Yeah I looks like VP2 hardware works as well. It also appears that Nvidia wants Intel and ATI to write drivers for their hardware.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Flash 10.1 updated with hardware acceleration support, codenamed "Gala"
- Thread starter aiqw9182
- Start date
- Sort by reaction score
Yeah I looks like VP2 hardware works as well. It also appears that Nvidia wants Intel and ATI to write drivers for their hardware.
Meanwhile, Adobe is adding hardware assisted decoding for even total crap chipsets like the Intel GMA 500 series which can run 720P now with little CPU use on Windows while Apple cannot decode HD video on even their flagship Mac Pro with a top of the line video card, but that's not Apple's fault....
I have a GMA 950 MacBook and am at the point where I'll likely need to switch to Windows 7 full time via bootcamp to have flash without significant audio/video distortion (Apple claims it needs a new logic board, but I can't stomach paying near the cost of an iPad for a 3 year old model when it works ok otherwise).
I know it's an older machine, but if Apple were to allow Adobe to support it that could make the difference in this case. It sure needs it more than my new Mini that's perfectly silent even with the pre-installed flash from Snow Leopard (not version 10.1)!
The Intel GMA 500 is a gray area since it isn't an Intel GMA product but a licensed PowerVR one. You would at least need to hit the GMA X4500HD or GMA 4500MHD before Intel provides that support in their GMA hardware.Meanwhile, Adobe is adding hardware assisted decoding for even total crap chipsets like the Intel GMA 500 series which can run 720P now with little CPU use on Windows while Apple cannot decode HD video on even their flagship Mac Pro with a top of the line video card, but that's not Apple's fault....
In addition there are plenty of devices/netbooks that also use Broadcom's HD/h.264 decoders to supplement Intel GMA. Adobe does support these Broadcom products for hardware based decoding and acceleration. XBMC does so as well.
Not true. The GeForce 6600 and later models (GF6) have H.264 acceleration hardware. Some of the early GF6 cores were MPEG1/2 and VC1 only.
For the first question, the answer is no. There is no fully hardware agnostic acceleration layer available--no one supports first generation (GF6/7) hardware in this application. For the second, only Apple knows. They may or may not be working on current-generation ATI support.
It's not a rant at all. I feel like I'm talking to a child when your posts come back. For that matter, I probably am.
Let's try it this way: (1) a 1.5 year old computer has nothing to do with the hardware capabilities of its components, (2) a 1.5 year old graphics card has nothing to do with its hardware capabilities, (3) The age of the core will define its featureset. The G84/86/G92 are 2007-era cores with an older implementation of the hardware featureset used in acceleration applications. Because Apple introduced acceleration a year later than nVidia's Windows VDPAU implementation, the cutoff for supported hardware was also about a year later.
On Windows, VP2 was the hardware of the time. VP1 was not supported. When OS X implemented it, VP3 was the current hardware. VP1 and VP2 were not supported. This is par for the course in terms of graphics features.
"Total crap chipsets" can still be newer and thus have new features, irrespective of their relative performance.
I didn't say they shouldn't get it or that they wouldn't benefit from it. I said that highly optimized task-specific acceleration is less important when you have overall powerful graphics. Specialized hardware like H.264 acceleration is designed to do a single task to make up for the main GPU's inability to perform adequately (whether that's in terms of power consumption or processing resources). This is why nVidia's VP3 appeared first on its lower-power cards.
No, I say that GPU cores that are from a prior generation of hardware technology are outdated because they are a generation old. Many of the cards from a generation back continue to outperform current-generation products, but because they are older, they do not contain the logic features of newer products. It has nothing to do with the age of the video card or the age of the computer, nor with the performance of either using commonly-shared features.
It's just that something designed in 2009 is going to have some new improvements that something designed in 2008 didn't have. That's life.
VDPAU is a generic API. Implementation is wholly up to the individual users. VDPAU Feature Set B was implemented by Apple. VDPAU feature Set A (VP2) was implemented for Windows presentations, and then updated to include Set B. When Set C is announced, both Windows and OS X will presumably be updated to include them.
Neither platform has gone backwards to support older featuresets (VP1 on Windows, VP1/VP2 on OS X).
VDPAU, the API, is for Unix-like systems. But the hardware features of the PureVideo engine, as defined by VDPAU feature sets, are hardware based and thus platform-independent. Much as "DirectX 9" shaders specify features also used by OpenGL, the VDPAU sets are a useful baseline for the hardware.
What prevented Apple from supporting hardware limited to VDPAU feature Set A then?
Apple's implementation in Quartz is based on Set B (VP3) hardware. Windows PureVideo is based on Set A (VP2) hardware; Linux's VDPAU for X is based on Set A as well, because it was a recreation for Linux of the Windows PureVideo feature set. Both PureVideo and VDPAU predate both the VP3 hardware release and Apple's own implementation of acceleration APIs.
The reason that VP2 isn't supported on OS X is the same reason that VP1 isn't supported on Windows or Linux--it wasn't the current implementation when the system was created, and Microsoft, nVidia, and Apple all made the same decision not to invest in coding for last-generation hardware.
As an illustrative timeline:
VP1/Set N/A [unsupported] (GF6) > VP2/Set A (GF8) > PureVideo (Windows) > VDPAU (Linux) > VP3/Set B (GF9/GT100) > Quartz acceleration (OS X) > VP4/(Presumed) Set C (GT200)
Each implements the then-current hardware featureset and builds going forward.
Thank you for this clarification. Also, I believe that Windows uses DXVA 2.0 today. PureVideo would be the name for the specific components of nVidia's hardware that decode video for playback.
DXVA is a higher-level API built on top of API implementations by the graphics manufacturers (PureVideo Engine and UVD from nVidia and AMD, respectively), but yes, you're correct that this is where most software applications go for access in Windows (as most go to QTKit on OS X for the same).
PureVideo APIs can also be implemented in OpenGL on Windows (though I don't know if they have been), which naturally would have nothing to do with DXVA.
Could you provide more details about interacting with PureVideo and UVD outside of DXVA?
Sure. PureVideo API is implemented in the graphics driver, exposed through the Aero DWM. DXVA is implemented in Windows Media Foundation, which sits on top of the DWM. MF provides modern features as well as integrates some legacy systems (DirectShow, VFW, and some others I can't remember off hand).
However, any software not relying on Media Foundation, e.g. through the use of an ICD (installable client driver), such as OpenGL, would interface in their own way with the DWM layer, rather than through WMF.
But as I said, I don't know if anyone has made any attempt to do so for OpenGL on Windows. The primary purpose of PureVideo is video playback in desktop applications, and I'm not aware of any OpenGL-based video players for Windows.
I suspected the interactions outside of DXVA were handled in the drivers provided by the hardware manufacturers. Though I do suspect most would use DXVA and hardware vendors would create the proper drivers regardless.
Regardless of how old the hardware is, it isn't old in Apple's product timeline. It's Apple that decides to use "old" hardware in their "new" products.
No matter how you spin this, it is unacceptable to have a 2 year old MBP that is capable of h.264 acceleration in Windows and not OS X. Especially when OS X is supposed to be "the world's most advanced OS".
It's not unreasonable to expect support when the hardware is capable. Doesn't matter if Apple got into the game late with h.264 acceleration (which is ironic considering how they embrace h.264). Part of paying a premium for the Apple brand is the expectation that the support is there as well.
As has already been said multiple times, neither the age of the computer or GPU, nor the raw power of the hardware is relevant.
This is a highly amusing statement to make for several reasons, and here is the first: Apple was a launch partner for the 8600M GT in the 2007 MacBook Pro. Second, Apple was the launch customer again for the 9400M.
Further, nVidia, to this day, sells products based on GF8 cores as "new". The 8600M was based on an existing core because nVidia designed it that way. It was in its time a high-performance notebook GPU, and even today it's a solid performer.
For example, the new GT210M used in many notebooks has less raw processing power than the 8600M (72 vs. 91 GFLOPS). The 8600M is OpenGL 3 and DirectX 10 compliant. If it were still used in notebooks, it would remain a competent performer on the whole. But the age of its core (and thus its featureset) and its power consumption eventually caught up with it.
I'm not spinning anything. All I'm doing is explaining.
OS X beats Windows to some things, and Windows beats OS X on some others. Users of each are occasionally left out. It can be frustrating, but that's the way it goes. How do you think owners of the 6600 felt when DXVA acceleration wasn't made available to them? They had computers that in some cases were less than a year old and didn't get it. As I've said multiple times now, the age of the computer is irrelevant to what logic features it contains. You can buy a brand new IGP that gets spanked by a five-year old computer with a then-high-end GPU. But that new IGP is going to have features that the older, more powerful card does not.
For another immediate example, Windows 7's Direct2D implemented last year some features that have been part of CoreImage since Tiger--and a number of video cards that are CoreImage compatible are not compatible with Direct2D because of evolution in the time between the Apple and MS implementations.
What does that even mean? Even assuming for the moment that it's true, being the most advanced on the whole doesn't mean there aren't specific things that other OSes do better or sooner.
For one, it's highly unreasonable when it comes to graphics cards. But more to the point, the hardware isn't capable of using the OS X API any more than the GeForce 6600 (which also had hardware H.264 acceleration) is capable of using the Windows API.
I get that you're trying to be funny, but it isn't ironic at all. H.264 has been running fine for years with both general hardware acceleration and software decoding (you've been watching it for years on both Windows and OS X, in and out of Flash).
Specialized hardware acceleration exists to solve some very specific problems and limitations. While all GPUs benefit from H.264 acceleration, most don't need it.
Said multiple times by an overinflated egotist, perhaps. That doesn't make it true, however.
Basically, this boils down to Apple is BEHIND in their support of hardware assisted technology present in "older" chipsets and so rather than add the older VP2 to OSX so their products that are less than two years old can have acceleration, they are instead choosing to "not bother". You defend this as a waste of time to add older features to the operating system, but this still boils down to poor support in OSX when the chipsets were new and total laziness (or perhaps greediness) to retrofit the operating system now simply because Apple no longer sells those chipsets.
Your dismissal of this as "normal" is what is so absurd. There is simply no technical reason the support for VP2 couldn't still be added to OSX and make many of their customers happy that have still very relevant equipment in their possession. But like so many fanboys who defend EVERYTHING Apple does, your opinions are simply based on "Apple knows best". The mere fact people on here are booting into Windows on a regular basis just to surf the web on machines that are only 2-3 years old says something about how poor Apple's support is for their own hardware. Ultimately, "buy a newer computer" is a very poor excuse for a lack of support of older hardware, particularly when newer hardware isn't needed to perform the vast majority of computing tasks. HD Video is one of those areas where hardware support makes a real difference and it is here that OSX fails to deliver and is by far the WORST of all three major operating systems out there in that regard. Until just recently there was NO support for 3rd party hardware-based acceleration in this operating system.
Making *EXCUSES* for this is what makes your posts so DISGUSTING. Furthermore, calling people children when you act like one yourself is all the more laughable and is tantamount to flaming as well. Wanting Apple to support hardware that is more than one year old is not childish. It's a completely realistic expectation and simply leads to the idea that Apple should not have a virtual monopoly on OSX hardware, especially if they are not going to support it for more than a year or two. This is becoming painfully apparent with the iPhone, which apparently is considered disposable after a mere three years, regardless of whether it's actually USEFUL or not. All these things are simply due to GREED on Apple's part and nothing more. You, apparently would simply call this a normal fact of life. It is small wonder then that Apple products have traditionally had a aura of smugness to them. Either you continually buy the latest hardware (no matter how foolish a use of money this might be) or you are looked down upon by that segment of the community that simply cannot fathom why someone would continue to use a PPC machine or not replace their Mac Pro every other year with another $3000+ machine, regardless of whether it's making that person any money to justify relatively small improvements in speed. Such people give whole new meanings to disposable income.
I personally see little reason to replace a MBP that is less than two years old with newer machines when Apple has done much to regress the features during that time. Such examples included removing the option for a matte screen (and later charging a surcharge for an adapter to limit glare), removing the expansion slot on the 15" "pro" model and replacing with a relatively pointless SD card reader (which would easily be added to the expansion slot or even a USB port if actually needed while limiting the true pro uses of the machine), removing the battery access door in order to be able to charge for replacement batteries and/or encourage hardware replacement over time (this is becoming endemic of all Apple portable products and it's simply unacceptable), removal of the extra firewire 400 port (forcing adapters and/or hubs to be added) and all together in certain models all while newer GPUs have relatively small improvements in actual performance (neglecting Apple's refusal to support features older hardware included that Windows supports such H264 hardware decoding). Then there's the new trackpad, which Apple supported newer features in Leopard, but ignored older models until Snow Leopard (thus trying to force an OS upgrade, despite the fact that Snow Leopard runs many applications significantly SLOWER than regular Leopard, which gives a bad impression for an OS that has supposedly been "optimized" since Leopard.
Yes, Apple is "normal" and we users are "children" that expect more from a company that sells products at inflated prices compared to similar hardware for Windows while fighting the use of such hardware with their OSX operating system.
If Apple wants its users to keep buying overpriced hardware, they need to supply the enhanced SUPPORT that would keep a user happy with that hardware. To me, that means supporting hardware features for more than a year or two. To you, it means buying a new computer every year to get features in software that may already exist in hardware that is several years old but Apple doesn't support out of pure greed or laziness. You can call that "normal" if you want, but some of us know better.
Your sad personal vendetta aside, the fact that it is true is what makes it true.
I'm not defending anything. I'm explaining what actually happened, something that is clearly beyond your ability to comprehend. One need not like the decision, but at the very least he should understand it.
The exact same thing can be said of nVidia and Microsoft and AMD in their implementations. It would be equally untrue in all cases. Your psychological need to make this about Apple when it's really just about the industry as a whole is as puzzling as your inability to comport yourself in a mature manner.
The technical reason is that VP2 is not VP3. The features in Apple's API require VP3 as minimum hardware. There's no way to "add" VP2 to that implementation.
They could have written an implementation with lesser requirements, but what would be the point? Performance of existing products doesn't degrade with the introduction of new features in new models, and the dedication of resources to supporting old hardware would not provide any future benefits.
When Microsoft sets the hardware requirements for a new version of DirectX, they do so using the current hardware. A DX9 card doesn't get to use all the features of DX10. Deal with it. Complaining that Microsoft is greedy and lazy and should have released all DX10 features in DX9 is ridiculous.
And you're done. Apparently, your psychological need also extends to making what has been a detailed explanation that has little do with Apple into a fanboy rant, where it is clear I have made zero normative statements about my opinion about Apple's conduct. You can't have a discussion on this forum without melodrama and personal entanglement, and you should know that it accomplishes nothing for you.
Bzzzt. Wrong. Thanks for playing. Third party hardware acceleration has been available since the introduction of Core technologies and Quartz Extreme; H.264 specialized acceleration has been available through QTKit since 10.6.0. 10.6.3 exposed it outside of QTKit for software that isn't properly integrated with OS X.
I believe you're already clarified this point in great detail.
It was probably a little earlier than 10.6.0.
Right you are! I forgot that it was implemented in 10.5 for the 9400M notebooks, though I don't know if QTKit enabled this for third-party software before 10.6.
BZZZZZZZZT wrong. If you read up on what 10.1 actually offers you'll find that it's not just about h.264 video. It offloads vector graphics to the GPU (something about tesselation blah blah, can't remember) and it also uses acceleration for bitmap rendering 'whenever possible', whatever that means. There are also other tweaks like freezing off-screen video content so that no CPU cycles are used for content that's not in view anyway. Not hardware acceleration as such, but an improvement over previous versions that would play all Flash content full blast even if it was happening outside the viewable area of a window, e.g. banners at the bottom of a page you're currently only seeing the top of, or content in minimized windows etc.
But this info was published before this new beta for Mac, at a point in time when, tentatively, only the Windows version would have hw acceleration. I'm no tech wizard so I have no idea if Apple has now unlocked access to everything Adobe needs to accelerate vector+bitmap graphics on Mac as well, or if this was just about H.264, but either way, Adobe has done quite a bit more than just "check the h.264 box"...
Perhaps I'm not understanding the specifics. Is there a reason VP2 cannot be supported in OS X other than Apple not wanting to start there?
Dear Adobe,
Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you! Thank you!
Thank you!
Apple's implementation requires at least VP3 hardware, just as Microsoft's requires at least VP2.
Apple must collaborate with the GPU manufacturers to support features--it does not and cannot write all the necessary code itself. When Apple started working with nVidia, its engineers had already moved on to VP3 and so that's what they worked together to build on. Whether the reason for that was technological or a resource management decision by nVidia and/or Apple can only be speculated from the outside.
It is known that nVidia and AMD have fewer resources dedicated to Apple, Linux, and OpenGL relative to DirectX. It is also known that PureVideo has a degree of DirectX affinity, which can create some extra work when building OpenGL support (needed by Apple).
To my knowledge, there is no technological roadblock to having enabled support in late 2007 at the same time Microsoft did. It appears that nVidia did not turn its attention to X, OpenGL, and Apple until 2008, as it focused on the DirectX implementation first. As Apple could not have done it without nVidia involvement, nVidia's schedule may have played a role. It might also be that Apple just didn't get around to it until nVidia's engineers had already moved on to VP3.
Whatever the reason, when Apple and nVidia sat down to work, VP3 was the hardware of the day, and that's what they wrote.