[Public WebGL] IE11 vs Chrome

John Davis [email protected]
Sat Nov 22 02:54:32 PST 2014


Is there an ETA on release?


On Wed, Nov 19, 2014 at 10:09 AM, Jamie Madill <[email protected]> wrote:

> Hey John & Florian,
>
> We were forcing D3D9 on Intel for a while due to performance problems. Now
> that we've mostly fixed the perf in ANGLE, in Chrome M40 (currently only
> available as dev or Canary) we've enabled D3D11 for more recent Intel
> drivers.
>
> There's a tracking Chrome bug here: http://crbug.com/363721.
>
> On Tue, Nov 18, 2014 at 6:05 PM, Florian Bösch <[email protected]> wrote:
>
>> - https://code.google.com/p/chromium/issues/list
>> - https://bugzilla.mozilla.org/
>> - https://bugs.webkit.org/
>> - https://code.google.com/p/angleproject/issues/list
>> - https://connect.microsoft.com/IE
>>
>> On Tue, Nov 18, 2014 at 11:58 PM, John Davis <[email protected]>
>> wrote:
>>
>>> Ticket?  Where?  BTW, Jamie just said the Intel updates are on their way.
>>>
>>> On Tue, Nov 18, 2014 at 5:56 PM, Florian Bösch <[email protected]> wrote:
>>>
>>>> So it's the ordinary DX11 path for IE. So for whatever reason, Chrome
>>>> decided to service your webgl with DX9. Maybe file a ticket?
>>>>
>>>> On Tue, Nov 18, 2014 at 11:54 PM, John Davis <[email protected]>
>>>> wrote:
>>>>
>>>>> Platform:Win32Browser User Agent:Mozilla/5.0 (Windows NT 6.3; WOW64;
>>>>> Trident/7.0; Touch; .NET4.0E; .NET4.0C; Tablet PC 2.0; .NET CLR 3.5.30729;
>>>>> .NET CLR 2.0.50727; .NET CLR 3.0.30729; InfoPath.3; BRI/2; rv:11.0) like
>>>>> GeckoContext Name:experimental-webglGL Version:WebGL 0.94Shading
>>>>> Language Version:WebGL GLSL ES 0.94Vendor:MicrosoftRenderer:Internet
>>>>> ExplorerUnmasked Vendor:MicrosoftUnmasked Renderer:Intel(R) HD
>>>>> Graphics 4000Antialiasing:AvailableANGLE:NoMajor Performance Caveat:No
>>>>>
>>>>> On Tue, Nov 18, 2014 at 5:42 PM, Florian Bösch <[email protected]>
>>>>> wrote:
>>>>>
>>>>>> I believe that this backend was selected for you, because of the HD
>>>>>> 4000, for whatever reason. Could some of the chrome guys comment on this?
>>>>>> Also please look at http://webglreport.com/ with IE and see what you
>>>>>> get.
>>>>>>
>>>>>> On Tue, Nov 18, 2014 at 11:38 PM, John Davis <
>>>>>> [email protected]> wrote:
>>>>>>
>>>>>>> Lol, that must be it, I have Intel drivers.
>>>>>>>
>>>>>>> ANGLE (Intel(R) HD Graphics 4000 Direct3D9Ex vs_3_0 ps_3_0)
>>>>>>>
>>>>>>> When's the new update come out?
>>>>>>>
>>>>>>> On Tue, Nov 18, 2014 at 9:58 AM, Jamie Madill <[email protected]>
>>>>>>> wrote:
>>>>>>>
>>>>>>>> This 3D Texture sample seems to use 100% of my GPU; it's possible
>>>>>>>> IE handles dropped frames a little better sometimes. For me, Canary behaves
>>>>>>>> better than stable Chrome.
>>>>>>>>
>>>>>>>> FYI you can also look at about:gpu in Chrome to determine if you're
>>>>>>>> using D3D11 or D3D9. Most AMD and nVidia systems should be on D3D11, and
>>>>>>>> we're recently moving to D3D11 on Intel systems with recent drivers.
>>>>>>>>
>>>>>>>> On Tue, Nov 18, 2014 at 5:13 AM, Tibor Ouden, den <
>>>>>>>> [email protected]> wrote:
>>>>>>>>
>>>>>>>>> If you go to : http://webglreport.com/ you will see if the webgl
>>>>>>>>> implementation uses ANGLE and what version of DirectX is used.
>>>>>>>>>
>>>>>>>>> 2014-11-18 10:54 GMT+01:00 Florian Bösch <[email protected]>:
>>>>>>>>>
>>>>>>>>>> On Tue, Nov 18, 2014 at 10:38 AM, John Davis <
>>>>>>>>>> [email protected]> wrote:
>>>>>>>>>>
>>>>>>>>>>> Will Chrome/Angle be using DX11 anytime soon?
>>>>>>>>>>>
>>>>>>>>>> As Brandon has said, it depends on your system if Angle is using
>>>>>>>>>> DX11 or DX9.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>> Not sure about everyone else, but on my end this demo is smooth
>>>>>>>>>>> as glass on IE11, but stutters/jerks on Chrome.
>>>>>>>>>>>
>>>>>>>>>> That's probably not got anything to do with the backend choice.
>>>>>>>>>> It runs smooth here in Chrome/Linux.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>> Any idea what the difference is?  Or how I can make it run
>>>>>>>>>>> better on Chrome?
>>>>>>>>>>>
>>>>>>>>>> Stuttering/Jerking/etc. is most often caused by engaging the JS
>>>>>>>>>> garbage collector a lot. Each time it kicks in, it stops the world for
>>>>>>>>>> anything between 80-150ms and nothing gets drawn during that time.
>>>>>>>>>>
>>>>>>>>>> It is possible that there's a particular bug with Chrome (due to
>>>>>>>>>> vsync and whatnot) that you might be affected by (Brandon and I talked
>>>>>>>>>> about this getting somewhat worse). You should compare your demo on
>>>>>>>>>> different browsers (IE, Chrome, Firefox, Opera) and on different OSes (OSX,
>>>>>>>>>> Windows, Android, iOS). This helps you spot if you did something wrong
>>>>>>>>>> (it's bad everywhere), or something suboptimal you did (it's good on
>>>>>>>>>> exactly one browser, but not on any other), or if it's a browser bug (it's
>>>>>>>>>> bad on only one browser).
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://khronos.org/pipermail/public_webgl_khronos.org/attachments/20141122/c40b5e81/attachment.html>


More information about the public_webgl mailing list