User-Agent: Mozilla/5.0 (Windows NT 5.1; rv:2.0b6pre) Gecko/20100904 Firefox/4.0b6pre Build Identifier: Mozilla/5.0 (Windows NT 5.1; rv:2.0b6pre) Gecko/20100904 Firefox/4.0b6pre ID:20100904041009 Rendering corruption with layers.accelerate-all set to true. As a aside: NOTICE THE TYPO in the Error Console message: "Direct3D 9 DeviceManager Initialized Succesfully." <--- s/Succesfully Reproducible: Always Steps to Reproduce: Issue 1: Garbage rending or IE9 Fish Tank demo -------- 1. Create a new profile 2. Open about:config and set layers.accelerate-all to true 3. Check Error Console to make sure you see: "Direct3D 9 DeviceManager Initialized Succesfully." NOTE: There's also "CoCreateInstance() FAILED." in Error Console -- just above 4. Restart the browser 5. Visit: http://ie.microsoft.com/testdrive/performance/fishIE%20tank/default.html 6. Notice the corruption (see screenshot #1). Issue 2: Garbage during tab opening animation -------- 7. Close the FishIE Tank tab 8. Click the "Open a new tab" tab bar button. 9. Notice the garbage text during tab opening animation. Issue 3: Garbage in TabCandy/TabView/Panorama :-) -------- 10. Make sure you have two tabs open, with at least one of the tabs having a web page loaded, such as the "Minefield Start Page." 11. Click the "Group Your Tabs" tab bar button. 12. Notice the corruption (see screenshot #2).

My Hardware, based on Error Console: Direct3D 9 DeviceManager Initialized Succesfully. Driver: ati2dvag.dll Description: ATI Radeon HD 4290 Version: 6.14.10.7111

Missed one: Issue 4: Scrolling a textbox with text results in smearing (until the region is updated about 2-3 seconds later).

How recent is your video driver? ATI unfortunately uses different version numbering for their Catalyst packages and the version files (e.g. you have 6.14.10.7111, and the latest catalyst package seems to be "10.8")

(In reply to comment #5) > How recent is your video driver? Catalyst 10.8 Asus M4A89GTD Pro/USB3 (AMD 890GX integrated graphics) If someone can get their hands on a board with the same chipset, should be pretty easy to reproduce.

Catalyst driver details: http://developer.amd.com/drivers/ccc/Pages/default.aspx

I see this on the following website: http://lazarus.interclue.com/download.html

Reproduced at this link: https://bug583548.bugzilla.mozilla.org/attachment.cgi?id=461865

Interestingly, I don't get these issue if I run these with Window XP in a VirtualBox VM. There, everything is fine. So there's a problem in the way Firefox is interacting with real hardware. This leads me to wonder if some of you are writing the hardware acceleration code against virtual hardware and not testing it on real hardware.

It looks like Issues 1, 3, and 4 were resolved in the intervening period. What remain are: Issue 2 and those represented in the videos (and of course, the typo). If no one else sees otherwise, I will later dupe this to bug 593888 and attach the videos there.

That's interesting. I wonder what we did to fix them.

It was also affecting https://bugzilla.mozilla.org/show_bug.cgi?id=589357 :) but now everything is fine

Though there are still some rendering bugs with Image animations temporal truncations and also Grafx Bot shows many failing reftests but non of those seems as hurting as this problem was :)

also experiencing currently the same text transition and image rendering problems (though im not sure if it's only for sliding images seems more a overall problem for changing images (javascript driven) :) also happens here http://www.arcinspirations.com/kobe/ if you move over the animated sections on top

(In reply to comment #15) > also experiencing currently the same text transition and image rendering > problems (though im not sure if it's only for sliding images seems more a > overall problem for changing images (javascript driven) :) also happens here > http://www.arcinspirations.com/kobe/ if you move over the animated sections on > top Hrm, those are looking pretty good for me. Could you add a screenshot of what you're seeing on this page?

(In reply to comment #16) > (In reply to comment #15) > > also experiencing currently the same text transition and image rendering > > problems (though im not sure if it's only for sliding images seems more a > > overall problem for changing images (javascript driven) :) also happens here > > http://www.arcinspirations.com/kobe/ if you move over the animated sections on > > top > > Hrm, those are looking pretty good for me. Could you add a screenshot of what > you're seeing on this page? Sure here Moving mouse from Drinking (highlighted) over to Gallery http://img844.imageshack.us/img844/8066/transitioncorrupt.png Moving mouse fast between all sections http://img8.imageshack.us/img8/9421/transitioncorruptextrem.png The Changing text on the Lazarus page (effect looks familiar ;)) http://img715.imageshack.us/img715/7690/transitioncorruptlazaru.png here also my latest Grafx Bot reftest http://brasstacks.mozilla.com/resultserv/data/results/6e95cc55-8613-41f9-bbe8-97bf1c6f6f22

Ehmm above Rendering Problems are gone in latest nightly on XP :)

(In reply to comment #19) > Ehmm above Rendering Problems are gone in latest nightly on XP :) I don't think the fixed the underlying bugs. It seems they merely blacklisted the GPU, as hw acceleration doesn't appear to be enabled. Will have to double-check when I get home.

(In reply to comment #20) > (In reply to comment #19) > > Ehmm above Rendering Problems are gone in latest nightly on XP :) > > I don't think the fixed the underlying bugs. It seems they merely blacklisted > the GPU, as hw acceleration doesn't appear to be enabled. Will have to > double-check when I get home. We should be blacklisting less than we used to.. or should be. Do double check. I suspect this was fixed by fixing of bug 596152.

@Bas Shouten What is blocking D3D9 on XP showing some actuall speedup in all the tests that get accellerated well under D2D and Win7. Chromes new GPU rendering http://blog.chromium.org/2010/09/unleashing-gpu-acceleration-on-web.html allready works under XP quite well even on your own http://demos.hacks.mozilla.org/openweb/HWACCEL/ test :) i get 66 fps with Chromes Hardware accelleration on XP and 11 either with layers on or of in Firefox :(

(In reply to comment #21) > We should be blacklisting less than we used to.. or should be. Do double check. > > I suspect this was fixed by fixing of bug 596152. Confirming this fixed. And yes, like CruNcher pointed out, Chrome 7's D3D9 acceleration on my XP system is almost twice the speed of Firefox, so something's wrong with the Firefox implementation. Mozilla/5.0 (Windows NT 5.1; rv:2.0b7pre) Gecko/20100916 Firefox/4.0b7pre ID:20100916041016 http://hg.mozilla.org/mozilla-central/rev/f38ef1080bfe