https://bugs.webkit.org/show_bug.cgi?id=112345
Summary: [Qt] requestAnimationFrame performance issues
Product: WebKit
Version: 528+ (Nightly build)
Platform: Unspecified
OS/Version: Unspecified
Status: NEW
Severity: Normal
Priority: P2
Component: WebKit Qt
AssignedTo: webkit-unassigned at lists.webkit.org
ReportedBy: milian.wolff at kdab.com
See also: https://lists.webkit.org/pipermail/webkit-qt/2013-March/003546.html
Short version is: window.requestAnimationFrame has performance issues on QtWebKit and performs worse than a simple window.setTimeout approach.
Code to reproduce the issue:
``` test.qml ```
import QtQuick 2.0
import QtWebKit 3.0
import QtWebKit.experimental 1.0
WebView {
id: webView
height: 500
width: 500
url: "test.html"
experimental {
preferences.developerExtrasEnabled: true
}
}
``````
``` test.html ```
<DOCTYPE html>
<html>
<head>
<title>rAF test</title>
<script type="text/javascript">
function animateRAF() {
window.requestAnimationFrame(animateRAF);
}
function animateTimer() {
window.setTimeout(17, animateTimer)
}
// abysmal performance
window.onload = animateRAF;
// nice performance
// window.onload = animateTimer;
</script>
<body>
<h1>rAF test</h1>
</body>
</html>
``````
The animateRAF version above makes my system noticeably sluggish and I see a high CPU load:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
25016 milian 20 0 2578m 55m 36m S 54.9 0.7 0:10.93 qmlscene
312 root 20 0 206m 92m 65m S 48.2 1.2 8:41.84 X
25024 milian 30 10 1827m 48m 38m S 12.0 0.6 0:01.89 QtWebProcess
Using animateTimer instead I do not see any of the above processed in the top 20 of top at all.
I am using a git checkout of WebKit master with bd8ce398ed4dd2ce92f8db61d7e97b43a534297a, i.e. https://bugs.webkit.org/show_bug.cgi?id=112095 applied. That helped a bit but rAF is still unusable for me.
perf top -G, while running the rAF version of the above, shows this:
- 46.09% [kernel] [k] ioread32 ◆
- ioread32 ▒
- 53.13% 0xffffffffa05e80f6 ▒
- 68.79% nouveau_gem_ioctl_pushbuf ▒
0xffffffffa02ee3f3 ▒
do_vfs_ioctl ▒
sys_ioctl ▒
system_call_fastpath ▒
__GI___ioctl ▒
- 31.21% nv84_fence_sync ▒
nouveau_fence_sync ▒
validate_sync.isra.3 ▒
validate_list ▒
nouveau_gem_ioctl_pushbuf ▒
drm_ioctl ▒
do_vfs_ioctl ▒
sys_ioctl ▒
system_call_fastpath ▒
__GI___ioctl ▒
34.81% _nouveau_gpuobj_rd32 ▒
nv84_fence_read ▒
nouveau_fence_done ▒
9.60% nouveau_dma_wait ▒
1.17% nv50_crtc_cursor_set ▒
drm_mode_cursor_ioctl ▒
drm_ioctl ▒
do_vfs_ioctl ▒
sys_ioctl ▒
system_call_fastpath ▒
__GI___ioctl ▒
0.88% nv50_vm_flush
I.e. the painting seems to be an issue. So apparently rAF triggers excessive repaints. Is this maybe also a driver problem of noveau?
--
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.