Benchmark: legacy vs ir-3 - MeasureThat.net

Script Preparation code:

function isJsArray(obj_0) {
    var tmp0_unsafeCast_0 = Array.isArray(obj_0);
    return tmp0_unsafeCast_0;
  }

function isArray(obj_0) {
    var tmp;
    if (isJsArray(obj_0)) {
      tmp = !obj_0.$type$;
    } else {
      tmp = false;
    }
    return tmp;
  }

function arrayConcat(args) {
    var len = args.length;
    var tmp0_unsafeCast_0 = Array(len);
    var typed = tmp0_unsafeCast_0;
    var inductionVariable = 0;
    var last = len - 1 | 0;
    if (inductionVariable <= last)
      do {
        var i = inductionVariable;
        inductionVariable = inductionVariable + 1 | 0;
        var arr = args[i];
        if (!(!(arr == null) ? isArray(arr) : false)) {
          typed[i] = [].slice.call(arr);
        } else {
          {
            typed[i] = arr;
          }
        }
      }
       while (!(i === last));
    return [].concat.apply([], typed);
  }

var args = (new Int32Array(10000)).map(_ => Math.random(1000000));

Tests:

legacy
arrayConcat([0, 1], args)
ir
[].concat.apply([], [[0, 1], [].slice.call(args)]);

Rendered benchmark preparation results:

Suite status: <idle, ready to run>

Previous results

Test case name	Result
legacy
ir

Fastest: N/A

Slowest: N/A

Latest run results:

Run details: (Test run date: 3 years ago)

User agent: Mozilla/5.0 (Linux; Android 10; ONEPLUS A6000) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/88.0.4324.181 Mobile Safari/537.36

Browser/OS: Chrome Mobile 88 on Android

View result in a separate tab

Test name	Executions per second
legacy	293512.7 Ops/sec
ir	1100.6 Ops/sec

Autogenerated LLM Summary (model llama3.2:3b, generated 2 months ago):

LLMs can make mistakes. Check important info.

**Benchmark Explanation**

MeasureThat.net is a website that allows users to create and run JavaScript microbenchmarks. The provided benchmark definition json represents two test cases: "legacy" and "IR-3".

**Legacy Approach**

The legacy approach is implemented using the `arrayConcat` function, which takes an array of arguments as input. It checks if each argument is an array using the `isArray` function and converts it to a typed array if necessary. The resulting arrays are then concatenated using the `concat` method.

Here's what's being compared in this approach:

* `Array.isArray()` vs `isJsArray()`: These two functions check if an object is an array, but they have different implementations.
	+ `isJsArray()` checks if the object is directly an array (i.e., `obj_0 === Array`). This implementation is likely faster because it avoids type checking.
	+ `isArray()` uses a different approach by checking if the object's `$type$` property is falsy. This implementation might be slower due to additional overhead.
* `arrayConcat()` vs `concat()`: The legacy approach uses its own `arrayConcat` function to concatenate arrays, while the IR-3 approach uses the built-in `concat()` method.

**IR-3 Approach**

The IR-3 approach is implemented using the same `arrayConcat` function as the legacy approach, but with a different implementation of the `isArray` function. The IR-3 approach also uses the built-in `concat()` method to concatenate arrays.

Here's what's being compared in this approach:

* Same functions and methods are used as in the legacy approach.
* `concat()` vs `arrayConcat()`: This is the same comparison as above, with the IR-3 approach using the built-in `concat()` method.

**Pros and Cons**

* Legacy Approach:
	+ Pros: Might be faster due to direct array checking in `isJsArray()`.
	+ Cons: Might have additional overhead due to custom implementation of `isArray` and `arrayConcat`.
* IR-3 Approach:
	+ Pros: Uses built-in methods, which might be optimized by the JavaScript engine.
	+ Cons: Might be slower than the legacy approach due to type checking in `isArray`.

**Library**

The `isArray()` function uses a library or extension that adds `$type$` property to arrays. This allows for faster checks without using `Array.isArray()`. The name of this library is not specified in the benchmark definition, but it's likely a polyfill or an extension written by Mozilla.

**Special JS Feature/Syntax**

There are no special JavaScript features or syntax mentioned in this benchmark. However, the use of typed arrays (`Int32Array`) and the `$type$` property might be considered advanced topics.

**Other Alternatives**

For similar benchmarks, you can try:

* `arrayConcat()` vs `concat()` using `Array.prototype.concat()`
* `isArray()` vs `isJsArray()` with different implementations (e.g., using `Object.prototype.toString.call()`)
* Using a library like Lodash or underscore.js for array manipulation
* Implementing the benchmark using a specific JavaScript engine or virtual machine (e.g., SpiderMonkey)

Related benchmarks:

legacy vs ir-3 (version: 0)

Comparing performance of: legacy vs ir

Created: 3 years ago by: Guest

Jump to the latest result

legacy

ir

Suite status: <idle, ready to run>

Experimental features:

Fastest: N/A

Slowest: N/A

Autogenerated LLM Summary (model llama3.2:3b, generated 2 months ago):