Flash 18.0.0.203 terrible lag

[Guest]

Hi, it would seem that build 203 introduced some problematic 'novelty'.

Problem encountered on:

• OSX 10.10.2
• Firefox, Chrome, Safari, all in their latest versions.

Example URL: EteRNA - Played by Humans. Scored by Nature.

• if you don't see numbers around the nucleobases, hit 'N'
• then click base 74, this should mutate the Cytosine (green) to an Adenine (A), normally instantly

How things used to be:

How things are now:

For the game, and its community, this is a catastrophe. From our team, I've been hearing "move away from Flash" for a long time already. With this, it has become nearly impossible to argue against that suggestion.

[markc888]

I am experiencing a similar issue with my swf following the FP update.

I use the shineMP3Encoder library to encode MP3s from bytearrays.

flash-kikko - experiments and open source code for flash - Google Project Hosting

Since the update the process takes around 1000% (yes that's one thousand) more time to complete.

[Guest]

hi mark,

Our game makes heavy use of "alchemized" C libraries (those are nucleic acids folding engines).

Am I correct when I'm guessing that your MP3 encoder is also a C library converted to SWF by means of Alchemy?

Nevermind, this URL alone answers my question: kikko/Shine-MP3-Encoder-on-AS3-Alchemy · GitHub

[markc888]

I've created a report on bugbase.com. Perhaps you could take a look and vote it up.

Bug#4020231 - Flash Player browser plugin has crippling performance on Alchemy libraries since 18.0....

[Guest]

It's the first time I'm wandering in this jungle on adobe.com, I think I would probably have posted there if I had known about bugbase. Will vote it up asap.

[Koofka]

We have encountered the exact same issue with our clients project that implements the Shine MP3 encoder and have opened a similar bugbase issue here, please up vote this as well.  From our testing the performance issue has been intermittent, but consistently problematic.

Bug#4019626 - MP3 Encoding Performance Intermittent on .203 Flash Player Update

[chris.campbell]

Thank you all for the details and the bug reports.  I'm going to take a look now and also alert our QA team to this thread and the two bug reports for immediate triage. 

[markc888]

Thanks Chris. It's reassuring to know the team are now aware of this.

[Kratika Agarwal]

Hi Koofka,

Thanks for reporting the issue.

The issue you reported Bug#4019626 - MP3 Encoding Performance Intermittent on .203 Flash Player Update is reproducible on Mac but not on win 7 with build 18.0.0.203.

We will provide the fix shortly.

[markc888]

Hi Kratika,

Thanks for looking into this for us.

Would you have any idea when a fix for the issue might become available?

Thanks,

Mark

[Koofka]

Hi Kratika,

Noted the "Success" notation on the bug reports in this thread today.  Is there a specific update you can share?  Our client who's product has been impacted is trying to make business decisions based on when or if the current issue will be resolved.

Thanks

Koofka

[markc888]

Could someone on the Adobe team PLEASE give us an update on this issue.

Thanks

[Guest]

agreed with Mark, an ETA, even imprecise would be much appreciated.

[chris.campbell]

We're hard at work on this bug, it's one of our top priorities.  The last I heard (as of this morning), we were *trying* to have a checked in later today.  Please note that it's not a trivial fix and we can't simply roll back the initial change due to security implications.  If all goes well, and the fix comes in, then we should be able to see this in the next beta (and the release on 8/11).

[Guest]

Thank you very much Chris, I appreciate a lot your taking the time to inform us.

[Kratika Agarwal]

Hi nando8888,

Could you please also confirm if your issue is fixed for your application for nuceobases issue, http://eterna.cmu.edu/game/solution/5962961/6030703/copyandview/

Thanks

[markc888]

Yes thanks Chris. I'm sure I can speak for all concerned that we very much appreciate the Adobe team working on this.

[chris.campbell]

I have a private test build available that we believe fixes many of the problems if anyone would like to test it out.  We're aware of at least one issue that remains and we're working on it.  It would be great to get feedback if anyone's up for it.  Please send me an email (ccampbel@adobe.com) and let me know what platform/browser installer you need.

Thanks,

Chris

[markc888]

Hi Chris,

I’ve tested my Flash app in Safari, Firefox and Chrome (Mac) and IE on a PC running Windows 7.

I can confirm that the lag noticed when using the ShineMP3Encoder library no longer happens.

Please thank the team for working through this to fix it.

Thanks again,

Mark

[Koofka]

Hi Chris,

Tested the private beta in Safari / Firefox (Mac) and IE / Firefox Windows 7 and an confirm the issue in .203 / .209 appears to have been fixed.  Wasn't able to test Chrome as the provided installer didn't update the PPAPI Flash Player.

Thanks to you and the team for addressing the fix so quickly and looking forward to getting it into the production version of the player.

Thanks.

[Kratika Agarwal]

Hi markc888,

The issue you reported on bugbase.com  Bug#4020231 - Flash Player browser plugin has crippling performance on Alchemy libraries since 18.0....\     has been fixed in Flash Player 18.0.0.209.

Please download latest version from https://get.adobe.com/flashplayer

Thanks

[markc888]

Hi Kratika,

Unfortunately, having uninstalled Flash Player and installed 18.0.0.209 I can't see any performance improvement using the shineMP3EncoderTest.swf.

[Koofka]

Our testing has also found the .203 performance issue for the Shine MP3 encoder not fixed with the .209 build.

To note, we have observed the problem on an intermittent basis with both .203 and .209.  Our clients application may create several successive short encoding sessions that, prior to .203, would take a sub-second times to complete encoding.  Since .203, we will see the at least 50% or more of the encoding sessions take 15 - 20 seconds to complete, while some encodings function at the previously expected speeds.

[Kratika Agarwal]

Hi nando8888,

I tried reproducing the issue on Mac 10.10 Safari with flash Player 18.0.0.194/18.0.0.203.

I could see performance improvement in 18.0.0.203.

Please see attached screenshot of Scout logs.

FP 18.0.0.203

FP 18.0.0.194

Could you please clear cache and try again. Is this result consistent for you i.e performance degradation in 18.0.0.203.

Please click on the images to view them  properly.

[Guest]

Hi, Safari forced me to update to .209 and then, for some reason, Scout won't "see" any sessions run in Safari.

I redid the test in Firefox (if I'm not mistaken, it's the same plugin). I cleared caches, restarted Firefox, and redid the test as presented in the opening comment:

I must clarify here: the problem has nothing to do with the mouseUp event itself. In my application, it's just the point where the code calls Alchemized functions. As explained in the two bug reports mentioned above, the issue has to do with Alchemy.

From what I know of that now abandoned product, the translated code is executed using a complex mix of Timer events and "controlled" exceptions (AlchemyYield for instance), in order to create "slices" of execution. I assume that the objective was to make sure that the Flash plugin would stay responsive no matter what, for the browser, or something along those lines. And this is also another wild guess on my part, but I think that possibly dozens of slices were allowed to run between timer ticks. The problem that appeared with build .203 and still present in build .209 could be that only one slice gets now executed between timer ticks, tremendously slowing down these function calls.

And no, the issue is not fixed by 18.0.0.209