They generally prove whatever point of view the designers of the benchmark happened to hold. What activities you include in a benchmark and how you weight them alters the result and this can be made to show your own favorite in a good light. Even if you accept that a benchmark is fair, its conclusions are only relevant to you and any projects you may be working on if the mix of activities used in the benchmark have some similarity to what you are trying to do.
- GB Emulator is derived from an open source emulator of a famous game console running a 3D demo.
You can see that three of the tests are heavily graphic oriented and there are some features from WebGL that stop the benchmark running in IE9 - but it does work in IE10. The original tasks in the V8 benchmark are:
OS kernel simulation benchmark, originally written in BCPL by Martin Richards.
One-way constraint solver, originally written in Smalltalk by John Maloney and Mario Wolczko.
Encryption and decryption benchmark based on code by Tom Wu.
Ray tracer benchmark based on code by Adam Burmister.
Regular expression benchmark generated by extracting regular expression operations from 50 of the most popular web pages.
Data manipulation benchmark that deals with splay trees and exercises the automatic memory management subsystem.
Solves NavierStokes equations in 2D, heavily manipulating double precision arrays. Based on Oliver Hunt's code.
Octane can be run on a web browser of your choice simply by visiting the web site and clicking Start. It also has a nice user interface and will work on some mobile browsers.