the first scenario, a user taps a button on the screen — perhaps to change the sort order of a list. The store updates, and this results in a re-rendering of the DOM (since that’s the point of a store).

Let me just delete one thing from the previous diagram:

In my experience, it is rare that the store-updating step goes beyond a few dozen milliseconds, and is generally followed by ten times that in DOM updating, layout, and paint. If I’ve got a site that’s taking longer than this, I’d be asking questions about why I have so much data in the browser and so much DOM, rather than on which thread I should do my processing.

So the question we’re faced with is the same one from above: the user tapped something on the screen, we’re going to work on that request for hopefully less than a second, why would we want to make the screen responsive during that time?

OK what about the second scenario, where a store update isn’t in response to a user interaction? Performing an auto-save, for example — there’s nothing more annoying than an app becoming unresponsive doing something you didn’t ask it to do.

Actually there’s heaps of things more annoying than that. Teens, for example.

Anyhoo, if you’re doing an auto-save and taking 100ms to process data client-side before sending it off to a server, then you should absolutely use a Web Worker.

In fact, any ‘background’ task that the user hasn’t asked for, or isn’t waiting for, is a good candidate for moving to a Web Worker.

The matter of value

Complexity is expensive, and implementing Web Workers ain’t cheap.

If you’re using a bundler — and you are — you’ll have a lot of reading to do, and probably npm packages to install. If you’ve got a create-react-app app, prepare to eject (and put aside two days twice a year to update 30 different packages when the next version of Babel/Redux/React/ESLint comes out).

Also, if you want to share anything fancier than plain data between a worker and the main thread you’ve got some more reading to do (comlink is your friend).

What I’m getting at is this: if the benefit is real, but minimal, then you’ve gotta ask if there’s something else you could spend a day or two on with a greater benefit to your users.

This thinking is true of everything, of course, but I’ve found that Web Workers have a particularly poor benefit-to-effort ratio.

Hey David, why you hate Web Workers so bad?

Good question.

This is a doweling jig:

I own a doweling jig. I love my doweling jig. If I need to drill a hole into the end of a piece of wood and ensure that it’s perfectly perpendicular to the surface, I use my doweling jig.

But I don’t use it to eat breakfast. For that I use a spoon.

Four years ago I was working on some fancy animations. They looked slick on a fast device, but janky on a slow one. So I wrote fireball-js, which executes a rudimentary performance benchmark on the user’s device and returns a score, allowing me to run my animations only on devices that would render them smoothly.

Where’s the best spot to run some CPU intensive code that the user didn’t request? On a different thread, of course. A Web Worker was the correct tool for the job.

Fast forward to 2019 and you’ll find me writing a routing algorithm for a mapping application. This requires parsing a big fat GeoJSON map into a collection of nodes and edges, to be used when a user asks for directions. The processing isn’t in response to a user request and the user isn’t waiting on it. And so, a Web Worker is the correct tool for the job.

It was only when doing this that it dawned on me: in the intervening quartet of years, I have seen exactly zero other instances where Web Workers would have improved the user experience.

Contrast this with a recent resurgence in Web Worker wonderment, and combine that contrast with the fact that I couldn’t think of anything else to write about, then concatenate that combined contrast with my contrarian character and you’ve got yourself a blog post telling you that maybe Web Workers are a teeny-tiny bit overhyped.

Thanks for reading

If you liked this post, share it with all of your programming buddies!

Follow us on Facebook | Twitter

Further reading

An Introduction to Web Workers

JavaScript Web Workers: A Beginner’s Guide

Using Web Workers to Real-time Processing

How to use Web Workers in Angular app

Using Web Workers with Angular CLI

Hire PHP Developer and Web Developer for your Online Business

Hire PHP Developer and Web Developer for your Online Business

PHP is widely used open-source scripting language it helps in making dynamically easy your websites and web application. Mobiweb Technology is your best technical partner and offering you solution for any kind of website and application...

PHP is widely used open-source scripting language it helps in making dynamically easy your websites and web application. Mobiweb Technology is your best technical partner and offering you solution for any kind of website and application development. To hire PHP developer and web developer at affordable prices contact Mobiweb Technology via [email protected]