Recent Flash plugin (version 30) is running many projects very slowly

-Bonfire-

ScratchDiogoh wrote:
ArnoHu wrote:
ArnoHu wrote:
RokCoder wrote:
ArnoHu wrote:
There is an update now on the Adobe bug report, which sheds some more light upon the issue (from: https://forums.adobe.com/thread/2499947):

Thanks for the reply. Timer functions are used in the core event loop for the Scratch interpreter as well as in the “timer” blocks that are used by kids for a variety of projects. We are looking at ways to mitigate / work around this change, but it's currently a little unclear how to overcome such a drastic performance change to the common library.

You know it isn't going to bode well when the latest official response starts with -
Unfortunately, reality is messy. We didn't create this particular problem…

Hi there, first of all thanks for bringing this to attention in the first place!

Yes indeed. I just tried the local workaround Adobe mentioned (TimerJitterMicroseconds=0 in mms.cfg), and it worked on Firefox. Of course that's not an option in reality - asking 30mio users to make a security-critical config change that would affect all Flash applications.

Browser vendors introduced timer jitters to JavaScript APIs beginning of the year already without breaking websites worldwide, so that's certainly possible. Performance-testing such a change in advance would have been nice, maybe even inviting the Scratch team to do so (probably the largest community still using Flash today).

Question to the Scratch team: Do you see a workaround within Scratch on the horizon? As Adobe does not seem to consider fixing that soon.

thisandagain mentioned on his profile today that the Scratch team has worked all weekend and is close to a fix. Thanks a ton!
incredibly

Anti [Anti Evil Kumquat v]

ScratchDiogoh

-Bonfire- wrote:
Total Cut

What has my signature?

Last edited by ScratchDiogoh (June 11, 2018 16:33:59)

thisandagain

Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).

ArnoHu

thisandagain wrote:
Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).

OK, I tested it with my chess project (https://scratch.mit.edu/projects/148769358/), and while it is certainly faster than before, it did not reach the old performance again either. More than that, I experience another regression now - the animations for moving pieces as well as progress bar rendering don't work any more. I will have a closer look at what's going there, but it might have to do with how I invoke pen-painting / moving sprites.

Last edited by ArnoHu (June 11, 2018 18:53:08)

thisandagain

ArnoHu wrote:
thisandagain wrote:
Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).

OK, I tested it with my chess project (https://scratch.mit.edu/projects/148769358/), and while it is certainly faster than before, it did not reach the old performance again either. More than that, I experience another regression now - the animations for moving pieces as well as progress bar rendering don't work any more. I will have a closer look at what's going there, but it might have to do with how I invoke pen-painting / moving sprites - a no screen-refresh block sends a message, and the message receiver does the rest.

Hmm. I'll take a look. Custom procedures with no screen-refresh share a code path with turbo mode so that may be the issue.

ArnoHu

thisandagain wrote:
ArnoHu wrote:
thisandagain wrote:
Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).

OK, I tested it with my chess project (https://scratch.mit.edu/projects/148769358/), and while it is certainly faster than before, it did not reach the old performance again either. More than that, I experience another regression now - the animations for moving pieces as well as progress bar rendering don't work any more. I will have a closer look at what's going there, but it might have to do with how I invoke pen-painting / moving sprites - a no screen-refresh block sends a message, and the message receiver does the rest.

Hmm. I'll take a look. Custom procedures with no screen-refresh share a code path with turbo mode so that may be the issue.

An additional finding, the browser reports the plugin not responding after some 30sec or so of busy think time. Like a message loop not returning from handling messages. This also didn't happen before.

heldlaw

I was working on a project with an infinite loop calling a no-refresh custom block drawing frames using pen. You know, exactly the stuff that was taking a major performance hit. Just now I reloaded the project, and the speed improvement was stunning. In terms of performance, I think I can call this patch a great success. Thank you Scratch team!

ArnoHu

thisandagain wrote:
ArnoHu wrote:
thisandagain wrote:
Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).

OK, I tested it with my chess project (https://scratch.mit.edu/projects/148769358/), and while it is certainly faster than before, it did not reach the old performance again either. More than that, I experience another regression now - the animations for moving pieces as well as progress bar rendering don't work any more. I will have a closer look at what's going there, but it might have to do with how I invoke pen-painting / moving sprites - a no screen-refresh block sends a message, and the message receiver does the rest.

Hmm. I'll take a look. Custom procedures with no screen-refresh share a code path with turbo mode so that may be the issue.

OK, interestingly animations worked when the engine's move came from the opening book, but not when the engine calculated it. Opening book is just a quick lookup, calculating of course is much more CPU intense and may take 2 to 45 seconds. Just for testing I added a 10ms wait block before triggering the animation, and surprisingly that solved it:

Calculate[...]
wait (0.01) secs
broadcast [MoviePiece] and wait

no screen-refresh was only on in one of the two described cases, so not sure if this is really related.

Last edited by ArnoHu (June 11, 2018 19:34:59)

ideapad-320

ScratchDiogoh wrote:
ideapad-320 wrote:
This Is Impacting
32544872 projects
How did you find it?

SATISTICS

ideapad-320

ideapad-320 wrote:
ScratchDiogoh wrote:
ideapad-320 wrote:
This Is Impacting
32544872 projects
How did you find it?
SATISTICS

oops caps lock.

ScratchDiogoh

I think we can Close because
I went to my Tester and is walking Fast and Well!

Raytracing

thisandagain wrote:
Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).

Wow that patch has an amazing difference! Now I can work on some 3D projects again

-ShadowOfTheFuture-

thisandagain wrote:
Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).

This patch has really helped a lot

I can finally start working on my pen projects again!

Psiborg

Good news indeed. Thanks for your work on this Scratch Team!

ScratchDiogoh

Very Good, Now We Can Make Pen Projects and 3D

Raytracing wrote:
thisandagain wrote:
Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).

Wow that patch has an amazing difference! Now I can work on some 3D projects again

-ShadowOfTheFuture- wrote:
thisandagain wrote:
Hi folks. We have deployed a patch to address the performance issues introduced in Flash 30. Please try out your projects again and let us know if you continue to see any major issues. Please note that while this fix *mostly* gets the performance back to where it was before Adobe's release, it is not exactly the same so you still may see some subtle differences (particularly in turbo mode).
This patch has really helped a lot I can finally start working on my pen projects again!

Scratch-Minion

My projects are mostly better with the patch. Thank-you Scratch Team for all the hard work!

There are some exceptions, one is especially strange.

My Rhombicosidodecahedron Morph project https://scratch.mit.edu/projects/222617296/ works ok when I am logged in online but lags terribly if I am not logged in!

I know that sounds really strange but I tested it a few times in two different browsers (Edge and Chrome under Windows 10) and the result was consistent; it always worked online when I was logged in and lagged terribly when I was online but not logged in.

Any ideas how being logged in could affect the patch?

(Maybe some Scratchers could run this project while they are logged in and not logged in and see if they experience the same).

———————

For me, the end project on the Featured Projects line on the Scratch homepage (3D Space Gems https://scratch.mit.edu/projects/227634998/ ) still lags terribly on the second object - the first wireframe object has always worked fine.
Could other Scratchers test this as the project author @huagoose put a warning on the project about lag and I am not sure whether it can be removed.

Last edited by Scratch-Minion (June 11, 2018 22:13:21)

Pezd

Thank you. My projects now run smoothly. I had a problem while loading the inside of a game though, and I only post it here because it might have happened from this last update since I have never seen anything like it before. The screen went black then the blocks loaded but I couldn't click or drag anything. Might be totally unrelated and just a random glitch with safari or something.

Scratch-Minion wrote:
…

For me, the end project on the Featured Projects line on the Scratch homepage (3D Space Gems https://scratch.mit.edu/projects/227634998/ ) still lags terribly on the second object - the first wireframe object has always worked fine.
Could other Scratchers test this as the project author @huagoose put a warning on the project about lag and I am not sure whether it can be removed.

All the shapes run fine for me.

hedgehog_blue

ideapad-320 wrote:
hedgehog_blue wrote:
ideapad-320 wrote:
I got a update! the
(timer)
block is slow. use these instead:
when gf clicked
forever
change [timer v] by (0.1)
wait (0.1) secs//keep timer going
end
set [timer v] to (0)//reset timer
(timer::variables)//find out timer
Actually, that workaround to the slow timer won't work, because then the automatic wait between scripts would cause it to also be slow. Instead, something related to the (current ) or (days since 2000) blocks would be better.

But back to the point: It's weird that the timer is also running slowly, because I'm surprised they didn't use a better method to create that feature. It appears to be going at the correct speed for me; are you sure that it is actually slow?
The timer reporter is slow.
set [sec v] to (current [second v]
set [min v] to (current [min v]
set [hr v] to (current [hour v])//reset timer

(((((current [hour v]) * (3600)) + ((current [minute v]) * (60))) + (current [second v])) - ((((hr) * (3600)) + ((min) * (60))) + (sec)))//fetch timer

I think we all already know how to work around a slow timer.

What I was asking, is what is your proof that timer is slow.

hedgehog_blue

-Bonfire- wrote:
ScratchDiogoh wrote:
——snip——
Anti [Anti Evil Kumquat v]

Stay on topic - somebody's signature isn't very related to their post.

Last edited by hedgehog_blue (June 12, 2018 01:33:56)

Raytracing

hedgehog_blue wrote:
-Bonfire- wrote:
ScratchDiogoh wrote:
——snip——
Anti [Anti Evil Kumquat v]
Stay on topic - somebody's signature isn't very related to their post.

The Anti Evil kumquat block Is very useful when protecting your signatures from evil kumquats that might eat your signature though. Evil kumquats a a serious problem in today's society.

Discuss Scratch