nVidia likely to remain accelerated video king?

[droidhacker]

(the drivers themselves are NOT open source, but Broadcom is working with the open source devs so that's a plus).

Really? Could have fooled me, they have the source code posted right here: http://www.broadcom.com/support/crystal_hd/

And its even in the KERNEL here: http://git.kernel.org/?p=linux/kerne...b385cc68fb7627

So how you go about suggesting that it isn't open source is REALLY beyond my comprehension.

So far, very few people have the card actually working 100% reliably under Linux and those that do tend to find that 1080p isn't renderable at all by the driver, especially in the gstreamer derivative.

http://wilsonet.com/?p=71

I'm not saying that Broadcom HD isn't a workable solution, but it's hardly the dream you make it sound like.

Sure beats nvidia, which just doesn't work sometimes, even on hardware that is supposed to be fine with it.

[gbeauche]

Really? Could have fooled me, they have the source code posted right here: http://www.broadcom.com/support/crystal_hd/

Yes, but this is not fully Open Source. This wraps around a 2.7 MB firmware. The EMGD kernel module has the same level of openess then.

Sure beats nvidia, which just doesn't work sometimes, even on hardware that is supposed to be fine with it.

It depends. The Broadcom solution has several problems:
- A single stream at a time, if not a single instance. i.e. device opened at a time. This is fixed for BCM70015 though.
- The need to read data back from the decoder, and upload the decoded frames to the GPU. This means this is a potentially slower approach (extra copies or memory used).

[IsawSparks]

Really? Could have fooled me, they have the source code posted right here: http://www.broadcom.com/support/crystal_hd/

And its even in the KERNEL here: http://git.kernel.org/?p=linux/kerne...b385cc68fb7627

So how you go about suggesting that it isn't open source is REALLY beyond my comprehension.

http://wilsonet.com/?p=71

Sure beats nvidia, which just doesn't work sometimes, even on hardware that is supposed to be fine with it.

I won't bother with the open source malarkey, but suffice to say you're wrong.

In terms of Nvidia not working with HD content. WHEN? WHERE?

All of my H264 MKVs, MP4s, WMV-HDs and Bluray rips work 100% and have since I had my first 8600GTS, 9600GT and 285GTX.

[kazetsukai]

All of my H264 MKVs, MP4s, WMV-HDs and Bluray rips work 100% and have since I had my first 8600GTS, 9600GT and 285GTX.

I had a first gen 8800GTS, a GTX260, and now a GTX470. The only thing that did not work was VC-1 on the 8800GTS, because the 8800GTS doesn't have the hardware for it. All of the later cards (including the 8600GTS and the newer 8800GTS cards) had this functionality.

I thought this site was above the Fanboyism denial that other hardware/software sites suffer from. Fact: NVIDIA has really decent hardware acceleration which makes high definition video generally a better experience. ATI doesn't, and while promising to bring an implementation over to Linux time and time again, they've continued to fail to do so. All I hear is whining and claims that it does nothing for the video experience, only from the people that -don't have it-.

Right now, NVIDIA + blob = accelerated video king. There is no better combination of performance + featureset on any other solution. Rather than coming here and posting why you think it isn't important or doesn't work well when it is and does, go write a letter to ATI saying how much you want the functionality.

If my driver and hardware vendor don't provide critical functionality, I don't get defensive on their behalf, I take the offense -against them-, on their behalf. Its what ultimately puts the features you want into your driver, and I believe what won ATI its superior Open Source driver. Start putting the pressure on your own side of the fence already.

[droidhacker]

Yes, but this is not fully Open Source. This wraps around a 2.7 MB firmware. The EMGD kernel module has the same level of openess then.

Look at all the hardware on your computer and try to find JUST ONE that has fully open source FIRMWARE.

Good luck.
OPEN SOURCE DRIVERS have nothing to do with the FIRMWARE.
OPEN SOURCE DRIVERS are the glue that goes between whatever kind of firmware the device has and the KERNEL.

And in this case, the drivers are OPEN SOURCE.

It depends. The Broadcom solution has several problems:
- A single stream at a time, if not a single instance. i.e. device opened at a time. This is fixed for BCM70015 though.

HOW MANY VIDEOS can you watch at a time?
Right. One's plenty!

- The need to read data back from the decoder, and upload the decoded frames to the GPU. This means this is a potentially slower approach (extra copies or memory used).

That is an ADVANTAGE not a disadvantage.
It means that the decoded output is AVAILABLE and not stuck in a proprietary mess. And if you had any concept of how things are implemented, you would see that the impact this has is NEGLIGIBLE.

[IsawSparks]

HOW MANY VIDEOS can you watch at a time?
Right. One's plenty!


That is an ADVANTAGE not a disadvantage.
It means that the decoded output is AVAILABLE and not stuck in a proprietary mess. And if you had any concept of how things are implemented, you would see that the impact this has is NEGLIGIBLE.

1. Bluray has menus and features (such as multiple camera angles etc) which need more than one stream. This includes rips. If I'm editing HD video, I need more than one HD stream playable at a time for realtime previews of edit. One stream is not plenty. Maybe on a netbook, where expectations and are lowered it's ok, but not in a Desktop Linux, Windows or Mac installation.

2. The firmware provides the functionality, not the driver. The driver just implements said functionality in wrappers which allow the firmware to pass of data to the app of choice. The firmware is where the work needs to be done, such as being able to support the correct stream implementations to play 1080p, which isn't fixed yet. Therefore the open sourceness of the driver (which it isn't despite what you argue) is moot if the firmware doesn't deliver the features needed.

THEREFORE

3. Broadcom HD chipset is just the same proprietary mess as any closed source driver delivers.

Nvidia's blob with the correct level of hardware (8600GTS and above) delivers a complete all in one solution of full OpenGL and HD video decoding with a lot less headache than jury rigging YET ANOTHER proprietary piece of hardware into a system to get HD Video playback kinda, sorta, but not really working.

[clearer]

Look at all the hardware on your computer and try to find JUST ONE that has fully open source FIRMWARE.

My graphics card, sound card, chipset -- actually pretty much everything in my computer is firmware free. The only thing that's not is my WiFi, which I am considering ditching for something that is firmware free. Just need to find a chip that fits the description first (recommendations are welcome).

OPEN SOURCE DRIVERS have nothing to do with the FIRMWARE.
OPEN SOURCE DRIVERS are the glue that goes between whatever kind of firmware the device has and the KERNEL.

And in this case, the drivers are OPEN SOURCE.

That's a very liberal view you have on what makes something open. Besides, "open source" is a worthless term anyway, but I won't go into that (so please just ignore it if it bugs you).

HOW MANY VIDEOS can you watch at a time?
Right. One's plenty!

Not if you want to play more than one video at a time on the same computer -- for example for trade shows and demonstrations or similar things. Video playback is used for more than watching porn.

The "firmware" discussion is a dead end -- it's been debated. Drivers that depend on binary blobs, regardless of what you call them or on what device you load them, makes the drivers incapable of being open. They run software ("firmwares" are compiled) which no-one has any chance of correcting, expanding or replacing, thus making them closed, even if there is some open glue, between the kernel and the real driver (i.e. the "firmware").

[droidhacker]

My graphics card, sound card, chipset -- actually pretty much everything in my computer is firmware free. The only thing that's not is my WiFi, which I am considering ditching for something that is firmware free. Just need to find a chip that fits the description first (recommendations are welcome).

Wow. You don't see it, so it must not be there, right? LOL.
I will GUARANTEE that your video card, whatever it is, has TONS of proprietary firmware.

That's a very liberal view you have on what makes something open. Besides, "open source" is a worthless term anyway, but I won't go into that (so please just ignore it if it bugs you).

Meaningless....

Not if you want to play more than one video at a time on the same computer -- for example for trade shows and demonstrations or similar things. Video playback is used for more than watching porn.

Well, we can see what you use your computer for...
Here's a suggestion: GET OUT OF YOUR DARK CAVE AND GET LAID!

The "firmware" discussion is a dead end -- it's been debated. Drivers that depend on binary blobs, regardless of what you call them or on what device you load them, makes the drivers incapable of being open. They run software ("firmwares" are compiled) which no-one has any chance of correcting, expanding or replacing, thus making them closed, even if there is some open glue, between the kernel and the real driver (i.e. the "firmware").

And.... EVERYTHING HAS IT. Doesn't matter HOW blinded you are to that fact.

[kazetsukai]

And.... EVERYTHING HAS IT. Doesn't matter HOW blinded you are to that fact.

Yes, but the difference here is, how much of your "open" driver is actually open? Almost none of it, since they crammed the majority of the meaningful code into the firmware.

I don't have to download some firmware file for my video card, one driver is enough. Sound, too. Most hardware ships with a static firmware which never leaves the device. Yours doesn't.

What are you getting so upset over?

[droidhacker]

1. Bluray has menus and features (such as multiple camera angles etc) which need more than one stream. This includes rips. If I'm editing HD video, I need more than one HD stream playable at a time for realtime previews of edit. One stream is not plenty. Maybe on a netbook, where expectations and are lowered it's ok, but not in a Desktop Linux, Windows or Mac installation.

You can't play bluray on linux to begin with, so how is that relevant?
Editing videos? So you're previewing two videos simultaneously? Good luck observing where you screwed up when your focus is split between two...

2. The firmware provides the functionality, not the driver. The driver just implements said functionality in wrappers which allow the firmware to pass of data to the app of choice.

What's your point?
That open source is a myth?
The point that ***I*** am making is that it is AS OPEN AS ANYTHING ELSE, and typically MORE open (certainly more open than ncrapia).

The firmware is where the work needs to be done, such as being able to support the correct stream implementations to play 1080p, which isn't fixed yet.

Isn't fixed? Evidence points to the contrary.
In fact, I haven't seen any evidence to suggest that there was EVER an issue.

Therefore the open sourceness of the driver (which it isn't despite what you argue) is moot if the firmware doesn't deliver the features needed.

As I said... IT IS AS OPEN AS ANYTHING ELSE.
If you are of the opinion that it isn't open, then NOTHING IS.
Your arguments prove NOTHING.

3. Broadcom HD chipset is just the same proprietary mess as any closed source driver delivers.

Therefore, by your logic, there is NO SUCH THING as open source.

Nvidia's blob with the correct level of hardware (8600GTS and above) delivers a complete all in one solution of full OpenGL and HD video decoding with a lot less headache than jury rigging YET ANOTHER proprietary piece of hardware into a system to get HD Video playback kinda, sorta, but not really working.

Nvidia's drivers are CRAP. I would absolutely NOT be willing to suffer with their garbage drivers just for unreliable and very badly broken video decode acceleration.