[javascript] fastest MD5 Implementation in JavaScript

There are many MD5 JavaScript implementations out there. Does anybody know which one is the most advanced, most bugfixed and fastest?

I need it for this tool.

I only need to support HTML5 browsers that support typed arrays (DataView, ArrayBuffer, etc.) I think I took the Joseph Myers code and modified it to support passing in a Uint8Array. I did not catch all the improvements, and there are still probably some char() array artifacts that can be improved on. I needed this for adding to the PouchDB project.

var PouchUtils = {};
PouchUtils.Crypto = {};
(function () {
    PouchUtils.Crypto.MD5 = function (uint8Array) {
        function md5cycle(x, k) {
            var a = x[0], b = x[1], c = x[2], d = x[3];

            a = ff(a, b, c, d, k[0], 7, -680876936);
            d = ff(d, a, b, c, k[1], 12, -389564586);
            c = ff(c, d, a, b, k[2], 17, 606105819);
            b = ff(b, c, d, a, k[3], 22, -1044525330);
            a = ff(a, b, c, d, k[4], 7, -176418897);
            d = ff(d, a, b, c, k[5], 12, 1200080426);
            c = ff(c, d, a, b, k[6], 17, -1473231341);
            b = ff(b, c, d, a, k[7], 22, -45705983);
            a = ff(a, b, c, d, k[8], 7, 1770035416);
            d = ff(d, a, b, c, k[9], 12, -1958414417);
            c = ff(c, d, a, b, k[10], 17, -42063);
            b = ff(b, c, d, a, k[11], 22, -1990404162);
            a = ff(a, b, c, d, k[12], 7, 1804603682);
            d = ff(d, a, b, c, k[13], 12, -40341101);
            c = ff(c, d, a, b, k[14], 17, -1502002290);
            b = ff(b, c, d, a, k[15], 22, 1236535329);

            a = gg(a, b, c, d, k[1], 5, -165796510);
            d = gg(d, a, b, c, k[6], 9, -1069501632);
            c = gg(c, d, a, b, k[11], 14, 643717713);
            b = gg(b, c, d, a, k[0], 20, -373897302);
            a = gg(a, b, c, d, k[5], 5, -701558691);
            d = gg(d, a, b, c, k[10], 9, 38016083);
            c = gg(c, d, a, b, k[15], 14, -660478335);
            b = gg(b, c, d, a, k[4], 20, -405537848);
            a = gg(a, b, c, d, k[9], 5, 568446438);
            d = gg(d, a, b, c, k[14], 9, -1019803690);
            c = gg(c, d, a, b, k[3], 14, -187363961);
            b = gg(b, c, d, a, k[8], 20, 1163531501);
            a = gg(a, b, c, d, k[13], 5, -1444681467);
            d = gg(d, a, b, c, k[2], 9, -51403784);
            c = gg(c, d, a, b, k[7], 14, 1735328473);
            b = gg(b, c, d, a, k[12], 20, -1926607734);

            a = hh(a, b, c, d, k[5], 4, -378558);
            d = hh(d, a, b, c, k[8], 11, -2022574463);
            c = hh(c, d, a, b, k[11], 16, 1839030562);
            b = hh(b, c, d, a, k[14], 23, -35309556);
            a = hh(a, b, c, d, k[1], 4, -1530992060);
            d = hh(d, a, b, c, k[4], 11, 1272893353);
            c = hh(c, d, a, b, k[7], 16, -155497632);
            b = hh(b, c, d, a, k[10], 23, -1094730640);
            a = hh(a, b, c, d, k[13], 4, 681279174);
            d = hh(d, a, b, c, k[0], 11, -358537222);
            c = hh(c, d, a, b, k[3], 16, -722521979);
            b = hh(b, c, d, a, k[6], 23, 76029189);
            a = hh(a, b, c, d, k[9], 4, -640364487);
            d = hh(d, a, b, c, k[12], 11, -421815835);
            c = hh(c, d, a, b, k[15], 16, 530742520);
            b = hh(b, c, d, a, k[2], 23, -995338651);

            a = ii(a, b, c, d, k[0], 6, -198630844);
            d = ii(d, a, b, c, k[7], 10, 1126891415);
            c = ii(c, d, a, b, k[14], 15, -1416354905);
            b = ii(b, c, d, a, k[5], 21, -57434055);
            a = ii(a, b, c, d, k[12], 6, 1700485571);
            d = ii(d, a, b, c, k[3], 10, -1894986606);
            c = ii(c, d, a, b, k[10], 15, -1051523);
            b = ii(b, c, d, a, k[1], 21, -2054922799);
            a = ii(a, b, c, d, k[8], 6, 1873313359);
            d = ii(d, a, b, c, k[15], 10, -30611744);
            c = ii(c, d, a, b, k[6], 15, -1560198380);
            b = ii(b, c, d, a, k[13], 21, 1309151649);
            a = ii(a, b, c, d, k[4], 6, -145523070);
            d = ii(d, a, b, c, k[11], 10, -1120210379);
            c = ii(c, d, a, b, k[2], 15, 718787259);
            b = ii(b, c, d, a, k[9], 21, -343485551);

            x[0] = add32(a, x[0]);
            x[1] = add32(b, x[1]);
            x[2] = add32(c, x[2]);
            x[3] = add32(d, x[3]);

        }

        function cmn(q, a, b, x, s, t) {
            a = add32(add32(a, q), add32(x, t));
            return add32((a << s) | (a >>> (32 - s)), b);
        }

        function ff(a, b, c, d, x, s, t) {
            return cmn((b & c) | ((~b) & d), a, b, x, s, t);
        }

        function gg(a, b, c, d, x, s, t) {
            return cmn((b & d) | (c & (~d)), a, b, x, s, t);
        }

        function hh(a, b, c, d, x, s, t) {
            return cmn(b ^ c ^ d, a, b, x, s, t);
        }

        function ii(a, b, c, d, x, s, t) {
            return cmn(c ^ (b | (~d)), a, b, x, s, t);
        }

        function md51(s) {
            txt = '';
            var n = s.length,
            state = [1732584193, -271733879, -1732584194, 271733878], i;
            for (i = 64; i <= s.length; i += 64) {
                md5cycle(state, md5blk(s.subarray(i - 64, i)));
            }
            s = s.subarray(i - 64);
            var tail = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0];
            for (i = 0; i < s.length; i++)
                tail[i >> 2] |= s[i] << ((i % 4) << 3);
            tail[i >> 2] |= 0x80 << ((i % 4) << 3);
            if (i > 55) {
                md5cycle(state, tail);
                for (i = 0; i < 16; i++) tail[i] = 0;
            }
            tail[14] = n * 8;
            md5cycle(state, tail);
            return state;
        }

        /* there needs to be support for Unicode here,
         * unless we pretend that we can redefine the MD-5
         * algorithm for multi-byte characters (perhaps
         * by adding every four 16-bit characters and
         * shortening the sum to 32 bits). Otherwise
         * I suggest performing MD-5 as if every character
         * was two bytes--e.g., 0040 0025 = @%--but then
         * how will an ordinary MD-5 sum be matched?
         * There is no way to standardize text to something
         * like UTF-8 before transformation; speed cost is
         * utterly prohibitive. The JavaScript standard
         * itself needs to look at this: it should start
         * providing access to strings as preformed UTF-8
         * 8-bit unsigned value arrays.
         */
        function md5blk(s) { /* I figured global was faster.   */
            var md5blks = [], i; /* Andy King said do it this way. */
            for (i = 0; i < 64; i += 4) {
                md5blks[i >> 2] = s[i]
                + (s[i + 1] << 8)
                + (s[i + 2] << 16)
                + (s[i + 3] << 24);
            }
            return md5blks;
        }

        var hex_chr = '0123456789abcdef'.split('');

        function rhex(n) {
            var s = '', j = 0;
            for (; j < 4; j++)
                s += hex_chr[(n >> (j * 8 + 4)) & 0x0F]
                + hex_chr[(n >> (j * 8)) & 0x0F];
            return s;
        }

        function hex(x) {
            for (var i = 0; i < x.length; i++)
                x[i] = rhex(x[i]);
            return x.join('');
        }

        function md5(s) {
            return hex(md51(s));
        }

        function add32(a, b) {
            return (a + b) & 0xFFFFFFFF;
        }

        return md5(uint8Array);
    };
})();

I wrote tests to compare several JavaScript hash implementations, including most MD5 implementations mentioned here. To run the tests, go to http://brillout.github.io/test-javascript-hash-implementations/ and wait a bit.

It seems that the YaMD5 implementation of R. Hill's answer is the fastest.

Node.js has built-in support

const crypto = require('crypto')
crypto.createHash('md5').update('hello world').digest('hex')

Code snippet above computes MD5 hex string for string hello world

The advantage of this solution is you don't need to install additional library.

I think built in solution should be the fastest. If not, we should create issue/PR for the Node.js project.

Currently the fastest implementation of md5 (based on Joseph Myers' code):

https://github.com/iReal/FastMD5

jsPerf comparaison: http://jsperf.com/md5-shootout/63

If the performance of your application is limited by a Javascript implementation of MD5, then you're really doing something wrong. Consider an architectural change (Hint: use MD5 less often)

I found a number of articles on this subject. They all suggested Joseph Meyers implementation.

see: http://jsperf.com/md5-shootout on some tests

in My quest for the ultimate speed i looked at this code, an i saw that it could be improved. So i created a new JS script based on the Joseph Meyers code.

see Improved Jospeh Meyers code

Just for fun,

this one is 42 lines long, fits in 120 characters horizontally, and looks good. Is it fast? Well - it's fast enough and it's approximately the same as all other JS implementations.

I just wanted something that doesn't look ugly in my helpers.js file and doesn't slow down my SublimeText with 20-mile long minified one-liners.

So here's my favourite md5.

//  A formatted version of a popular md5 implementation.
//  Original copyright (c) Paul Johnston & Greg Holt.
//  The function itself is now 42 lines long.

function md5(inputString) {
    var hc="0123456789abcdef";
    function rh(n) {var j,s="";for(j=0;j<=3;j++) s+=hc.charAt((n>>(j*8+4))&0x0F)+hc.charAt((n>>(j*8))&0x0F);return s;}
    function ad(x,y) {var l=(x&0xFFFF)+(y&0xFFFF);var m=(x>>16)+(y>>16)+(l>>16);return (m<<16)|(l&0xFFFF);}
    function rl(n,c)            {return (n<<c)|(n>>>(32-c));}
    function cm(q,a,b,x,s,t)    {return ad(rl(ad(ad(a,q),ad(x,t)),s),b);}
    function ff(a,b,c,d,x,s,t)  {return cm((b&c)|((~b)&d),a,b,x,s,t);}
    function gg(a,b,c,d,x,s,t)  {return cm((b&d)|(c&(~d)),a,b,x,s,t);}
    function hh(a,b,c,d,x,s,t)  {return cm(b^c^d,a,b,x,s,t);}
    function ii(a,b,c,d,x,s,t)  {return cm(c^(b|(~d)),a,b,x,s,t);}
    function sb(x) {
        var i;var nblk=((x.length+8)>>6)+1;var blks=new Array(nblk*16);for(i=0;i<nblk*16;i++) blks[i]=0;
        for(i=0;i<x.length;i++) blks[i>>2]|=x.charCodeAt(i)<<((i%4)*8);
        blks[i>>2]|=0x80<<((i%4)*8);blks[nblk*16-2]=x.length*8;return blks;
    }
    var i,x=sb(inputString),a=1732584193,b=-271733879,c=-1732584194,d=271733878,olda,oldb,oldc,oldd;
    for(i=0;i<x.length;i+=16) {olda=a;oldb=b;oldc=c;oldd=d;
        a=ff(a,b,c,d,x[i+ 0], 7, -680876936);d=ff(d,a,b,c,x[i+ 1],12, -389564586);c=ff(c,d,a,b,x[i+ 2],17,  606105819);
        b=ff(b,c,d,a,x[i+ 3],22,-1044525330);a=ff(a,b,c,d,x[i+ 4], 7, -176418897);d=ff(d,a,b,c,x[i+ 5],12, 1200080426);
        c=ff(c,d,a,b,x[i+ 6],17,-1473231341);b=ff(b,c,d,a,x[i+ 7],22,  -45705983);a=ff(a,b,c,d,x[i+ 8], 7, 1770035416);
        d=ff(d,a,b,c,x[i+ 9],12,-1958414417);c=ff(c,d,a,b,x[i+10],17,     -42063);b=ff(b,c,d,a,x[i+11],22,-1990404162);
        a=ff(a,b,c,d,x[i+12], 7, 1804603682);d=ff(d,a,b,c,x[i+13],12,  -40341101);c=ff(c,d,a,b,x[i+14],17,-1502002290);
        b=ff(b,c,d,a,x[i+15],22, 1236535329);a=gg(a,b,c,d,x[i+ 1], 5, -165796510);d=gg(d,a,b,c,x[i+ 6], 9,-1069501632);
        c=gg(c,d,a,b,x[i+11],14,  643717713);b=gg(b,c,d,a,x[i+ 0],20, -373897302);a=gg(a,b,c,d,x[i+ 5], 5, -701558691);
        d=gg(d,a,b,c,x[i+10], 9,   38016083);c=gg(c,d,a,b,x[i+15],14, -660478335);b=gg(b,c,d,a,x[i+ 4],20, -405537848);
        a=gg(a,b,c,d,x[i+ 9], 5,  568446438);d=gg(d,a,b,c,x[i+14], 9,-1019803690);c=gg(c,d,a,b,x[i+ 3],14, -187363961);
        b=gg(b,c,d,a,x[i+ 8],20, 1163531501);a=gg(a,b,c,d,x[i+13], 5,-1444681467);d=gg(d,a,b,c,x[i+ 2], 9,  -51403784);
        c=gg(c,d,a,b,x[i+ 7],14, 1735328473);b=gg(b,c,d,a,x[i+12],20,-1926607734);a=hh(a,b,c,d,x[i+ 5], 4,    -378558);
        d=hh(d,a,b,c,x[i+ 8],11,-2022574463);c=hh(c,d,a,b,x[i+11],16, 1839030562);b=hh(b,c,d,a,x[i+14],23,  -35309556);
        a=hh(a,b,c,d,x[i+ 1], 4,-1530992060);d=hh(d,a,b,c,x[i+ 4],11, 1272893353);c=hh(c,d,a,b,x[i+ 7],16, -155497632);
        b=hh(b,c,d,a,x[i+10],23,-1094730640);a=hh(a,b,c,d,x[i+13], 4,  681279174);d=hh(d,a,b,c,x[i+ 0],11, -358537222);
        c=hh(c,d,a,b,x[i+ 3],16, -722521979);b=hh(b,c,d,a,x[i+ 6],23,   76029189);a=hh(a,b,c,d,x[i+ 9], 4, -640364487);
        d=hh(d,a,b,c,x[i+12],11, -421815835);c=hh(c,d,a,b,x[i+15],16,  530742520);b=hh(b,c,d,a,x[i+ 2],23, -995338651);
        a=ii(a,b,c,d,x[i+ 0], 6, -198630844);d=ii(d,a,b,c,x[i+ 7],10, 1126891415);c=ii(c,d,a,b,x[i+14],15,-1416354905);
        b=ii(b,c,d,a,x[i+ 5],21,  -57434055);a=ii(a,b,c,d,x[i+12], 6, 1700485571);d=ii(d,a,b,c,x[i+ 3],10,-1894986606);
        c=ii(c,d,a,b,x[i+10],15,   -1051523);b=ii(b,c,d,a,x[i+ 1],21,-2054922799);a=ii(a,b,c,d,x[i+ 8], 6, 1873313359);
        d=ii(d,a,b,c,x[i+15],10,  -30611744);c=ii(c,d,a,b,x[i+ 6],15,-1560198380);b=ii(b,c,d,a,x[i+13],21, 1309151649);
        a=ii(a,b,c,d,x[i+ 4], 6, -145523070);d=ii(d,a,b,c,x[i+11],10,-1120210379);c=ii(c,d,a,b,x[i+ 2],15,  718787259);
        b=ii(b,c,d,a,x[i+ 9],21, -343485551);a=ad(a,olda);b=ad(b,oldb);c=ad(c,oldc);d=ad(d,oldd);
    }
    return rh(a)+rh(b)+rh(c)+rh(d);
}

But really, I posted it merely out of aesthetic considerations. Also, with the comments it's exactly 4000 bytes. Please don't ask why. I can't come up with a proper explanation for my OCD/rebel behaviour. Also, thank you Paul Johnston, thank you Greg Holt. (Side note: you guys omitted a few var keywords so I took the liberty of adding them.)

Much faster hashing should be possible by calculating on graphic card (implement hashing algorithm in WebGL), as discussed there about SHA256: Is it possible to calculate sha256 hashes in the browser using the user's video card, eg. by using WebGL or Flash?

While selecting library it's also important to see if it supports modern frameworks such as Bower, passes jslint, supports plugin model for JQuery or module systems such as AMD/RequireJS in addition to being in active development and have more than 1 contributors. There are couple of options that satisfies some or all of these additional criteria:

• CryptoJS: This is perhaps the most expansive library where each algorithm can be used separately without adding fat in to your JS code. Plus it as encoder/decoders for UTF8, UTF16 and Base64. I maintain github repository that is registered as Bower package plus instructions on how to use it with RequireJS.
• Spark MD5: This is based on JKM code that other answer mentions which is also the faster implementation. However in addition, Spark implementation adds AMD support, passes jslint plus has incremental mode. It doesn't have Base64 o/p but it does have raw o/p (i.e. array of 32-bit int insead of string).
• JQuery MD5 plugin: Very simple down to earth but doesn't seem to have raw mode.
• JavaScript-MD5: Not as fancy or fast as Spark but simpler.

Example from CryptoJS:

//just include md5.js from the CryptoJS rollups folder
var hash = CryptoJS.MD5("Message");
console.log(hash.toString()); 

There is a performance comparison between above libraries at http://jsperf.com/md5-shootout/7. On my machine current tests (which are admittedly old) shows that if you are looking for speed Spark MD5 is your best bet (and so is plain JKM code). However if you looking for more comprehensive library then CryptoJS is your best bet although it is 79% slower than Spark MD5. However I would imagine CryptoJS would eventually achieve same speed as it is bit more active project.

It bothered me that I could not find an implementation which is both fast and support Unicode strings.

So I made one which supports Unicode strings and still shows as faster (at time of writing) than the currently fastest ascii-only-strings implementations:

https://github.com/gorhill/yamd5.js

Based on Joseph Myers' code, but uses TypedArrays, plus other improvements.

I would suggest you use CryptoJS in this case.

Basically CryptoJS is a growing collection of standard and secure cryptographic algorithms implemented in JavaScript using best practices and patterns. They are fast, and they have a consistent and simple interface.

So if you want to calculate the MD5 hash of your password string then do as follows:

<script src="https://cdnjs.cloudflare.com/ajax/libs/crypto-js/3.1.9-1/core.js"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/crypto-js/3.1.9-1/md5.js"></script>
<script>
    var passhash = CryptoJS.MD5(password).toString();

    $.post(
      'includes/login.php', 
      { user: username, pass: passhash },
      onLogin, 
      'json' );
</script>

So this script will post the hash of your password string to the server.

For further info and support on other hash calculating algorithms you can visit:

http://code.google.com/p/crypto-js/

You could also check my md5 implementation. It should be approx. the same as the other posted above. Unfortunately, the performance is limited by the inner loop which is impossible to optimize more.

Maybe this package was useful
https://www.npmjs.com/package/pure-md5

console.time('latin');_x000D_
const latin = md5('hello');_x000D_
console.timeEnd('latin');_x000D_
_x000D_
console.log('??????: ', md5('??????'));_x000D_
console.log('?: ', md5('?'));

<script src="https://unpkg.com/pure-md5@latest/lib/index.js"></script>

js-md5 supports UTF-8 string, array, ArrayBuffer, AMD....

and fast. jsperf

Why not try http://phpjs.org/functions/md5/?

Unfortunately performance is limited with any emulated script, however this can render real md5 hash. Although I would advice against using md5 for passwords, as it is a fast-rendered hash.

As of 2020 the fastest MD5 implementation is probably written in WASM (Web Assembly).

hash-wasm is a library that implements MD5 hash in WASM.

You can find the benchmarks here.

You can either install it with npm:

npm i hash-wasm

or just add a script tag

<script src="https://cdn.jsdelivr.net/npm/hash-wasm"></script>

then use the hashwasm global variable.

Example:

async function run() {
  console.log('MD5:', await hashwasm.md5('The quick brown fox jumps over the lazy dog'));
}

run();

outputs

MD5: 9e107d9d372bb6826bd81d3542a419d6