Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×

Comment Re:Ten times as fast as which Firefox version? (Score 1) 378

I never said IE9 is slow or not competitive for javascript, I said that FF4 has the second fastest Javascript engine next to Chrome's and that IE9's Sunspider results are slightly flawed due to it eliminating dead code. IE9's dead code elimination is fine in the Sunspider test but it shows that the test itself is flawed and IE9's dead code elimination had many flaws that suggested they were only doing it to improve their Sunspider score. The V8 and Kraken benchmarks do a better job at testing modern javascript performance.

Comment Re:Ten times as fast as which Firefox version? (Score 1) 378

It explains multiple flaws and I admit I do not know if they were fixed or not. Although dead code elimination is valid, it only works on poorly written code typically found in badly designed benchmarks. The math-cordac test in Sunspider isn't trying to test how well browsers can eliminate the code, it's trying to find out how fast they can do the math. So while it's fine for IE9 to be eliminating the dead code in math-cordac, the test itself needs to be updated so that all browsers actually do the math that it's trying to test.

Comment Re:Ten times as fast as which Firefox version? (Score 1) 378

I posted this already just below this but here it is again: http://blog.mozilla.com/rob-sayre/2010/11/17/dead-code-elimination-for-beginners/
I admit that I do not know whether IE9 fixed all the flaws in their dead code elimination but it used to be incredibly flawed and dead code elimination only works for poorly written code that you would typically only find in badly written benchmarks anyways. However, the real problem is the test itself and so Sunspider should be updated so that IE9 actually does the math that Sunspider is trying to test the performance of.

I said to ignore Peacekeeper because my original post was about JAVASCRIPT and HARDWARE ACCELERATION but Peacekeeper tries to test overall browser performance. I can't find any links for you but the major problem with Peacekeeper's Javascript benchmarks is that they call getTime() from within loops instead of outside of the loop which makes the tests mostly about how fast they can call getTime() instead of running the actual code that they're trying to test.

Comment Re:Ten times as fast as which Firefox version? (Score 1) 378

For Javascript performance you can see the SS/V8/Kraken results from this comparison but note that IE9 cheats on Sunspider with faulty dead code removal that is removing code that needs to be executed and ignore Peacekeeper because it's flawed in many ways and also tests more than Javascript...

http://www.zdnet.com/blog/hardware/ie9-vs-chrome-10-vs-firefox-4-rc-vs-opera-1101-vs-safari-5-the-big-browser-benchmark/11890

For info about why FF4 RC is slower than IE9 in some hardware acceleration tests which they will be fixing soon (probably after FF4 release) see this:

http://weblogs.mozillazine.org/roc/archives/2011/03/investigating_p.html

Comment Re:Ten times as fast as which Firefox version? (Score 2) 378

What they're saying is that the benchmark is badly designed and is testing the javascript minimum timeout more than anything else. Properly designed hardware acceleration tests try to render more and more stuff until the framerate goes below 60 instead of rendering hardly anything and seeing how high the framerate goes.

Slashdot Top Deals

You knew the job was dangerous when you took it, Fred. -- Superchicken

Working...