Mostly DOM

These benchmarks are mostly testing DOM performance, not JavaScript. But they can be significantly affected by JS performance (strings, garbage collection, etc).

• ​Hixie's DOM Core Performance tests - pure test of basic DOM operations
• ​QuirksMode W3C DOM vs. innerHTML - this was meant to compare performance of using DOM methods to innerHTML to build content, but is also interesting as a cross-browser metric.

Mostly JavaScript

These could be converted to pure JS tests that can run even command-line.

• ​Ron's Small JavaScript Benchmarks - fairly math-heavy.
• ​Bits In Byte Function Compare - pure JS, a lot of numerics stuff.
• ​CORDIC - pure JS (in fact it's just a JS file). Numeric-sheavy.
• ​MD5 in JS #1 - MD5 implemented in JS
• ​MD5 in JS #2 - another version of MD5
• ​jorendorff.com JavaScript speed test (Google Cache copy) - A wide variety of JS tests, includings strings, math and basic processing.
• ​JavaScript for and while loop benchmark - compares performance of different approaches to looping.

Mixed JS and DOM

also covers some DOM stuff. The correctness checks seem wrong, as both WebKit and Firefox 2 fail a lot of them.

• ​BenchJS by 24fun.com - a grab back of JS, DOM and rendering benchmarks. Computing total is currently broken on the live site.
• ​Celtic Kane's JavaScript Speed Tests - mostly pure JS, but also some DOM and rendering benchmarks.
• ​VisiBone JavaScript Unit Tests - these are both correctness and performance tests. 1 and 2 are pure JS, but 3 * JS/Bench - ​http://www.computerbytesman.com/js/jsbench/dobench.htm - mostly pure JS, but also some DOM stuff
• ​DHTML JavaScript Benchmark - mix of DOM and JS
• iBench (JavaScript, HTML DOM, and XML DOM tests) - you need to install this on your own server (requires a Windows IIS installation to run). This is available from ​http://www.lionbridge.com/lionbridge/en-US/services/outsourced-testing/zdm-eula.htm The JavaScript test is mostly JavaScript but has a tiny handful of DOM dependencies.