157903 - timerout has vast speed diff between ie5/6 ns4/7/moz

Status: RESOLVED DUPLICATE of bug 64516

(Core :: DOM: Core & HTML, defect)

Description

[Richard Bozman]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Win 9x 4.90; en-US; rv:1.1b) Gecko/20020717
BuildID:    2002071704

DHTML  snow in javascript displays at much different droping speeds in
ie5/6  ns4.79 and netscape 7 or moz daily.

Speed changes with cursor movement on screen. (this is not a follow
cursor script. timing is affected by moving the mouse in all browsers
tested.

Reproducible: Always
Steps to Reproduce:
1.tried it in all july dailys.
2.
3.

Actual Results:  same, drop speeds vary a lot between browsers.

Expected Results:  same timing

this script tests for ie, ns4, and ns6+

Comment 1

[Phil Schwartau]

This is a DHTML or Layout performance issue, not JS Engine. 
Here is the principal function: 


function snowIE_NS6() {  // IE and NS6 main animation function
    for (i = 0; i < no; ++ i) {  // iterate for every dot
      yp[i] += sty[i];
      if (yp[i] > doc_height-50) {
        xp[i] = Math.random()*(doc_width-am[i]-30);
        yp[i] = 0;
        stx[i] = 0.02 + Math.random()/10;
        sty[i] = 0.7 + Math.random();
        doc_width = ns6up?window.innerWidth-50 : document.body.clientWidth-50;
        doc_height = ns6up?window.innerHeight : document.body.clientHeight;
      }
      dx[i] += stx[i];
      if (ie4up){
      document.all["dot"+i].style.pixelTop = yp[i];
      document.all["dot"+i].style.pixelLeft = xp[i] + am[i]*Math.sin(dx[i]);
      }
      else if (ns6up){
      document.getElementById("dot"+i).style.top=yp[i];
      document.getElementById("dot"+i).style.left=xp[i] + am[i]*Math.sin(dx[i]);
      }   
    }
    setTimeout("snowIE_NS6()", 10);
}



On my WinNT(SP6) 500MHz 128RAM, this takes up about 60% CPU in IE6.
But in Mozilla trunk 20020715xx, it pins the CPU at 100%.

In IE6 the snowflakes fall smoothly in microscopic increments.
In Moz the movements seem to coalesce and are very awkward-looking.

Reassigning to DOM Level 0. Note the DHTML performance tracking bug 21762.
If this bug is not a duplicate of any other, we should add it as a
dependency to that tracking bug -

Comment 2

[Johnny Stenback (:jst)]

While the page is loading I too see the weird movement of the snowflakes, but
once the page finishes loading the movement of the snowflakes is nice and
smooth. We do use 100% cpu tho...

I bet this is a dupe of the other "snowflake" bug whose number I don't have handy...

Comment 3

[Phil Schwartau]

This looks like it: bug 64516  "95-99% CPU usage for DHTML 'snow effect'".
I will dupe this bug against it and add the URL to the info in that bug.

[e-mail]: thank you for this report; you will be cc'ed on the 
duplicate bug so you can follow its progress -

*** This bug has been marked as a duplicate of 64516 ***