IE9 Javascript slower than IE8 - that's weird!

Question

I'm having difficulty in explaining why the following code runs slower in IE9 than IE8!

Here's some test code that runs smoothly in about half a second in (latest) FF/Chrome/Safari on OS X, WinXP, and Win7 plus IE7 & 8 on WinXP (for simplicity I removed the tweak that makes it work in IE6).

For some reason I can't explain, it's awful in IE9, slow and clunky. Reducing the time for the setTimeout makes it a bit quicker but no less jerky.

I've tried removing and benchmarking a number of what might be the choke points (Math.min for example ... all with no change.

I'm stumped ... can anyone please point me in the right direction?... preferably one that does not require browser-sniffing?

Here's the test code ...

<div id = 'panel' class='noShow' style='background-color: #aaa;'>
    <div id = 'wrapper' class='slideWrapper'>
        <p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p><p>xxxxxxxxxxx</p>
    </div>
</div>


<script type = 'text/javaScript'>

    var e = document.getElementById('panel');
    var w = document.getElementById('wrapper');

    w.style.overflow = 'hidden';                                                
    w.style.height = '1px'; 
    var sh = w.scrollHeight;
    show();
    function show()
    {
        setTimeout(function()
           {
                w.style.height = Math.min(sh, (w.offsetHeight + Math.ceil(sh/15))) + 'px';  
                if(
                     (w.offsetHeight < sh)
                ) 
                {
                    show(e);
                }
                else 
                {
                    w.style.height = 'auto'; 
                }                       
        }, 20); 
    }   
</script>

Answer 1

IE9 defaults to software rendering on VMware Virtual Machines (as of 7/8/2011) there is supposed to be a hotfix released in the near future but no timeline is given per this blog post.