Slashdot Mirror
Freeciv As Benchmark of HTML5 Canvas Javascript Performance
Andreas(R) writes "The Freeciv.net crew has benchmarked their web client, which is a rich web application using the HTML5 canvas element. This shows how fast Firefox, Google Chrome, Safari and Internet Explorer perform using the latest HTML5 web standards."
If it was just GPOs and WSUS, IE8 would dominate simply for security reasons. The main reason for IE6 is the combination of idiotic managers/developers that have locked a lot of applications into IE6 only. As for 2012, we got approval to upgrade to IE7 six months ago. Thanks, Oracle.
I think what info we've released publicly with IE9 is promising. New and vastly improved javascript engine, hardware accelerated rendering, lots of new standards support (and we're highly active in ECMAScript v5 and joined the SVG working group). Oh, and did I mention, we now do rounded borders?!
A year ago I experimented with HTML5, and made (you guessed it) a Tetris clone, which took advantage of Canvas elements.
I noted that when drawing entire images, it was all very fast. Drawing a frame took about 12ms in Firefox and Opera. (limited by the precision of the timer)
Then I tried combining all the images into one, and drawing a region from the tileset. Talk about slowdown! Wow! Separate 64x64 images blitted fast, but as soon as it was dealing with a 512x512 image, the time to render jumped to about 500ms.
I did some quick pixel math and concluded both Opera and Firefox must've been making a copy of the entire tileset every time I tried to blit a region from it. It's the only thing that added up. When I boosted the size to 1024x1024, it jumped to over 2000ms for a frame. Completely ridiculous! ;)
Perhaps someone else could chime in about whether this bug has been fixed? Note: I was blitting from Image elements to Canvas elements. Canvas to Canvas always worked fine for me.
And I believe the trend will be for consumer CPUs to aim for lower heat and power, rather than higher speed. Unfortunately, the abstraction layers just keep piling on there.
Give it another few years, and we might not be able to emulate Commodore 64 games on the desktop any more.
If you were blocking sigs, you wouldn't have to read this.
No, it's a question of scalability, which is often more important than raw speed. With some systems, they perform well in relatively restricted hardware, but the performance improvement when you add more does not scale linearly with the extra RAM, CPU, and so on. With others, you get more constant overhead, but better scalability. Think of the overall performance as constant overhead + scalability load * resources. With XP, it sounds like the constant overhead is lower (which makes sense, as it had to run on 200MHz chips), but the scalability load is higher (which also makes sense, because it wasn't designed for 4+ cores and 2+GB of RAM).
Or, to put it another way, if XP gets 80% of the maximum theoretical performance out of a 200MHz Pentium with 128MB of RAM, but only 50% of the maximum theoretical performance from a 2GHz Core 2 Duo with 4GB of RAM, while Vista gets 50% and 70%, respectively, what the grandparent said would be true and contain no contradictions.
Various things in modern operating systems are optimised to take advantage of lots of spare RAM (for example, aggressive pre-fetching of data from the disk). Splitting services up into concurrent tasks has more overhead from context switching, but lets you scale better to multiple processors. Older desktop operating systems treated RAM as a very scarce resource and were heavily optimised for the single-CPU case, because hardly anyone had more than one CPU.
I am TheRaven on Soylent News
Based on your description I doubt the problem is with IE 8, rather I suspect the problem is with a mis-behaving browser plugin. New blank tabs should open nearly instantly and each tab loads in a separate thread
For instance, the Sun Java SSV Helper plugin for IE tends to cause a lot of the problems that you are describing including taking 3-4 seconds to open new tabs at times. I have no idea exactly what the Java SSV Helper plugin does but I have yet to encounter a Java applet that won't run without it, so I just disable it.
I have also seen the Adobe PDF Link Helper plugin cause problems (although the latest version of Adobe Reader 9 appears to have fixed most of those problems.)
Try starting Internet Explorer using the No Add-Ons shortcut and see if you still have problems. If performance is improved then you can launch IE the normal way and go to Manage Add-Ons and try disabling add-ons one by one until you find the ones that are causing problems.