Adobe's Flash Player 10.2 to Offer Up to Ten-Fold Improvement in CPU Efficiency

[Bregalad]

We need new measuring tools

Every computer has CPU monitoring software so it's easy to see when your CPU is being pushed hard.

Moving the processing from the CPU to the GPU will dramatically reduce CPU usage and everyone will see that. However, since almost nobody has GPU monitoring software the huge increase in GPU load will go unnoticed.

Thankfully, modern GPUs are very good at video decoding so shifting that work from CPU to GPU should reduce the energy required.

That still leaves questions like:
- how much more efficient than a modern CPU is a low end video chip? perhaps not much
- what overhead does Flash add to the process? could be a lot
- is that overhead something that GPUs do efficiently? not likely

In the end the only way to know whether Flash 10.2 is actually more efficient may be in measurement of heat production and battery life.

 

[Dalton63841]

I will stick to no flash, completely uninstalled it. I have noticed VERY impressive increase in battery life, and if I need to view a video I just use the iPad UserAgent in Safari.

 

[totoum]

If it's "open" then why does it require licensing fees?

Are there not patents for proprietary technology involved in h.264?

not "proprietary" technologies but "patented" technologies,as for you question,open doesn't mean free,they're two different things,H.264 is an open standard but not a free standard, i'll copy and paste what I said in another thread:

H.264 is open,you can access the spec and know how exactly to code and decode a H.264 video,if it weren't open an encoder like x264 wouldn't exist because the devs wouldn't have access to the knowledge of how to create an H.264 video.

An exemple of a "closed" propriotery codec would be codecs developed by On2 before google released them.
Take VP7 , only On2 devs knew exactly how to make a VP7 video,that info hadn't been released to the public so if you wanted an encoder or decoder you HAD to get it from On2 .That is not the case with H.264 there are many different encoders and decoders that are used to create and playback H.264

Another exemple is the Apple Prores codec (used in video editing),you can only get an encoder or decoder from apple,no one else.

 

[Cs_swagg]

Wirelessly posted (Mozilla/5.0 (iPod; U; CPU iPhone OS 4_2_1 like Mac OS X; en-us) AppleWebKit/533.17.9 (KHTML, like Gecko) Version/5.0.2 Mobile/8C148 Safari/6533.18.5)

I really like how he calls the MacBook air underpowered, however the 11" model (presumably the baseline model) gets 10% CPU usage but his pc registered about 20-25%.....

 

[achtung!]

i have 2 words for you:
WHO CARES!

cheers

 

[hagjohn]

Every computer has CPU monitoring software so it's easy to see when your CPU is being pushed hard.

Moving the processing from the CPU to the GPU will dramatically reduce CPU usage and everyone will see that. However, since almost nobody has GPU monitoring software the huge increase in GPU load will go unnoticed.

Thankfully, modern GPUs are very good at video decoding so shifting that work from CPU to GPU should reduce the energy required.

That still leaves questions like:
- how much more efficient than a modern CPU is a low end video chip? perhaps not much
- what overhead does Flash add to the process? could be a lot
- is that overhead something that GPUs do efficiently? not likely

In the end the only way to know whether Flash 10.2 is actually more efficient may be in measurement of heat production and battery life.

I have a GPU monitor on my desktop sidebar and with playing 720p youtube (normal size with 10.2 beta), I don't have any load on the GPU. I get slight increases for a second up to 10% and then it drops to 0 again. I have to run 2 720p youtube video's at the same time (normal size) in order for the GPU to jump up and stay to 20 some %.

Very far from scientific results but there you go.

 

[AppleFan1984]

Because the codec uses patented technology. It is "open" in the sense that it is openly licensed and controlled by a standards committee rather than a single company.

Ah, so it's several companies as opposed to a single company. Thanks for clearing that up.

You said h.264 is proprietary. Proprietary requires ownership. No one owns h.264. They own patents utilized in h.264. These patents were contributed to h.264 in exchange for licensing fees.

If you can't download the source and freely modify it and redistribute it, it doesn't fit most people's definition of "open".

 

[BaldiMac]

Ah, so it's several companies as opposed to a single company. Thanks for clearing that up.

No it was developed by a standard committee. It is not owned by any company or group of companies.

If you can't download the source and freely modify it and redistribute it, it doesn't fit most people's definition of "open".

That would be "open source." And a codec doesn't have source code.

 

[dwsolberg]

If Adobe can get Flash to stop turning my computer into a battery-powered heater, I'm all for it.

I also love Click to Flash: All of the Flash games and video without the extremely annoying, invasive ads.

 

[Kenrik]

What's new?

Flash still fails to reduce CPU usage, now enlists GPU to do the work.

Still fails to be efficient.

Someone who gets it... !!!

Notice how the "Underpowered" Macbook Air 11.6 used less CPU than the PC..?

That's because it has a more powerful graphics card...

The they are doing is playing a shell game with the processor usage. Now you see it now you don't (but it's still there)

 

[totoum]

If you can't download the source and freely modify it and redistribute it, it doesn't fit most people's definition of "open".

H.264 isn't software, it has no source code,the "source" would be the specification document which explains the standard,that document is a freely available as a pdf file you can download here: http://www.itu.int/rec/T-REC-H.264-201003-I/en

If you've done studies and understand what it all says you can create your own H.264 decoder or encoder software and choose to make it open source or not

A few peeople got together and made their own open source H.264 encoder,called it x264,the source code is available for download and you can browse it here: http://www.videolan.org/developers/x264.html

A few other people got together a made this proprietary decoder for windows called coreavc : http://corecodec.com/products/coreavc

Apple uses its own encoder and decoders and since it isn't open source the code isn't available.

But none of these people (Apple included) would be able to make any of this if H.264 wasn't open.

 

[BigMac2]

Ah, so it's several companies as opposed to a single company. Thanks for clearing that up.


If you can't download the source and freely modify it and redistribute it, it doesn't fit most people's definition of "open".

I think you don't really understand what is a open standard. H.264 is open because any one who want to create an new device or apps for creating or playing h.264 content can do so. The licensing fee and the consortium is there to make sure that every one respect the standard and make the interoperability between different device possible without depending on one developing team. Once you've licence the specs, you're free to code you own h.264 codec soft or hard without interoperability issue and depending on others sources codes.

This is why the same mp4 file can play on iPod, PSP, XBox, google chrome, Flash and hardware decode on GPU and mobile device. This is why you will never see direct hardware acceleration for MKV, AVI container and other opensource codec or container from hardware maker. Hardware maker need those consortium to make sure there investment into hardware acceleration will not be compromise by a video format with multiple incompatible variant (ex: DIVX).

 

[Kristenn]

Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 4_2_1 like Mac OS X; en-us) AppleWebKit/533.17.9 (KHTML, like Gecko) Mobile/8C148)

It no longer matters.

That sounds a little distorted to me. An improvement is an improvement. Like most of the world I have flash installed and use it. The plug in has only crashed twice. No big deal.

 

[SandynJosh]

I doubt it. The reason why Jobs hates flash has everything to do with business. It's a competitor to their store. Just before I came here I read that for the upcoming slates that "Microsoft is encouraging developers to create HTML5-based web applications for Windows 7 slates and host the apps on their own websites." My 1st thought was that Steve Jobs is not going to like that, since that means iPad/iPhone users will be able to use those apps, and that's when I decided to check macrumors. It's all about money, watching videos and playing free games on flash cuts into Apple's bottom line, and so would what Microsoft is doing.

Actually, having Apple and MS pushing for HTML5-based applications in place of Flash is a much stronger case for replacing Flash than if just Apple was pushing alone.

I suspect that may have been why the improved Flash was shown running on both an Apple OSX laptop and a MS Windows "whatever" laptop. You may have noticed that no Android, "we run Flash," devices were demoed.

 

[AppleFan1984]

Go check for your self with those examples:
http://dev.sencha.com/deploy/css3-ads/

The flash ECMA interpreter look like Netscape Navigator or IE6 in terms of efficiency. All modern browser now got more efficient javascript interpreter than Flash.

When did Adobe drop ActionScript in favor of ECMAScript?

The URL you provided is an interesting page, but it only discusses appearances and provides neither benchmarks nor the means to run each animation separately to obtain your own measurements.

That prompted me to do a little searching - here's a sampling of what I could find in the way of actual benchmarks:

http://www.themaninblue.com/writing/perspective/2010/03/22/

Vector charting test
HTML5: Avg (5.53) fps
Flash: Avg (21.29) fps

Bitmap Gaming Test
HTML5: Avg (8.13) fps
Flash: Avg (15.44) fps

http://www.craftymind.com/guimark2/

My results with HTML5 canvas :
Chrome 7.0.517.44 : CPU ~30%
Firefox 3.6.10 : CPU ~50%
IE 8.0.6001 : not working…

My results with Flash Player 10.1
Chrome 7.0.517.44 : CPU ~7% and stable memory usage.
Firefox 3.6.10 : CPU ~7% and stable memory usage.
IE 8.0.6001 : CPU ~1% and stable memory usage.

http://www.yopsolo.fr/wp/2010/11/16/flash-vs-html5-performance-test/

The fastest map renderings were achieved in Flashplayer (83ms), followed in distance by Chrome (446ms) and Firefox (610ms). Seems as if the days of flash aren’t counted yet (no big surprise).

http://vis4.net/blog/de/2010/04/map-rendering-speed-flash-vs-html5/

Developer Chris Black shows us two versions of the same animation, one done in Canvas and JavaScript and one done in Flash. He first runs it on a brand new iPod Touch (HTML5) and then on an Android Nexus One (HTML5 and Flash). The frame rate is much higher and steady on the Flash version — 57 frames per second versus 40fps in Canvas on the Nexus One and 22fps on the iPod.

http://www.wired.com/epicenter/2010/09/video-watch-flash-hand-html5-a-beating-on-mobiles/

What he found was that on the iPod Touch 4G (roughly identical hardware to the iPhone 4 sans 3G modem), a simple HTML5 canvas of a ball bouncing on a paddle rendered at a choppy 22 fps in Apple Mobile Safari browser. What's worse, the animation stopped entirely when zooming out or in.

By contrast the Nexus One averaged around 40 fps when using HTML5 in its built-in Chrome-derivative browser. The animation felt slightly sluggish, but was tolerable. And zooming out and in no longer stopped the animation.

Finally, Mr. Black tested Flash 10.1 running on Nexus One (in-browser). The results were a silky-smooth 57 fps, near the target of 60 fps. What's more, after running the test for 10 minutes, the Flash animation consumed only half the battery that running the equivalent HTML5 animation did.

http://www.dailytech.com/UPDATED+Be...formance+is+Far+From+Magical/article19681.htm


Note that all of those are performance benchmarks, leaving open the other side of the issue for debate: which is more efficient to develop in?

 

[electric]

Be careful what you wish for

http://www.thewildernessdowntown.com/

Great project but it too eats up the processor and does not work on the ipad, sorta like flash

 

[AppleFan1984]

I think you don't really understand what is a open standard. H.264 is open because any one who want to create an new device or apps for creating or playing h.264 content can do so.

Well, that and a few million dollars.

 

[AppleFan1984]

http://www.thewildernessdowntown.com/

Great project but it too eats up the processor....

Indeed it does. Spike at 95% here, averaging about 87%.

And I had to use Safari or Chrome, meaning the other 85% of the world can't run it.

 

[totoum]

Well, that and a few million dollars.

Of course not,again x264 is open source encoder that makes H.264 video,it's used by apps like handbrake: http://handbrake.fr/

Do you really think the handbrake developers paid "a few million dollars"? of course not.

The developpers don't need to pay a thing,however,if you as a user of the software use it (or any other software) to make H.264 video then you might end up having to pay.

edit: and vlc uses open source H.264 decoders,doubt the vlc team paid millions of dollars.

 

[BaldiMac]

Well, that and a few million dollars.

You don't need a few million dollars to create an encoder or decoder. You don't have to pay MPEG LA a penny unless you distribute it to 100,000 end users. And then it's only in certain circumstances.

 

[joueboy]

Comparison!

Now that they have it. I want to see a comparison between the html5 and see which one is more efficient. I think what we should look here is, is the new flash competitive on efficiency or just an improvement? Imagine how much battery consumption you could save on your portable device. If it's just an improvement then thank you but no thank you then.

 

[AppleFan1984]

You don't need a few million dollars to create an encoder or decoder...

...provided you include a clause that your codec isn't to be used for any commercial purpose:

So what are the rates, and who has to pay?

An excellent question that cuts to the heart of the matter. Remember, patents cover making, using, and selling the patented technology, so the MPEG-LA actually offers two licenses: one for codec developers (who make and sell the patented H.264 technology) and one for video content and service providers (who use it to distribute H.264 encoded content). The rates vary significantly; the yearly royalties for distributing an encoder range from free to $5m, while the royalties for distributing for-pay content are subject to complex rules about distribution but also range from free to $5m. In any event, MPEG-LA has said to us that only the parties at the top and bottom of the H.264 tool chain are generally required to pay royalties; that is, the party who makes the encoder, and the party who distributes the encoded file to the end users. You can think of that as the first and last transaction, if you like -- the person who sells the encoder and the person who sells the content are the ones who have to pay.

That ultimately means products that come with an H.264 codec don't also come with a license to use the codec commercially -- in order to distribute H.264 content in a way that makes money, the distributor has to pay for a separate license. So products like Windows 7, Mac OS X, Final Cut Pro, Avid, and modern video cameras aren't licensed to distribute video for commercial use -- they all have fine print somewhere that says they're for personal and non-commercial use only.

http://www.engadget.com/2010/05/04/know-your-rights-h-264-patent-licensing-and-you/


But even that apparently provides no clear exemption:

On Thursday, the patent pool organization announced that for the H.264 license used by free web video, it will continue to waive royalty fees through the entire life of the license. Previously, MPEG-LA said it would waive royalties for licensees only until December 31, 2015.

This means that if you use H.264 solely for free web video, you will never have to pay a fee to the MPEG-LA. It does not rule out the possibility, however, that some other patent holder outside the MPEG-LA will come calling.

http://www.theregister.co.uk/2010/08/26/mpegla_v_google/


The MPEG-LA's summary of fees is available here:
http://www.mpegla.com/main/programs/AVC/Documents/AVC_TermsSummary.pdf

 

[Grfelix]

Well Said

#1: Where was this all along? You needed the HTML5 revolution in order to take action?

#2: It's too late, anyway. People are moving on.

#3: The shirt. My God, the shirt.

Well said,

Glad I'm not the only one who noticed point #3!!

 

[SandynJosh]

:
Originally Posted by Consultant View Post
Flash still fails to reduce CPU usage, now enlists GPU to do the work.
Still fails to be efficient.

Someone who gets it... !!!
Notice how the "Underpowered" Macbook Air 11.6 used less CPU than the PC..?

That's because it has a more powerful graphics card...

The they are doing is playing a shell game with the processor usage. Now you see it now you don't (but it's still there)

Neither of you "get it!"

One of you think (counter to the demonstration) that CPU usage didn't improve, and the other one of you think that all of the improvement went to the GPU.

While both HTML-5 and Flash offload work to the GPU, and while the MBA has a boatload less power (CPU & GPU) than a MBP, the net result of the demonstration was that on "underpowered" devices the revised Flash was much more efficient than the previous incarnation. The logical extension of this is that iDevices can be expected to respond as well too.

 

[totoum]

...provided you include a clause that your codec isn't to be used for any commercial purpose:

All this still has nothing to do with whether or not H.264 is open or not,what you're talking about is whether it's free or not,two different things.

If it were not open we wouldn't be debating whether or not developpers have to pay millions of dollars to develop an encoder or decoder because if it weren't open they wouldn't be able to develop an encoder or decoder period,not even for a trillion dollars.