JavaScript memory limit

In Chrome and Chromium OS, the memory limit is defined by the browser, and you can inspect the limit with the following command in the Developer Tools command-line by hitting F12:

> window.performance.memory.jsHeapSizeLimit
1090519040

On my Windows 10 OS, it is about 1 GB.

On Chrom(e/ium), you can get around the heap size limit by allocating native arrays:

var target = []
while (true) {
    target.push(new Uint8Array(1024 * 1024)); // 1Meg native arrays
}

This crashes the tab at around 2GB, which happens very rapidly. After that Chrom(e/ium) goes haywire, and repeating the test is not possible without restarting the browser.

I also recommend reading TrackJS’s blogpost about Monitoring JavaScript Memory before you get deep into the weeds trying to diagnose or measure anything memory related in the browser.

You can also search comp.lang.javascript for javascript memory limit.

See also these Stack Overflow posts:

1. Maximum size of an Array in Javascript, which suggests you can store up to 2³²-1 = 4,294,967,295 = 4.29 billion elements.

2. Maximum number of arguments a JavaScript function can accept

There is additional knowledge on the JS9 astronomical image display library website: Dealing with Memory Limitations.

(I was trying to find a good answer, and the “there is no upper limit” answer provided here was just silly to me. I cannot run into a production issue for a multi-million dollar project and say to management, “Well, I assumed there is no upper limit and everything would be okay.” Try to do a proof-of-concept, e.g. loading lots of combobox controls in your JavaScript UI framework of choice, etc. You may discover your framework has some performance degradation.)

Here are some components that I’ve found scale very well both in CPU performance and memory performance:

1. Microsoft Monaco editor
   • This is used by several commercial projects:
     1. Postman, as of v7.1.1-canary08
     2. VS Code

Here are some examples of frameworks with well-known performance degradation:

1. Angular: Poor change detection approach.
   • For each async event, compare each of the bindings (Model-Dom binding) to its old value to decide if to re-render.
     1. NG1: >2500 watchers, performance grinds to a halt
     2. NG2: the same problem remains but you have a long tiring workaround: Switch to immutables and spread ChangeDetectionStrategy.onPush all over your app to turn off the default problematic strategy
2. React
   • Again, Immutable collections of JS objects only scale so far.
     1. create-react-app internally uses Immutable.JS, and Immutable.JS can only create about 500k immutable collections before it dies.

Here are some other things to think about:

1. Use array.slice for manipulating arrays to minimize additional array allocations; array.slice will modify the array in place, which will reduce garbage collection and overall heap size.